Introduction to Bioinformatics: - Pathogenomics of Innate Immunity

Introduction to Bioinformatics: - Pathogenomics of Innate Immunity
M.Sc. in Molecular Medicine,
Institute of Molecular Medicine,
Trinity College Dublin,
Ireland.
Introduction to Bioinformatics:
February 2005
David Lynn (M.Sc., Ph.D.)
The development of this course was supported by
The Dublin Molecular Medicine Centre (DMMC).
&
The Conway Institute of Biomolecular and Biomedical Research,
University College Dublin.
http://www.binf.org/course2005/
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
Table of Contents
Introduction to bioinformatics ...............................................................................................1
Database formats and structure .............................................................................................4
Sequence formats & Accession Numbers ............................................................................10
Day 1: Interrogating sequence databases: SRS Sequence Retrieval System at the EBI
to find sequences by their annotation and Entrez...........................................................13
Day 1: Nucleic Acid sequence analysis (Feb. 25th 2005).....................................................20
Translating DNA ............................................................................................................21
Reverse complement ......................................................................................................21
DNA analysis .................................................................................................................23
Primer design..................................................................................................................23
Gene prediction ..............................................................................................................26
Alternative splicing ........................................................................................................28
Promoter characterization...............................................................................................33
Day 1: Protein Sequence Analysis (Feb. 25th 2005) ............................................................39
Physico-chemical properties...........................................................................................40
Cellular localization .......................................................................................................42
Signal peptides ...............................................................................................................44
Transmembrane domains................................................................................................46
Post-translational modifications .....................................................................................49
Motifs and domains ........................................................................................................51
Secondary structure prediction.......................................................................................52
Day 2: Accessing Complete Genomes (Feb. 28th 2005)......................................................55
UCSC Genome Bioinformatics – The Golden Path .......................................................56
Ensembl at EBI...............................................................................................................62
NCBI Genomic Biology .................................................................................................65
OMIM (On-line Mendelian Inheritance in Man) ...........................................................72
Day 2: Alignments and homology Searching (Feb. 28th 2005) ..........................................73
2 sequence alignments....................................................................................................73
Similarity (homology) searching....................................................................................76
Multiple sequence alignment..........................................................................................88
Day 2: Phylogenetic trees (Feb. 28th 2005) ..........................................................................95
Tree calculation methods: Distance matrix (NJ), Maximum Parsimony (MP),
Maximum Likelihood (ML). .........................................................................................97
A PHYLIP protocol......................................................................................................102
Bootstrapping ...............................................................................................................108
WebPhylip and PISE ....................................................................................................113
Further readings in bioinformatics ....................................................................................121
ii
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
Acknowledgements:
This course was designed and implemented by David Lynn and Andrew Lloyd while
working at the Education and Research Centre (ERC) at St. Vincent’s University
Hospital, Dublin. The course and manual grew naturally from The ABC
Bioinformatics Course, an earlier Irish National Centre for BioInformatics (INCBI)
project based on GCG and the WWW, to which Aoife McLysaght (TCD) was a
major contributor. That in turn owes a debt of gratitude to the ABCT tutorial
designed by Rodrigo Lopez when he was the Norwegian EMBnet node. This course
would never have got off the ground without the encouragement of Cliona
O’Farrelly, the Research Director at the Education and Research Centre (ERC) at St.
Vincent’s University Hospital. The development of this course was funded by the
Dublin Molecular Medicine Centre and the Conway Institute, University College
Dublin.
Any suggestions for improvement, notification of typos and the like should be sent
on to [email protected]
iii
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
Introduction to Bioinformatics
This course is designed to impress upon you that computers and the Internet can not
only make your work as a biologist easier and more productive but also enable you
to answer questions that would be impossible without computational help. Thus
there are some computational analyses that you could conceivably do on the back of
an envelope or with a pocket calculator and there are others so computationally
demanding that you would not attempt them without electronic help. An example of
the first would be to scan the following DNA sequence for ecoRI restriction
endonuclease sites (GAATTC):
>Adhr D.melanogaster
ATGTTCGATTTGACGGGCAAGCATGTCTGCTATGTGGCGGATTGCGGAGGGAGACCAGC
AAGGTTCTCATGACCAAGAATATAGCGAAACTGGCCATTCGGAAAATCCCCAGGCCATC
GCTCAGTTGCAGTCGATAAAGCCGAGTACTTCTGGACCTACGACGTGACCATGGCAAGA
ATTCATATGAAGAAGTACTGATGGTCCAAATGGACTACATCGATGTCCTGATCAATGGT
GCTACGCTGATAACATTGATGCCACCATCAATACAAATCTAACGGGAATGATGAACACG
TGTTACCCTATATGGACAGAAAAATAGGAGGAATTCGTGGGCTTATTGTTCGGTCATTG
GATTGGACCCTTCGCCGGTTTTCTGCGCATATAGTGCAGTGTAATTGGATTTACCAGAA
GTCTAGCGGACCCTCTTTACTATTCCCAGCTGTGATGGCGGTTTGTTGTGGTCCTACAA
GGGTCTTTGTGGACCGGGGTTTTTAGAATACGGACAATCCTTTGCCGATCGCCTGCGGC
GAGCGCCCCATCGGTTTGTGGTCAGAATATTGTCAATGCCATCGAGAGATCGGAGAATG
GATTGCGGATAAGGGTGGACTCGAGTTGGTCAAATTGCATTGGTACTCGACCAGTTCGT
GCACTATATGCAGAGCAATGATGAAGAGGATCAAGAT
(This sequence is written in Fasta format - see below for sequence formats.)
A computer could do it quicker, but it is still trivial to do it by eye. Especially as one
of the sites has been picked out in bold. Can you find the other(s)? Sequence
analyses impossible without a computer include, but are not limited to, most
operations that involve the sequence databases. The DNA databases (Genbank
EMBL DDBJ) are curated by three different groups in Bethesda, MD, Hinxton, UK
and Mishima, JP but, because they exchange information on a daily basis, should be
effectively the same in content. The DNA databases are doubling in size about every
year; they currently (15 June 2003) comprise:
32,528,249,295 bases, from 25,592,865 reported sequences.
So finding all of the ecoRI sites in GenBank or even the whole of a printed copy of
the human genome (3,200,000,000 bp) would take more than a few minutes.
1
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
This course will introduce you to some of the more commonly used bioinformatics
tools, tell you how to use them and, more importantly, how to use them "correctly"
or at least more effectively. Most of the analysis will be carried out on the World
Wide Web (WWW). This is partly because it is available to all comers without
requiring direct access to the necessary computers, which serve as database and
software repositories. But it is also partly because a well-designed Web site can be
particularly user-friendly and intuitive in its operations.
There are likely to be network related problems trying to make 25 simultaneous
connections over the Internet to the same site. Try doing the course exercises late in
the evening, early in the morning (best for speed!) or at weekends.
This module in bioinformatics is designed to give you a flavour of what analytical
and informative tools are available on the World Wide Web.
Bioinformatics
Bioinformatics has been described as the storage, retrieval and analysis of biological
sequence information. In this short course we will be taking a broader definition:
how computers can maximise the biological information available to you. This will
touch on determining the 3-D structure of bio-molecules and trying to relate this to
their function as well as accessing the relevant literature. I hope that, by the end of
the course, everyone will be adopting a more explicitly evolutionary understanding
of ‘their’ molecule. The formal course practicals can be carried out entirely on the
World Wide Web using Netscape or the other Web-browser. Nevertheless, we
recommend using locally installed (FREE) software for the phylogenetic trees part
of the course.
You should note that several important types of bioinformatic analysis are not freely
accessible on the Web, but are available on various password controlled computers.
In particular, types of analysis that require large amounts of computational
power/time are best carried out off the web. Analyses of many genes are also often
better done in an environment where a computer program does the pointing and
clicking for you. For the record, the GCG package is a suite of programs which
carry out almost all the analyses that a molecular biologist might want to do with/on
DNA or protein sequences (secondary structure prediction, two sequence alignment,
conceptual translation of DNA, restriction site analysis, primer design, as well as
2
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
homology searching, multiple sequence alignment etc.). For phylogenetic inference
and tree drawing, the PHYLIP package (versions available for PCs, Macs and Unix)
will answer most needs. Both of these software packages and a variety of other
sequence analysis packages are available at the Irish National Centre for
BioInformatics (INCBI) contact Kevin Byrne ([email protected]). UCD also has
a GCG site-licence. GCG and PHYLIP are “packages” because they are internally
consistent: if you have run one GCG program you can run any other.
The web, by contrast, is a total mess: the same program is implemented with
different defaults at different sites; it is often not clear what those defaults, options
and parameters are; the results are not easily transferred to a different program. So it
is free, but there is a cost! You are advised to validate any analysis against the
results yielded by other sites.
3
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
Databases:
Databases are of course the core resource for bioinformatics. There is plenty of
software for analysing one or a few sequences, but many of the computationally
interesting and biologically informative programs access databases of information.
Frequently used classes are the biological sequence databases. These include:
-
EMBL (European Mol Biol Lab)
-
GenBank
-
DDBJ (DNA DB of Japan)
These three DNA databases exchange their data on a daily basis and so should be
identical as to content. They are, however, rather different in format:
Each of the database cited above consists of a (very large number) of entries, each
consisting of a single sequence preceded by a quantity of 'annotation' that puts the
sequence in its biological, functional and historical context. Without the annotation,
GenBank would be a meaningless string of 32 billion As Ts Cs and Gs. Compare
and contrast the two extracts from a) EMBL and b) Genbank (DDBJ has the same
look-and-feel as Genbank):
a) EMBL
ID
AC
DT
DT
DE
KW
OS
OC
OC
RN
RP
RX
RA
RT
RL
ECRECA
standard; DNA; PRO; 1391 BP.
V00328; J01672;
09-JUN-1982 (Rel. 01, Created)
12-SEP-1993 (Rel. 36, Last updated, Version 4)
E. coli recA gene.
.
Escherichia coli
Bacteria; Proteobacteria; gamma subdiv; Enterobacteriaceae;
Escherichia.
[1]
1-1374
MEDLINE; 80234673.
Sancar A., Stachelek C., Konigsberg W., Rupp W.D.;
"Sequences of the recA gene and protein";
Proc. Natl. Acad. Sci. U.S.A. 77:2611-2615(1980).
b) GenBank
LOCUS
DEFINITION
ACCESSION
KEYWORDS
SOURCE
ORGANISM
REFERENCE
AUTHORS
TITLE
JOURNAL
ECRECA
1391 bp
DNA
BCT
12-SEP-1993
E. coli recA gene.
V00328 J01672
.
Escherichia coli.
Escherichia coli
Eubacteria; Proteobacteria; gamma subdiv; Enterobacteriaceae;
Escherichia.
1 (bases 1 to 1374)
Sancar,A., Stachelek,C., Konigsberg,W. and Rupp,W.D.
Sequences of the recA gene and protein
Proc. Natl. Acad. Sci. U.S.A. 77 (5), 2611-2615 (1980)
You can see that these two are obviously talking about the same sequence from
E.coli, but the information is encoded in a rather different way. This makes no
4
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
difference to us reading the text, but causes problems when writing a program to
interrogate a database.
Each database entry has a name, called ID or LOCUS, which tries to be mnemonic
and marginally informative. More importantly each has an accession number which
is arbitrary but which remains attached to the sequence for the rest of time. The
organism might become reclassified, the gene may get renamed and the ID is thus
subject to change, but by noting the accession number you should always be able to
identify and retrieve the sequence. Note also that the original publication is cited.
Usually there will be other papers documenting functional analysis, mutations,
allelic variations, 3-D structure and so on.
Further down in the entry is annotation about the sequence itself, so that the
sequence is parsed into meaningful bits called a features table:
a) EMBL
FT
FT
FT
FT
FT
FT
FT
FT
FT
FT
FT
FT
FT
FT
FT
FT
FT
FT
source
mRNA
RBS
CDS
mutation
mutation
1. .1391
/organism="Escherichia coli"
/db_xref="taxon:562"
191. .>1391
/note="messenger RNA"
229. .233
/note="ribosomal binding site"
239. .1300
/db_xref="SWISS-PROT:P03017"
/transl_table=11
/gene="recA"
/product="recA gene product"
/protein_id="CAA23618.1"
353. .353
/note="g to a in recA441 (E to K)"
720. .720
/note="g to a in recA1 (G to D)"
b) GenBank
FEATURES
source
mRNA
RBS
gene
CDS
mutation
mutation
Location/Qualifiers
1..1391
/organism="Escherichia coli"
/db_xref="taxon:562"
191..>1391
/note="messenger RNA"
229..233
/note="ribosomal binding site"
239..1300
/gene="recA"
239..1300
/gene="recA"
/codon_start=1
/transl_table=11
/product="recA gene product"
/db_xref="SWISS-PROT:P03017"
353
/gene="recA"
/note="g to a in recA441 (E to K)"
720
/gene="recA"
5
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
/note="g to a in recA1 (G to D)"
Again you can see that the information exchange between Genbank and EMBL
includes all significant portions of the annotation. Such useful signals and data as
the open reading frame (CDS for CoDing Sequence), the ribosome binding site,
intron boundaries, signal peptides, variants/mutations may be recorded.
Protein databases:
-
SwissProt
-
PIR (Protein Information Resource)
-
GenPept
a) Swissprot
ID
AC
DT
DT
DT
DE
GN
OS
OC
OC
...
...
CC
CC
CC
CC
CC
CC
CC
CC
CC
CC
RECA_ECOLI
STANDARD;
PRT;
352 AA.
P03017; P26347; P78213;
21-JUL-1986 (REL. 01, CREATED)
21-JUL-1986 (REL. 01, LAST SEQUENCE UPDATE)
15-DEC-1998 (REL. 37, LAST ANNOTATION UPDATE)
RECA PROTEIN.
RECA OR LEXB OR UMUB OR RECH OR RNMB OR TIF OR ZAB.
ESCHERICHIA COLI, AND SHIGELLA FLEXNERI.
BACTERIA; PROTEOBACTERIA; GAMMA SUBDIVISION; ENTEROBACTERIACEAE;
ESCHERICHIA.
-!- FUNCTION: RECA PROTEIN CAN CATALYZE THE HYDROLYSIS OF ATP IN THE
PRESENCE OF SINGLE-STRANDED DNA, THE ATP-DEPENDENT UPTAKE OF
SINGLE-STRANDED DNA BY DUPLEX DNA, AND THE ATP-DEPENDENT
HYBRIDIZATION OF HOMOLOGOUS SINGLE-STRANDED DNAS. IT INTERACTS
WITH LEXA CAUSING ITS ACTIVATION AND LEADING TO ITS AUTOCATALYTIC
CLEAVAGE.
-!- INDUCTION: IN RESPONSE TO LOW TEMPERATURE. SENSITIVE TO
TEMPERATURE THROUGH CHANGES IN THE LINKING NUMBER OF THE DNA.
-!- DATABASE: NAME=E.coli recA Web page;
WWW="http://monera.ncl.ac.uk:80/protein/final/reca.htm".
KW
DNA DAMAGE; DNA RECOMBINATION; SOS RESPONSE; ATP-BINDING; DNABINDING;
KW
3D-STRUCTURE.
FT
INIT_MET
0
0
FT
NP_BIND
66
73
ATP.
FT
CONFLICT
112
112
D -> E (IN REF. 5).
FT
TURN
4
4
FT
HELIX
5
21
FT
HELIX
23
25
FT
TURN
29
30
etc etc
b) PIR
>P1;RQECA
recA protein - Escherichia coli
C;Species: Escherichia coli
C;Date: 31-Jul-1980 #sequence_revision 14-Nov-1997 #text_change 14-Nov-1997
C;Accession: G65049; A93847; A93846; S11931; S63525; S63979; A03548
...
C;Comment: The recA protein plays an essential role in homologous
recombination, in induction of the SOS response, and in initiation of
stable DNA replication.
6
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
C;Genetics:
A;Gene: recA
A;Map position: 58 min
C;Superfamily: recA protein
C;Keywords: ATP; DNA binding; DNA recombination; DNA repair; P-loop; SOS
response
F;67-75/Region: nucleotide-binding motif A (P-loop)
F;141-145/Region: nucleotide-binding motif B
F;73/Binding site: ATP (Lys) #status predicted
Note that these two entries refer to the same gene from E.coli despite differences in
the way the data is encoded. However, in contrast to the difference between EMBL
and Genbank, the quality of the annotation is quite different. The 3-D structure of
this gene has been worked out and this information is reflected in the SwissProt
entry as the position of every alpha-helix and beta-sheet is noted. In general, the
quality of the annotation and the minimization of internal redundancy makes
SwissProt the preferred database to use. However, note that PIR records the Genetic
Map position of the gene; so it is probably good to scrutinize both databases to
abstract maximal information.
SwissProt also gives added value by incorporating a large number of DR (database
reference) tags, pointing to equivalent information in other databases.
a) SwissProt:
DR
DR
DR
DR
DR
DR
DR
DR
DR
DR
DR
DR
DR
DR
DR
EMBL; V00328; G42673; -.
EMBL; X55553; -; NOT_ANNOTATED_CDS.
EMBL; AE000354; G1789051; -.
EMBL; D90892; G1800085; -.
PIR; A03548; RQECA.
PIR; S11931; S11931.
PDB; 1REA; 31-OCT-93.
PDB; 2REB; 31-OCT-93.
PDB; 2REC; 01-APR-97.
PDB; 1AA3; 23-JUL-97.
SWISS-2DPAGE; P03017; COLI.
ECO2DBASE; C039.3; 6TH EDITION.
ECOGENE; EG10823; RECA.
PROSITE; PS00321; RECA; 1.
PFAM; PF00154; recA; 1.
When these are used as hypertext links they can enable a WWW browser to locate an
extraordinary depth of detail about a given entry, 3-D structure (PDB), protein
motifs (Prosite), families of related genes (Pfam), the DNA sequence (EMBL) and a
couple of specialist E.coli added-value databases. SRS is one program that makes
these hypertext links. The PIR cross-references are far fewer and less explicit; its
reference to Genbank (GB:U00096) refers to the whole E.coli genome, whereas
SwissProt points specifically to the gene (DR EMBL; V00328)
7
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
b) PIR
...
A;Cross-references:
UWGP:b2699
GB:AE000354;
GB:U00096;
NID:g2367149;
PID:g1789051;
All these databases are made up of entries, concatenated one after the other in plain
readable text. As such they are far bigger than necessary if you are trying to analyze
the sequence rather than interrogate or browse the annotation. For these purposes,
special high-compressed databases can be constructed. Frequently these are not
readable by humans because they have been optimized for speed reading computers.
One of the simplest compression protocols is called Fasta format in which the
annotation is edited down to a single title line followed by the sequence. The
sequence at the top of the chapter is in Fasta format. All protein databases use the
one-letter amino acid code, can you think why this might be?
Sequence Related Databases
Not all biologically relevant Databases consist of sequences and annotation. There
are databases of journal abstracts, taxonomy, 3-D structures, mutations and
metabolic pathways. Some of the most useful of these are databases which specialise
in particular entities that can be found dispersed in the "whole sequence" databases.
You notice one of the cross-references for the SwissProt entry is:
DR PROSITE; PS00321; RECA; 1.
Prosite is a database of protein motifs. PS00321 is a family of proteins that all have
the motif:
PA A-L-K-F-[FY]-[STA]-[STAD]-[VM]-R
and are all believed to bind DNA, hydrolyze ATP and act as a recombinase. One of
the members of this family is the recA gene in E.coli which gives its name to
PS00321. In the pattern above, the residues within [square brackets] are alternatives.
Convince yourself that ALKFFAAVR could belong to the family but ALKFAAAVR
could not. There are more than 1000 other families classified in a similar way.
Finding a Prosite link in a SwissProt gene is a great help in finding other proteins
related by structure and/or function.
Interpro - http://www.ebi.ac.uk/interpro/
8
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
You should also be aware of the Interpro project which incorporates and sorts data
from a diversity of protein motif and domain databases into one searchable metadatabase.
9
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
Sequence formats
As we have seen comparing database entries above, there are dozens of different
ways in which you can store or represent the same fundamental information.
Databases are often compiled in, highly conventionalized, readable English text.
Computers, being not so bright, will have difficulty reading and interpreting the
information unless the conventions are quite rigidly obeyed. There are a very large
number of ways you can write, store and transmit simple one-dimensional sequence
files. A common sequence interchange program called 'readseq' recognizes at least
22 different file formats. If a computer program does not recognize the format of an
input sequence it may not work or, worse, misinterpret header lines as sequence data
or otherwise mangle your analysis. Some commonly used file/sequence formats are
shown below:
1) GCG (a software package):
TRANSLATE of: ecrgcg check: 4152 from:
1 to: 1062 generated symbols 1 to: 354.
ECRECA.RECA
1062
ecrgcg.pep Length: 354 Oct 15, 1998
1 MAIDENKQKA LAAALGQIEK
51 ALGAGGLPMG RIVEIYGPES
101 DPIYARKLGV DIDNLLCSQP
151 TPKAEIEGE*
Type: P
Check: 9572
..
2) Fasta (named for a widely used homology searching program) – single title line
beginning >:
>ECRGCG TRANSLATE of: ecrgcg
MAIDENKQKALAAALGQIEK
ALGAGGLPMGRIVEIYGPES
TPKAEIEGE*
1 to: 1062
3) Staden (named after Rodger Staden - early, but still extant, software writer) –
same as raw sequence:
MAIDENKQKALAAALGQIEK
ALGAGGLPMGRIVEIYGPES
TPKAEIEGE*
4) NBRF/PIR (named after the protein database):
>P1;ecrgcg.pep
ecrgcg.pep, 354 bases, 218 checksum.
MAIDENKQKA LAAALGQIEK
ALGAGGLPMG RIVEIYGPES
TPKAEIEGE*
10
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
Accession numbers
The information above makes you aware of the diversity of ways in which
something so simple as a one-dimensional sequence may be represented. Another
source of confusion is the variety of identifying numbers attached to sequences and
knowing to which database they refer. Accession numbers are used as unique and
unchanging numbers. They are not mnemonic, although databases also have a less
stable, more memorable nomenclature: HBB_HUMAN, HSHBB, HUMHBB 2HBB
are all human beta globin IDs in various databases,
ƒ
GenBank/EMBL accession numbers: originally a letter followed by 5 digits
(X32152, M22239). When the number of sequences exceeded 2,600,000 - 2
letters followed by 6 digits (AL234556, BF345788).
ƒ
SwissProt. Still one letter followed by 5 digits, letter is either O,P,Q.
P23445.
ƒ
PIR: the ‘other’ protein database, one letter followed by 5 digits, but
numbers confusable with EMBL/GenBank: B93303 is chimp haemoglobin in
PIR but a random genomic clone fragment in EMBL.
ƒ
GenPept.
Conceptual translations from DNA that have not yet been
annotated well enough to get into SwissProt. three letters and five digits,
e.g.: AAA12345.
ƒ
Trembl (Translated EMBL): O, P or Q followed by 5 letters/digits.
ƒ
PDB protein structure records: 1 digit and three letters 1HBA, 1TUP
More recently, an attempt has been made to reduce the redundancy in the databases
(there were 180 copies of D. melanogaster alcohol dehydrogenase each with its own
accession number). One result is RefSeq - NCBI’s “reference sequence” database
RefSeq: Two letters, and underscore bar, and six digits,
mRNA records (NM_*)
NM_000492
genomic DNA contigs (NT_*)
NT_000347
curated/annotated Genomic regions (NG_*) NG_000567
Protein sequence records (NP_*)
NP_000483
We will see how RefSeq is becoming the central resource for gene characterization,
expression studies, and polymorphism discovery. Because of the high level of
necessary curation, it is not anywhere close to being comprehensive even for those
species that are included.
11
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
Accession numbers give the community a unique label to attach to a biological
entity, so we all know we are talking about the same thing. Sequences in databases
evolve as their real biological counterparts do. They need to be updated, corrected
and merged and we need to know which version of the sequence entry is being
referred to. GenBank has used gi numbers and, more recently, version numbers for
this. Each small change made to a Genbank record gets the next gi number e.g.
gi6995995 and so is totally arbitrary. Version numbers are appended to the accession
number after a dot – V00234.2, NM_000492.2.
The other programs to use in the course are many and varied. We have tried to put
links to them all on the course website:
http://www.binf.org/course2005/
A few overall points for the course:
•
Take the opportunity to compare and contrast different methods of doing a
particular analysis.
•
By all means take the defaults but be aware that changing them will almost
certainly get more or better information.
•
The Web is free, and you get what you pay for, so use the Web with care &
caution.
•
As with lab work it takes time to get the protocol working. Once you have one
that works for you, write it down, bookmark and remember it. But note, the Web
changes rapidly and you cannot afford to use outmoded technology for long.
12
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
DAY 1: Interrogating (sequence) databases
SRS - http://srs.ebi.ac.uk/
The DNA databases are enormously rich information resources partly because they
are so big, but it would make little sense if it consisted of a long list of As Ts Cs and
Gs. At the moment there are more than 3 million individual entries in EMBL. An
entry could be a fragment as short as 3 base pairs (e.g. M23994) or a large contig
consisting of many genes, including complete eukaryotic chromosomes (e.g.
X59720). The value of the database lies substantially in the quality of the annotation
which puts the sequence in its biological context.
As a biologist you may need to be able to interrogate the Database to find particular
sequences or a set of sequences matching given criteria, such as:
The sequence published in Cell 31: 375-382
All sequences from Aspergillus nidulans
Sequences submitted by Peter Arctander
Flagellin or fibrinogen sequences
The glutamine synthase gene from Haemophilus influenzae
The upstream control region of Bacillus subtilis Spo0A
SRS (Sequence Retrieval System) is a very powerful, WWW-based tool, developed
by Thure Etzold at EMBL and subsequently managed by Lion Biosciences, for
interrogating databases and abstracting information from them.
One of the neatest features of SRS is the fact that interrelated databases can be crossreferenced with WWW hypertext links. This means that you can discover the protein
sequence, the cognate DNA sequence, a family of related proteins in other species, a
Medline reference to read an abstract of the original publication, a 3-D structure - all
with a few point-and-clicks with the mouse.
There are several SRS servers on the Web. We will be using
http://srs.ebi.ac.uk/
13
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
at the EBI in England because a) it has a large number of interlinked databases b)
connectivity to the UK is good c) they are attempting to interconnect their SRS
server with their clustalW server and blast server. If the SRS server at the EBI is
slow you might try any of:
http://srs.hgmp.mrc.ac.uk/
http://srs.sanger.ac.uk/
The three servers (EBI, HGMP and Sanger) are all located within a few metres of
each other on the Wellcome Trust Genome Campus at Hinxton in England.
The documentation for SRS is getting better. With experience and practice you will
get to use as much of SRS's power as necessary to obtain the results you need. I will
show below, as a worked example, a series of instructions to obtain the sequences of
all the mammalian osteonectin proteins in SwissProt, and download them locally to
carry out a multiple sequence alignment using, say, clustalW. It should also be
possible to do the multiple alignment on the EBI clustalW server.
Use your browser (Netscape?) to go to http://srs.ebi.ac.uk/ or one of the other SRS
servers at the top of the Course page. You should see the following options:
Click on ‘Library Page’.
This takes you to what is called the TOP PAGE. This page allows you to choose the
database(s) that you wish to search.
The databases may be of various types,
including:
Sequence: Swissprot, sptrembl, PIR (Protein) or EMBL, emblnew (DNA)
Sequence related: prosite, blocks, prints (protein motifs and alignments), repbase
(restriction enzymes),
Protein3Dstructure: PDB, HSSP
For more information about the contents of the database click on the relevant blue
underlined hypertext link - UniProt say.
ƒ
Click the box [_] to the left of UniProt.
ƒ
Click on the Query Form tab at the top of the page
This will move you to a Query Form Page that permits you to submit particular
queries (such as have been suggested at the beginning of this chapter) to the
14
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
databases. At the top of this page will be a note of which database(s) you have
chosen to search and a block of four text-insert boxes which you can use to enter
your question.
to the left you will see some things you can change:
1. [Reset] - which clears the screen.
2. combine search terms &(AND) - which enables you to apply other logical
(boolean) operators.
3. Use wildcards which means that "bact" will be interpreted as bact* and look
for bacteria, bacteriophage, etc.
4. Number of entries to display per page (default is 30)
Your question can be entered into one of more of the text-insert boxes, thus:
ƒ
Click [All text] change to [Description] and insert osteonectin in
box
Note: it does not have to be osteonectin it could be ubiquitin or haemoglobin or
hemoglobin or actin & alpha. Separate keywords in the same box have to be linked
by a logical (Boolean) operator such as
and:
&
or:
|
but not:
!
ƒ
Click the next [All text] change to [Taxonomy] and insert
mammalia in box
ƒ
Click Search
a new window appears with Query "([uniprot-Description:osteonectin*] & [uniprotTaxonomy:mammalia*]) " found 6 entries. This is how SRS interprets what you
have entered in the boxes and the numbers of "hits" found.
ƒ
Under Display options change [UniprotView] to [FastaSeqs]
ƒ
Click [Save]
Make sure view is [FastaSeqs].
Click [Save]
ƒ
Click Netscape's File ....Save As
ƒ
Save as type – Text File (*.txt).
ƒ
Change selection .../wgetz to .../osteo.pro and then Click Save.
15
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
This should dump the concatenated fasta format protein sequences into a local file
called osteo.pro. You can use this file as input for clustalw (say in week 4 of this
course). There may be local security difficulties with downloading sequences onto a
public terminal - check with your neighbours or your demonstrator.
Query manager: a powerful tool
A quick example will show how you can combine very complex queries to zero in
on the sequence(s) you need.
Having selected your database(s) go to the Query Form Page and enter:
ƒ
[Description] calmodulin
you should get about 2000 entries.
ƒ
Click [QUERY] tab at the top of the page to get a new page and enter:
ƒ
[Organism name] human (or indeed Homo sapiens)
this will get you a large number of sequences.
ƒ
Click [RESULTS] tab at the top of the page
A new window should appear with the results for all the queries you have entered in
the current SRS session. In the top box of this page enter "Q1 & Q2" (leave off the
quotes!) Note: Your mileage may vary here. Q1 and Q2 may refer to earlier queries
in this SRS session (osteonectin?) so use good judgement.
You have just used a boolean logical expression to yield sequences which are a)
human and b) have "calmodulin" in the SwissProt description. This shows you how
it can be unreliable to depend on the annotation to get homologous sequences.
Nevertheless, the list should contain the SwissProt entry for CALM_HUMAN which
is what you want.
Questions
1. Can you think of a better way to find other mammalian calmodulin genes ?
2. If you do a search in SwissProt for "calmodulin" using the [AllText] descriptor
instead of [Description] you find many more entries, why do you think you get
more entries under this search?
3. There are more entries in SwissProt under [Organism] dog than [Author] dog, but
more for [Author] wolf than [Organism] wolf. Why do you think this is so ?
16
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
4. Searching [Organism] mouse in SwissProt yields some plant sequences: prove this
by finding sequences matching [Organism] mouse & [Taxon] viridiplantae.
Why is this so? (Clue: append wildcard *).
You should be able to reveal the full SwissProt entry for any protein sequence. If
you do this you will see several (? blue, underlined) hypertext links to related
databases. Almost certainly at least one of these will be EMBL and one to Medline.
Probably one will be the prosite motif database. If the 3-D structure is known, one
link will be to PDB. Investigate these other databases to get as much relevant
information as possible about your sequence.
Aside: Displaying 3-D structures is not “fitted as standard” on all terminals. You
may need to get a copy of the RasMol 3-D structure viewer and install it in such a
way that your Netscape/IE will recognise it and connect suitable (3-D sequence) file
to it.
To display a PDB entry of 3-D coordinates as a rotatable, colorable model you need
to click on the [save] button.
The change the "use mime type" choice-box to
chemical/x-pdb and then click on the [save] box. This should fire up CHIME a
WWW implementation of RasMol.) Your mileage may vary!
It is this, interlinked databases, aspect of SRS which gives it a large part of its
power. You can extend your search to include other sequences related in some
particular (or peculiar!) way. The Prosite link allows you to find members of a
protein family. The EMBL link allows you to find the introns and the intron splice
junctions, not to mention the ribosome-binding site, the stop codon and the journal
reference for the original sequence. The Medline link will give you an abstract etc.
You will probably find that:
The PubMed server at http://www.ncbi.nlm.nih.gov/Entrez/ is a far better tool for
browsing Medline that what is offered with SRS. Especially powerful is its facility
for finding [Related entries].
Additional questions:
17
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
“Effective researchers know how to find things out”
1. Who submitted the serum amyloid A (SAA) gene sequence for Canis familiaris?
2. What prosite motif defines the recA family of prokaryotic proteins? Which
Dublin-based phylogeneticists used multiple-sequence alignment to define this
motif?
3. What are the first and last 5 bases in the intron of the yeast actin gene with EMBL
accession number V01288?
4. What is the map position of one of the human SAA genes (SwissProt: P02735)?
What cross-reference database is most likely to have map position?
5.
What
mutation
at
what
position
causes
phenylketonuria
(PKU)?
(hint: EMBL K03020) but then try SwissProt: P00439.
6. What bases define the ribosome binding site of the Bacteroides fragilis glnA
gene? Perhaps start from the E.coli homolog SwissProt: P06711.
7. Why is the name Saarinen associated with life-threatening cardiac arrythmias?
(Hint: not because of architectural flaws...try voltage gated potassium
channels)
8. Are there more publicly available DNA sequences from Rodents or Prokaryotes?
What about protein sequences?
9. Get a sample of mammalian introns. See what common features they have?
Think how these common features might help splicing out the introns.
18
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
Entrez - http://www.ncbi.nlm.nih.gov/Entrez/
Entrez is the US equivalent of SRS and is available from the NCBI webpage. You
will most likely be familiar with Entrez for interrogating Medline, but the same
engine can be pointed at DNA and protein databases. It is handy if you are familiar
with the Entrez system and you want a sequence whose name or accession number
you already know. At the top of the Entrez page change the Search [__] choice box
from PubMed to the appropriate sort of database – the available options are listed on
the Entrez page. If you want the sequence alone – to paste into some analysis page –
change the Display [__] choice box to FASTA then click on [Save] or [Display]
depending on whether you want a permanent or transitory copy of you proteins.
Entrez has a more complex syntax for less straight forward queries.
19
M.Sc. in Molecular Medicine Bioinformatics Course
DAY 1: Nucleic Acid Sequence Analysis
TOPICS:
1. Translating DNA in 6 frames.
2. Reverse complement & other tools.
3. Calculating some properties of DNA/RNA sequences.
4. Primer design.
5. Gene prediction.
6. Alternative splicing.
7. Promoter characterisation.
8. Other resources.
20
Feb. 2005
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
1) Translating DNA in 6-frames:
Translate tool - http://www.expasy.ch/tools/dna.html
This tool allows the 6-frame translation of a nucleotide (DNA/RNA) sequence to a
protein sequence in order to locate open reading frames in your sequence.
ƒ
Go to URL above.
ƒ
Paste your sequence in the box provided & click “TRANSLATE
SEQUENCE”.
ƒ
You can choose 3 options
o Verbose – puts Met & Stop to highlight start & stop codons.
o Compact – useful if you want to use output in other programs.
o Includes nucleotide sequence – nucleotide sequence is above the
translation.
ƒ
This returns a 6-frame translation of your sequence. You can then choose the
correct frame.
See Appendix II for the genetic code
2) Reverse Complement & other tools:
There are many cases where you might want to obtain the reverse complement of a
DNA sequence, for example the reverse complement is needed as a negative control
when doing a DNA hybridisation experiment.
Search launcher at Baylor College –
http://searchlauncher.bcm.tmc.edu/seq-util/seq-util.html
This tool contains a number of different applications for nucleic acid sequence
analysis: For each application you can click on the following [H] [O] [P] [E] =
[H]:Help/description; [O]:full Options form; [P]:search Parameters; [E]:Example
search. On all the Baylor pages (and everywhere else possible) it is important to
investigate the options [O] to see a) what are the defaults and b) what options seem
worth changing. The following programs are available:
Readseq:
Converts nucleic acid/protein sequences between any of 30 different formats. It is
often appropriate to convert to FASTA format. A large number of input formats are
permitted. See help for details [H].
RepeatMasker:
21
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
RepeatMasker is a program that screens DNA sequences for interspersed repeats
known to exist in mammalian genomes as well as for low complexity DNA
sequences. The output of the program is a detailed annotation of the repeats that are
present in the query sequence as well as a modified version of the query sequence in
which all the annotated repeats have been masked (replaced by Ns). On average,
over 40% of a human genomic DNA sequence is masked by the program. This is
important in primer design so that you do not design a primer that spans a region
with repeats. It is also important before doing a homology search as repeats in your
sequence may hit other repeats in the genome (although BLAST now does this for
you).
Primer Selection -PCR primer selection (See primer design later).
WebCutter- restriction maps using enzymes w/ sites >= 6 bases.
6 Frame Translation - translates a nucleic acid sequence in 6 frames.
Reverse Complement - reverse complements a nucleic acid sequence.
Reverse Sequence - reverses sequence order.
Sequence Chopover - cut a large protein/DNA sequence into smaller ones with
certain amounts of overlap.
HBR - Finds E.coli contamination in human sequences.
Exercise: Paste in your own sequence of interest or alternatively examine an
example output for each application by clicking [E] beside each program. Pay
particular attention to the options available: these will give you clues about standard
practice.
22
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
3) Oligo Calculator - http://www.pitt.edu/~rsup/OligoCalc.html
Tool to calculate the length, %GC content, Melting temperature (Tm) the
midpoint of the temperature range at which the nucleic acid strands separate,
Molecular weight, & what an OD = 1 is in picoMolar of your input nucleic acid
sequence.
Many of these parameters are useful in primer design (see next section) and in other
areas of molecular biology.
ƒ
Go to URL above.
ƒ
Paste your sequence in the box provided & click “Calculate”.
Example:
>gi|10834993|ref|NM_000641.1| Homo sapiens interleukin 11 (IL11), mRNA
Length = 2281
% GC content = 55
Tm = 87 °C
Molecular Weight = 704856 daltons (g/M)
OD of 1 = 41 picoMolar
4) Primer design - Originally written in Jan 2002 by Dr. Norma O’Donovan
(Thanks!).
The recommended site (although there are several others available on the web) is
GeneFisher:
http://bibiserv.techfak.uni-bielefeld.de/genefisher/help/wwwgfdoc.html
The submission form:
http://bibiserv.techfak.uni-bielefeld.de/cgi-bin/gf_submit?mode=STARTUP&sample=dna
The input form is straightforward and well documented.
Primer Design Tips:
ƒ
Primer Length: usually between 18 and 24 base pairs
ƒ
% GC: Optimum GC content is 45-55%
ƒ
Annealing Temperature: Should be between 55oC and 65oC and ideally the
annealing temperature of the 2 primers should be similar. A quick equation
(Wallace formula) for calculating the annealing temperature of the primer is:
2 x (no. of As + Ts) + 4 x (no. of Gs + Cs)
The lower of the 2 primer annealing temperatures is the highest temperature that
can be used for annealing. (Usually when optimising PCR you would start with an
annealing temp. a few degrees below the Tm of the primers).
23
M.Sc. in Molecular Medicine Bioinformatics Course
ƒ
Feb. 2005
G/C clamps: The 3’ end of the primer should be able to form G/C clamps, i.e.
several consecutive G/C or C/G base pairs, between the 3' end of the primer and the
template DNA.
ƒ
Length of PCR product: The optimum size is 100 – 500 base pairs for
conventional PCR. Shorter products can be used for real time PCR or longer
products can be amplified using special polymerases.
Things to avoid:
1. Complimentarity within a primer or between 2 primers (especially in the ends), used
in the same reaction, as this may cause primer dimers.
2. Strings of a single nucleotide (more than 3).
3. Non-specific binding of primers to related sequences – check the specificity of the
primers by doing a BLAST search of the database (non-redundant and genomic)
with each of the primer sequences.
Primers for RT-PCR
The same rules as above apply but there are a few extra considerations. If you are
doing RT-PCR with total RNA there may be genomic DNA contamination present in
the RNA. (You can DNase treat to remove it or purify poly-A mRNA). If it is not
removed you must ensure that your primers specifically amplify the cDNA
(complementary to mRNA). Ideally the primers should not amplify the genomic
DNA at all but if that is not possible the genomic product should be distinguishable
from the cDNA product on a gel, based on size. Therefore the primers must span at
least one intron in the genomic DNA. To identify the position of introns in the
sequence align the mRNA sequence with the genomic sequence using a pairwise
BLAST sequence alignment (http://www.ncbi.nlm.nih.gov/blast/bl2seq/bl2.html).
Alternatively,
for
human
or
mouse
sequences,
on
the
UCSC
website
(http://genome.ucsc.edu/) you can do a BLAT search with the mRNA which will
identify the intron/exon structure of the gene.
Example:
Intron 1
Intron 2
100bp
400bp
Exon 1
Exon 2
Exon 3
Forward 2
Genomic
DNA
100bp
150bp
F2
cDNA
Intron 3
Forward 1
800bp
Exon 4
150bp
F1
Exon 1 Exon 2 Exon 3 Exon 4
24
R1
200bp
Reverse 1
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
If the forward and reverse primers are designed in exon 4 the PCR product obtained
from the cDNA will be the same size as the genomic PCR product. If the forward
primer is in exon 1 and the reverse primer is in exon 4 the cDNA product will be
approx. 600bp whereas the PCR product from genomic DNA would be about
1900bp, which probably wouldn’t be amplified in conventional PCR.
25
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
5) Gene Prediction
Gene prediction is an area under intensive research in bioinformatics and an entire
course could be dedicated to it alone. Here we will introduce the GENSCAN
program since it was one of the major programs used to predict genes in the human
genome. This program should be useful in predicting genes in most vertebrate
species, although caution should be used when dealing with other species especially
prokaryotes where other programs are more suitable.
For links to other programs used in gene prediction take a look at:
The Institute for Genomic Research - http://www.tigr.org/software/
OR
The Deambulum Nucleic Acids Sequence Analysis page at Infobiogen http://www.infobiogen.fr/services/deambulum/english/prog2.html - STRUC
GENSCAN - http://genes.mit.edu/GENSCAN.html
Program designed to predict complete gene structures, including exons, introns,
promoter and poly-adenylation signals, in genomic sequences.
For information about Genscan, click here link. It will tell more…
ƒ
Paste your sequence in the box provided & change the “print options:” to
“Predicted CDS and peptides”. Other defaults are OK.
ƒ
Click “Run GENSCAN”.
ƒ
The Genscan prediction for the genomic region around human hepcidin is
shown below. Genscan predicts that the initial exon begins at position 2566
and ends at 2477, then there is an intron, then there is the 2nd exon (355 –
296). The 3rd exon is at 206bp – 102bp. Genscan also predicts the termination
signal and the polyA tail. The gene is on the complementary strand.
ƒ
The predicted coding sequence (CDS) & predicted peptide match exactly the
known sequences for hepcidin.
ƒ
READ the output carefully then try your own sequence.
It should be
reasonably self-explanatory.
Example: chr19:42030157-42032793 genomic sequence of human hepcidin.
26
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
GENSCANW output for sequence 09:24:40
GENSCAN 1.0
Date run: 5-Aug-103
Time: 09:24:40
Sequence 09:24:40 : 2637 bp : 53.92% C+G : Isochore 3 (51 - 57 C+G%)
Parameter matrix: HumanIso.smat
Predicted genes/exons:
Gn.Ex Type S .Begin ...End .Len Fr Ph I/Ac Do/T CodRg P.... Tscr..
----- ---- - ------ ------ ---- -- -- ---- ---- ----- ----- -----1.04
1.03
1.02
1.01
PlyA
Term
Intr
Init
-
31
206
355
2566
26
102
296
2477
6
105
60
90
2
1
1
0
0
0
118
119
99
42
89
94
49 0.878
-16 0.646
173 0.999
1.05
2.01
1.02
17.44
Click here to view a PDF image of the predicted gene(s)
Click here for a PostScript image of the predicted gene(s)
Predicted peptide sequence(s):
Predicted coding sequence(s):
>09:24:40|GENSCAN_predicted_peptide_1|84_aa
MALSSQIWAACLLLLLLLASLTSGSVFPQQTGQLAELQPQDRAGARASWMPMFQRRRRRD
THFPICIFCCGCCHRSKCGMCCKT
>09:24:40|GENSCAN_predicted_CDS_1|255_bp
atggcactgagctcccagatctgggccgcttgcctcctgctcctcctcctcctcgccagc
ctgaccagtggctctgttttcccacaacagacgggacaacttgcagagctgcaaccccag
gacagagctggagccagggccagctggatgcccatgttccagaggcgaaggaggcgagac
acccacttccccatctgcattttctgctgcggctgctgtcatcgatcaaagtgtgggatg
tgctgcaagacgtag
Explanation
Gn.Ex : gene number, exon number (for reference)
Type : Init = Initial exon (ATG to 5' splice site)
Intr = Internal exon (3' splice site to 5' splice site)
Term = Terminal exon (3' splice site to stop codon)
Sngl = Single-exon gene (ATG to stop)
Prom = Promoter (TATA box / initation site)
PlyA = poly-A signal (consensus: AATAAA)
S
: DNA strand (+ = input strand; - = opposite strand)
Begin : beginning of exon or signal (numbered on input strand)
End
: end point of exon or signal (numbered on input strand)
Len
: length of exon or signal (bp)
Fr
: reading frame (a forward strand codon ending at x has frame x mod
3)
Ph
: net phase of exon (exon length modulo 3)
I/Ac : initiation signal or 3' splice site score (tenth bit units)
Do/T : 5' splice site or termination signal score (tenth bit units)
CodRg : coding region score (tenth bit units)
P
: probability of exon (sum over all parses containing exon)
Tscr : exon score (depends on length, I/Ac, Do/T and CodRg scores)
Comments
The SCORE of a predicted feature (e.g., exon or splice site) is a
log-odds measure of the quality of the feature based on local sequence
properties. For example, a predicted 5' splice site with
score > 100 is strong; 50-100 is moderate; 0-50 is weak; and
below 0 is poor (more than likely not a real donor site).
The PROBABILITY of a predicted exon is the estimated probability under
GENSCAN's model of genomic sequence structure that the exon is correct.
This probability depends in general on global as well as local sequence
properties, e.g., it depends on how well the exon fits with neighboring
exons. It has been shown that predicted exons with higher probabilities
are more likely to be correct than those with lower probabilities.
27
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
6) Splice site prediction / Alternative splicing
Introduction to splicing:
Taken from http://www.bioinformatics.ucla.edu/HASDB/
The first requirement for proper splicing is some way to distinguish exons from
introns. This is accomplished using certain base sequences as signals. These
consensus base sequences, as they are known, allow the spliceosome (the cellular
machinery that does the splicing) to identify the 5' and 3' ends of the intron. For
example, in eukaryotes, the base sequence of an intron begins with 5' GU, and ends
with 3' AG [See Figure below] These sequences base pair with complementary
spliceosomal RNA so that the pre-mRNA is aligned properly with the spliceosome.
Each species has additional bases associated with these splice sites, but GU and AG
are the only ones that are conserved across all eukaryotes. For example, the
consensus sequence at the 5' splice site of vertebrate introns is AGGUAAGU
(Stryer, 1995). Introns also have another important sequence signal called a branch
site containing a tract of pyrimidine bases and a special adenine base, usually
approximately 50 bases upstream from the 3' splice site. More information on the
mechanism of splicing is available at the above website but will not be discussed in
this course.
Alternative splicing:
The central dogma of molecular biology was that 1 gene = 1 protein, however more
and more examples have been discovered where this is not the case and multiple
possible mRNA transcripts can be produced from 1 gene and if translated these
transcripts can code for very different proteins. This phenomenon is known as
alternative splicing. There are 4 basic ways in which alternative splicing can occur:
28
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
1) Splice / Don't Splice
First, an intron can either be spliced out of the RNA (as in the simple model of RNA
splicing), or it can be retained and included in the coding region of the RNA. This
phenomenon is known as splice/don't splice and the choice could have several
different results. For example, if the intron includes an in-frame stop codon, then a
splice variant that includes the intron may result in a shorter, non-functional protein.
If the intron is spliced out, then the resultant mRNA would have an open reading
frame which would be translated into the functional protein. In this case, the
alternative splicing acts like an on/off switch. Another potential outcome of splice/
don't splice is simply that two functional mRNAs could be made, each with a unique
base sequence. This would create two different proteins, each with a unique amino
acid sequence, and possibly with different but related functions. In this case, the
alternative splicing acts like a switch between producing mRNAs coding for two
different proteins.
2) Competing 5' or 3' Splice Sites
A second mechanism for alternative splicing is the presence of competing 5' splice
sites for one 3' site within one intron. Alternatively, there can be competing 3' splice
sites for one 5' site within one intron. The competing site that is closest to the other
end of the intron is called the proximal site, while the competing site that is farthest
from the other end of the intron is called the distal splice site. The selection of each
splice site would result in mRNAs that differed by the stretch of bases between the
proximal and distal splice sites. Like the possible outcomes of splice/don't splice,
competing 5' or 3' sites could act like an on/off switch, or this mechanism could act
like a switch between the production of mRNAs coding for two different proteins.
29
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
3) Exon Skipping
A third mechanism for alternative splicing is called exon skipping. This occurs when
an exon that would usually be included in the mature mRNA is spliced out with the
neighboring introns, and is therefore skipped. There can also be multiple exon
skipping in which more than one exon (with intervening introns) is skipped at once.
This mechanism has the potential to produce many different mRNA's. For example,
if a gene has 8 exons, one variant might include all of them, while another variant
skips exon 7, and another variant skips exons 2 and 3, and yet another variant skips
exons 4 and 5, etc... Hence, exon skipping has the potential to lead to many different
mRNAs that could function as on/ off switches or as a switch between maturation of
mRNAs for different proteins.
4) Mutually Exclusive Exons
A mechanism of alternative splicing related to exon skipping is called mutually
exclusive exons. In this case, the mRNA would include either exon 1 or 2, not both.
For example, if a gene has 4 exons, one splice variant might include exons 1, 2 and
4, while another splice variant might include exons 1, 3 and 4. Again, there is the
potential for an on/off switch and for a switch between mRNAs for two proteins. It
is important to note that more than one of these modes of splicing could happen at
the same time. For example, it is possible that a gene could be alternatively spliced
through both exon skipping and competing 5' splice sites at the same time. It is also
important to note that research into alternative splicing is in the early stages, and that
other modes of alternative splicing may be discovered in the future.
30
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
The Human Alternative Splicing Database at UCLA –
http://www.bioinformatics.ucla.edu/HASDB/
Used ESTs to locate alternative splices. Project has resulted in a publication of over
six thousand alternatively spliced isoforms of human genes.
You can search the database using any of the following identifiers:
ƒ
ƒ
ƒ
ƒ
ƒ
Gene Symbol: search by a gene symbol (e.g. TCN1)
UniGene Sequence Identifier: search by a UniGene sequence identifer (e.g.
Hs#S3362)
UniGene Cluster Identifier: search by a UniGene cluster identifier (e.g.
Hs.2012)
Gene Title: search by a gene title (e.g. transcobalamin I (vitamin B12
binding protein, R binder family) )
GeneBank Sequence Identifier: search by a GeneBank sequence identifier
(e.g. J05068)
You can also search for tissue-specific alternative transcripts by clicking “Search By
Tissue”.
Example: HLA-G (gene symbol)
HLA-G is a nonclassical MHC 1 molecule that inhibits NK cell function. At least 7
variants have been characterized and these variants may have very different
functions. Search HLA-G at HASDB to view the variants determined by this project.
NOTE: On day 2 we will see how the genome browser at UCSC can be used to help
identify alternative splice variants.
31
M.Sc. in Molecular Medicine Bioinformatics Course
32
Feb. 2005
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
7) Promoter Analysis & Recognition:
A promoter is a sequence that is used to initiate and regulate transcription of a gene.
Most protein-coding genes in higher eukaryotes have polymerase II dependent
promoters.
Features of pol II promoters:
• Combination of multiple individual regulatory elements.
• Most important elements are transcription factor binding sites.
• CAAT or TATA boxes are neither necessary nor sufficient for promoter
function.
• In many cases, order and distances of elements are crucial for their function.
•
Sequences between elements within a promoter are usually not conserved
and of no known function.
Figure 14-19: Taken from “Modern Genetic Analysis” (W.H. Freeman & Company).
The promoter region in higher eukaryotes. The TATA box is located approximately 30 base
pairs from the mRNA start site. Usually, two or more promoter-proximal elements are
found 100 and 200 bp upstream of the mRNA start site. The CCAAT box and the GC-rich
box are shown here. Other upstream elements include the sequences GCCACACCC and
ATGCAAAT.
Promoter identification
Polymerase II promoters are generally defined as the region of a few hundred base
pairs located directly upstream of the site of initiation of transcription. (More distal
regions and parts of the 5' UTR may also contain regulatory elements and may be
part of the promoter). The exact length of a promoter can often only be defined
experimentally. However, for an initial in silico analysis it may be sufficient (and
also necessary) to restrict the region to about 300 to 1000 bp upstream of the
transcription start site. Therefore, identification of the transcription start site directly
leads to the location of the promoter of a gene. The transcription start site can be
defined by mapping a 5' full-length mRNA/cDNA (including the complete 5' UTR)
to the genomic sequence. The second possibility is to use PromoterInspector, a tool
that is able to predict promoter regions in genomic sequences.
33
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
PromoterInspector – http://www.genomatix.de/shop/evaluation.html
PromoterInspector is a program (part of the Genomatix suite of programs) that
predicts eukaryotic pol II promoter regions with high specificity (~85%) in
mammalian genomic sequences. PromoterInspector focuses on the genomic context
of promoters rather than their exact location. The sensitivity of PromoterInspector is
about 50% which means that the current version predicts about every second
promoter in the genome. Therefore, your promoter may not be found.
PromoterInspector predicts the approximate location of a promoter region and not
the exact location of the Transcription Start Site (TSS). The predicted regions may
contain the promoter or overlap with the promoter. The strand orientation of the
predicted promoter region can only be derived from the location of the
corresponding gene. PromoterInspector predicts promoter regions by identification
of the conserved promoter context independently of the occurrence of specific
elements like CCAAT or TATA boxes. To identify transcription factor binding sites
in a promoter you can use MatInspector professional (see below).
ƒ
Go to the Genomatix link above.
ƒ
Before you can use this program you will have to register and obtain a user
name and password. Do this by filling in the form & clicking “Register”
ƒ
Once you have obtained a user name and password by e-mail you can use
PromoterInspector. Please note that this is commercial software and
academics receive only limited access. Full access is expensive.
ƒ
Log into GenomatixSuite and click on PromoterInspector.
ƒ
There are two ways to supply a set of input sequences: Either enter your
sequences directly into the form or, if your browser supports this option, a
sequence file can be uploaded.
ƒ
In both cases, the input sequences must be in one of the supported formats.
ƒ
Please note that the web version of PromoterInspector accepts up to 100000
base pairs sequence input! To keep computing times reasonable the program
should NOT be used more than 3 times per day per user!
ƒ
Supply your VALID email address & click “Start PromoterInspector”.
ƒ
When the analysis is finished, an email with the URL of the results will be
sent to this address. You can then point your browser to this address.
34
M.Sc. in Molecular Medicine Bioinformatics Course
ƒ
Feb. 2005
The result will be available for 30 days on the server. After that period it will
be deleted! Depending on the number and length of sequences in your input,
the computation may take a while.
ƒ
PromoterInspector creates an output file that contains a list of promoter
regions for every sequence. Start and end of the regions correspond to sense
strand numbering. Please note that predictions are not strand specific.
Example: >chr15:56167697-56191947 (reverse complemented) genomic sequence
around the human ADAM 10 gene.
ƒ
To extract the sequence in between the coordinates 4841 & 5836 I suggest
pasting the entire sequence into WORD & using the line numbering function.
Since each line has 50 letters you can divide the coordinates by 50 to
determine the lines on which your promoter sequence is located. The
promoter sequence can then be pasted into MatInspector to look for
transcription factor binding sites.
MatInspector professional: Identification of transcription factor binding sites
35
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
MatInspector professional is a tool that utilizes a library of matrix descriptions for
transcription factor binding sites to locate matches in sequences of unlimited length.
A large library of predefined matrix descriptions for protein binding sites exists and
has been tested for accuracy and suitability. Similar and/or related matrices have
been grouped into matrix families. The matrix library contains 592 weight matrices
in five species groups (fungi, insects, plants, vertebrates, and miscellaneous).
Transcription factor binding sites (TF-sites)
Individual TF-sites build the basis of the promoter. These are relatively short
stretches of DNA (10 - 20 nucleotides), sufficiently conserved in sequence to allow
specific recognition by the corresponding transcription factor. TF-acquisition by
DNA binding is the sole function of a TF-site! TF-sites are generally best described
by nucleotide weight matrices. MatInspector professional is a good tool for detection
of TF-sites in DNA sequences and benefits from a large library of precompiled and
quality checked nucleotide weight matrices.
Using MatInspector professional:
Once you have identified the promoter region using the PromoterInspector program
you can then search for potential transcription factor binding sites.
ƒ
Go to the link for the program above.
ƒ
You can use the same user name and password that you obtained for
PromoterInspector.
ƒ
Click on “MatInspector”.
ƒ
Enter your sequence in the box provided in one of the supported formats.
ƒ
You do not need to change any of the other parameters until the “Library
Selection” section.
ƒ
Here you can choose whether to search for transcription factor binding sites
that have pre-compiled weight matrices (i.e. TFs recognised by the program)
or you can search your own string of letters by choosing “User-defined
IUPAC string (IUPAC)”. You can also cut your sequence with a variety of
restriction enzymes but we will not be dealing with this feature here.
ƒ
Only the IUPAC symbols ABCDGHKMNRSTUVWY can be used (e.g. R is
A or G), all other letters are ignored. Specify the maximum number of
36
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
mismatches allowed in matches to the string. The number of mismatches
should not exceed 50% of the string-length.
ƒ
Click “Continue”.
ƒ
On the next page you can change matrix and output parameters.
ƒ
You can read about these parameters in detail by clicking the [?] at the top of
the page.
ƒ
Choose “Vertebrates” as the matrix group, change the core similarity to 1.00
(at least for 1st run), sort matches by quality & leave the other defaults as they
are.
ƒ
Fill in your e-mail address and click “Submit Query”.
ƒ
The output is shown below. Remember because a transcription factor binding
site exists does not mean that it is functional. This program just gives you a
starting point to experimentally characterise your promoter sequence.
Example: promoter region for human ADAM 10 gene identified by
Pro
moterInspector. Coordinates 4750-5000bp (TSS @ 5000bp).
Other Resources on the web for nucleic acid sequence analysis
37
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
There are many resources available on the web for nucleic acid sequence analysis for
a starting point take a look at:
Deambulum - http://www.infobiogen.fr/services/deambulum/english/menu.html
38
M.Sc. in Molecular Medicine Bioinformatics Course
Day 1: Protein Sequence Analysis
TOPICS
1. Physico-chemical properties.
2. Cellular localization.
3. Signal peptides.
4. Transmembrane domains.
5. Post-translational modifications.
6. Motifs & domains.
7. Secondary structure.
8. Other resources.
39
Feb. 2005
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
ExPASy - http://www.expasy.ch/
The ExPASy (Expert Protein Analysis System) protein and proteomics server of the
Swiss Institute of Bioinformatics (SIB) is dedicated to the analysis of protein
sequences and structures. Besides the tools that we will introduce in this manual
there are many other applications available at this website that you should take some
time to have a look at.
1) Physico-chemical properties:
ProtParam tool
- http://www.expasy.ch/tools/protparam.html
Calculates lots of physico-chemical parameters of a protein sequence. The computed
parameters include the molecular weight, theoretical pI, amino acid composition,
atomic composition, extinction coefficient, estimated half-life, instability index,
aliphatic index and grand average of hydropathicity (GRAVY)
Example: Human BRCA 1
You can paste the gene sequence from the Course Website.
ƒ
At ExPASy Æ “Proteomics and sequence analysis tools” Æ “Primary
structure analysis”.
ƒ
Click on the “ProtParam” link.
ƒ
Paste your sequence in the box provided
ƒ
The sequence must be written using the one letter amino acid code:
ƒ
Press the “Compute parameters” button.
ƒ
The output for this sequence is shown below.
Number of amino acids: 1863
Molecular weight: 207720.8
Theoretical pI: 5.29
Amino acid composition:
Ala (A) 84
4.5%
Arg (R) 76
4.1%
Etc etc
Thr (T) 111
6.0%
Trp (W) 10
0.5%
Tyr (Y) 31
1.7%
Val (V) 101
5.4%
Asx (B) 0
0.0%
Glx (Z) 0
0.0%
40
M.Sc. in Molecular Medicine Bioinformatics Course
Xaa (X) 0
Feb. 2005
0.0%
Total number of negatively charged residues (Asp + Glu): 283
Total number of positively charged residues (Arg + Lys): 213
Atomic composition:
Carbon
C
8908
Hydrogen H
14246
Nitrogen N
2554
Oxygen
O
3014
Sulfur
S
74
Formula: C8908H14246N2554O3014S74
Total number of atoms: 28796
Extinction coefficients:
Conditions: 6.0 M guanidium hydrochloride
0.02 M phosphate buffer
pH 6.5
Extinction coefficients are in units of M-1 cm-1 .
The first table lists values computed assuming ALL Cys residues appear as half cystines, whereas the
second table assumes that NONE do.
Ext. coefficient
Abs 0.1% (=1 g/l)
276
nm
102140
0.492
278
nm
102194
0.492
279
nm
100935
0.486
280
nm
99220
0.478
282
nm
95840
0.461
Ext. coefficient
Abs 0.1% (=1 g/l)
276
nm
98950
0.476
278
nm
99400
0.479
279
nm
98295
0.473
280
nm
96580
0.465
282
nm
93200
0.449
Estimated half-life:
The N-terminal of the sequence considered is M (Met).
The estimated half-life is: 30 hours (mammalian reticulocytes, in vitro).
>20 hours (yeast, in vivo).
>10 hours (Escherichia coli, in vivo).
Instability index:
The instability index (II) is computed to be 54.68
This classifies the protein as unstable.
Aliphatic index: 69.01
Grand average of hydropathicity (GRAVY): -0.785
41
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
2) Cellular localization:
PSORT - http://psort.nibb.ac.jp/form2.html
PSORT, a program to predict the subcellular localization sites of proteins from their
amino acid sequences. This program makes use of the fact that proteins destined for
particular subcellular localizations have distinct amino acid properties particularly in
their N-terminal regions. These properties can be used to predict whether a protein is
localized in the cytoplasm, nucleus, mitochondria, or is retained in the ER, or
destined for the lysosome (vacuolar) or the peroxisome. There is a detailed page of
output that we can probably ignore. At the end of the output the percentage
likelihood of the subcellular localization is given. If you want to learn more about
the output and how subcellular localization is determined please see the user manual
at:
http://psort.nibb.ac.jp/helpwww2.html
Example: Human ETS-1 protein.
ƒ
At ExPASy Æ “Post-translational modification prediction”.
ƒ
Click on the “PSORT” link.
ƒ
For animal/yeast sequences click the link to “PSORT II Prediction”.
ƒ
Paste your sequence in the box provided.
ƒ
The sequence must be written using the one letter amino acid code:
ƒ
Press the submit button.
ƒ
The output for this sequence is shown below.
There are a number parameters measured by this program which you can read about
as links from the output file. By scrolling to the bottom of the output you can see the
probability that this sequence is nuclear, cytoplasmic, peroxisomal, vacuolar or
cytoskeletal. PSORT predicts that ETS-1 is nuclear with a high probability. The fact
that ETS-1 is localized in the nucleus has been previously experimentally
determined.
Results of Subprograms
PSG:
a new signal peptide prediction method
N-region: length 8; pos.chg 2; neg.chg 1
H-region: length 6; peak value
1.89
PSG score: -2.51
42
M.Sc. in Molecular Medicine Bioinformatics Course
GvH:
Feb. 2005
von Heijne's method for signal seq. recognition
GvH score (threshold: -2.1): -10.14
possible cleavage site: between 54 and 55
>>> Seems to have no N-terminal signal peptide
ALOM: Klein et al's method for TM region allocation
Init position for calculation: 1
Tentative number of TMS(s) for the threshold 0.5:
number of TMS(s) .. fixed
PERIPHERAL Likelihood = 3.61 (at 98)
ALOM score:
3.61 (number of TMSs: 0)
0
MITDISC: discrimination of mitochondrial targeting seq
R content:
0
Hyd Moment(75): 6.78
Hyd Moment(95): 6.47
G content:
0
D/E content:
2
S/T content:
3
Score: -6.01
Gavel: prediction of cleavage sites for mitochondrial preseq
cleavage site motif not found
NUCDISC: discrimination of nuclear localization signals
pat4: none
pat7: none
bipartite: none
content of basic residues: 11.3%
NLS Score: -0.47
KDEL: ER retention motif in the C-terminus: none
ER Membrane Retention Signals: none
SKL: peroxisomal targeting signal in the C-terminus: none
SKL2: 2nd peroxisomal targeting signal:
none
VAC: possible vacuolar targeting motif: none
RNA-binding motif: none
Actinin-type actin-binding motif:
type 1: none
type 2: none
NMYR: N-myristoylation pattern : none
Prenylation motif: none
memYQRL: transport motif from cell surface to Golgi: none
Tyrosines in the tail: none
Dileucine motif in the tail: none
checking 63 PROSITE DNA binding motifs:
Ets-domain signature 1 (PS00345):
LWQFLLELL at 337
Ets-domain signature 2 (PS00346):
KPKMNYEKLSRGLRYY at 381
*** found ***
*** found ***
checking 71 PROSITE ribosomal protein motifs:
none
checking 33 PROSITE prokaryotic DNA binding motifs:
none
NNCN: Reinhardt's method for Cytplasmic/Nuclear discrimination
Prediction: nuclear
Reliability: 55.5
COIL: Lupas's algorithm to detect coiled-coil regions
total: 0 residues
43
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
Results of the k-NN Prediction
k = 9/23
73.9 %: nuclear
13.0 %: cytoplasmic
4.3 %: peroxisomal
4.3 %: vacuolar
4.3 %: cytoskeletal
>> prediction for QUERY is nuc (k=23)
3) Signal peptides:
Proteins destined for secretion, operation with the endoplasmic reticulum, lysosomes
and many transmembrane proteins are synthesized with leading (N-terminal) 13 – 36
residue signal peptides.
SignalP - http://www.cbs.dtu.dk/services/SignalP/
The SignalP WWW server can be used to predict the presence and location of signal
peptide cleavage sites in your proteins. It can be useful to know whether your protein
has a signal peptide as it indicates that it may be secreted from the cell. Furthermore,
proteins in their active form will have their signal peptides removed, if you can
determine the length of the signal peptide then you can calculate the size of the
protein minus the signal peptide.
Example: Human Beta-defensin; sp|Q09753|BD01_HUMAN
ƒ
At ExPASy Æ “Post-translational modification prediction”.
ƒ
Click on the “SignalP” link.
ƒ
Paste your sequence in the box provided
ƒ
The sequence must be written using the one letter amino acid code:
ƒ
It is recommend that the N-terminal part only (not more than 50-70 amino
acids) of the sequences is submitted. A longer sequence will increase the risk of
false positives and make the graphical output difficult to read. The new version
now automatically truncates input sequences.
ƒ
Choose one or more group of organisms for the prediction by clicking the
check-box next to the group(s):
ƒ
If no groups are indicated, predictions from all three groups will be returned.
ƒ
A graphical output (in Postscript format) of the prediction will be available, if
44
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
the "Include graphics"-button is checked.
ƒ
Press the "Submit sequence" button.
A WWW page will return the results when the prediction is ready. Response time
depends on system load. The output for this sequence is shown below
C score = raw cleavage site score
The output score from networks trained to recognize cleavage sites vs. other
sequence positions. Trained to be: High at position +1 after the cleavage site and low
at all other positions.
S score = signal peptide score
The output score from networks trained to recognize signal peptide vs. non-signalpeptide positions. Trained to be: High at position before the cleavage site and low at
all other positions.
Y score = combined cleavage site score
The prediction of cleavage site location is optimized by observing where the C-score
is high and the S-score changes from a high to a low value.
For each sequence, SignalP will report the maximal C, S, and Y scores, and the mean Sscore between the N-terminal and the predicted cleavage site. These values are used to
distinguish between signal peptides and non-signal peptides. If your sequence is
predicted to have a signal peptide, the cleavage site is predicted to be immediately
before the position with the maximal Y-score.
The Human beta-defensin protein has a predicted signal peptide from position 1 to 21
and a potential cleavage site exists between positions 21 and 22. These predictions
correspond exactly to the SWISS-PROT annotation for this protein (accession Q09753).
SignalP-NN result:
45
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
# data
>Sequence
length = 68
# Measure Position Value Cutoff
max. C
22
0.710
0.32
max. Y
22
0.761
0.33
max. S
14
0.998
0.87
mean S
1-21
0.943
0.48
D
1-21
0.852
0.43
# Most likely cleavage site between
signal peptide?
YES
YES
YES
YES
YES
pos. 21 and 22: ASG-GN
SignalP-HMM result:
# data
>Sequence
Prediction: Signal peptide
Signal peptide probability: 1.000
Signal anchor probability: 0.000
Max cleavage site probability: 0.818 between pos. 21 and 22
4) Transmembrane domains:
Tmpred - http://www.ch.embnet.org/software/TMPRED_form.html
The TMpred program makes a prediction of membrane-spanning regions and their
orientation. The algorithm is based on the statistical analysis of TMbase, a database
of naturally occurring transmembrane proteins. The prediction is made using a
46
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
combination of several weight-matrices for scoring. The presence of transmembrane
domains is an indication that the protein is located on the cell surface.
Example: Human chemokine receptor 4 protein sequence NP_003458.1
ƒ
At ExPASy Æ “Topology prediction”.
ƒ
Click on the link to Tmpred.
ƒ
Paste your sequence in the box provided in one of the supported formats e.g.
ƒ
plain text, SwissProt_ID or AC, etc.
ƒ
You may change the minimal and maximal length of the hydrophic part of the
transmembrane helix but unless you have reason to do so you should accept the
defaults i.e. 17 and 33. ~22 residues is the same length as the width of a lipid
bilayer.
ƒ
Click the “Run Tmpred” button to start the search.
ƒ
The output is given in 3 parts 1, 2 and 3 (see below).
ƒ
Part 1: lists all the significant predictions of possible transmembrane helices
ƒ
in this case there are 7 helices predicted but at this stage we do not know the
orientation of the helices so there are 2 tables, the first with the helices orientated
from the inside to the outside and vice versa for the second.
ƒ
Part 2: shows which inside->outside helices correspond to the outside -> inside
helices and indicates which orientation is most likely.
ƒ
Part 3: proposes the strongly preferred model for the transmembrane domain
ƒ
structure of the protein and also an alternative model.
ƒ
A graphic of the prediction is also available (not shown here)
ƒ
These predictions correspond well but not exactly to the SWISS-PROT
annotation for this protein (accession P30991).
Tmpred output
Sequence: MEG...HSS, length: 352
Prediction parameters: TM-helix length between 17 and 33
1. Possible transmembrane helices
The sequence positions in brackets denominate the core region.
Only scores above 500 are considered significant.
Inside to outside helices :
from
to
7 found
score center
47
M.Sc. in Molecular Medicine Bioinformatics Course
39
78
114
155
204
240
286
(
(
(
(
(
(
(
46)
85)
114)
157)
206)
240)
286)
62
105
133
175
223
261
305
(
(
(
(
(
(
(
Outside to inside
from
47 ( 47) 63 (
78 ( 78) 96 (
111 ( 114) 132 (
155 ( 157) 173 (
204 ( 204) 223 (
240 ( 242) 259 (
283 ( 286) 305 (
62)
103)
130)
173)
223)
259)
305)
1962
1623
1352
1716
2052
2840
1241
Feb. 2005
54
95
122
165
214
251
295
helices :
7 found
to
score center
63)
2568
55
96)
1331
86
132)
1740
122
173)
1197
165
223)
2404
214
259)
2037
251
305)
1703
294
2. Table of correspondences
Here is shown, which of the inside->outside helices correspond to which of the
outside->inside helices.
Helices shown in brackets are considered insignificant.
A “+”-symbol indicates a preference of this orientation.
A “++”-symbol indicates a strong preference of this orientation.
3978114155204240286-
Inside->outside
62 (24) 1962
105 (28) 1623 ++
133 (20) 1352
175 (21) 1716 ++
223 (20) 2052
261 (22) 2840 ++
305 (20) 1241
| outside->inside
|
47- 63 (17) 2568 ++
|
78- 96 (19) 1331
|
111- 132 (22) 1740 ++
|
155- 173 (19) 1197
|
204- 223 (20) 2404 ++
|
240- 259 (20) 2037
|
283- 305 (23) 1703 ++
3. Suggested models for transmembrane topology
These suggestions are purely speculative and should be used with extreme caution
since they are based on the assumption that all transmembrane helices have been
found.
In most cases, the Correspondence Table shown above or the prediction plot that is
also created should be used for the topology assignment of unknown proteins.
2 possible models considered, only significant TM-segments used
---Æ STRONGLY preferred model: N-terminus outside
7 strong transmembrane helices, total score : 14594
# from
to length score orientation
1
47
63 (17)
2568 o-I
2
78 105 (28)
1623 I-o
3 111 132 (22)
1740 o-I
4 155 175 (21)
1716 I-o
5 204 223 (20)
2404 o-I
6 240 261 (22)
2840 I-o
7 283 305 (23)
1703 o-I
----Æ alternative model
7 strong transmembrane helices, total score : 11172
# from
to length score orientation
1
39
62 (24)
1962 I-o
2
78
96 (19)
1331 o-I
3 114 133 (20)
1352 I-o
4 155 173 (19)
1197 o-I
5 204 223 (20)
2052 I-o
48
M.Sc. in Molecular Medicine Bioinformatics Course
6
7
240
286
259 (20)
305 (20)
2037 o-I
1241 I-o
49
Feb. 2005
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
5) Post-translational modifications:
After translation has occurred proteins may undergo a number of posttranslational
modifications. These can include the cleavage of the pro- region to release the active
protein, the removal of the signal peptide and numerous covalent modifications such
as, acetylations, glycosylations, hydroxylations, methylations and phosphorylations.
Posttranslational modifications such as these may alter the molecular weight of your
protein and thus its position on a gel. There are many programs available for
predicting the presence of posttranslational modifications, we will take a look at one
for the prediction of type O-glycosylation sites in mammalian proteins. Remember
these programs work by looking for consensus sites and just because a site is found
does not mean that a modification definitely occurs.
NetOGlyc - http://www.cbs.dtu.dk/services/NetOGlyc/
Prediction of type O-glycosylation sites in mammalian proteins.
This program
works by comparing the input sequence to a database of 299 known and verified
mucin type O-glycosylation sites extracted from O-GLYCBASE.
Example: Human CD1D sp|P15813|CD1D_HUMAN
ƒ
At ExPASy Æ “Post-translational modification”.
ƒ
Click on the link to “NetOGlyc”.
ƒ
Paste your sequence in the box provided in FASTA format.
ƒ
Check “generate graphics” and click the submit button.
ƒ
The output for this program is shown below (graphics not shown).
ƒ
This program predicts potential O-glycosylation sites at Threonine 64 and
Serine 214.
50
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
NetOGlyc 2.0 Prediction Results
Name: Sequence
Length: 335
MGCLLFLLLWALLQAWGSAEVPQRLFPLRCLQISSFANSSWTRTDGLAWLGELQTHSWSNDSDTVRSLKPWSQGTFSDQQ
WETLQHIFRVYRSSFTRDVKEFAKMLRLSYPLELQVSAGCEVHPGNASNNFFHVAFQGKDILSFQGTSWEPTQEAPLWVN
LAIQVLNQDKWTRETVQWLLNGTCPQFVSGLLESGKSELKKQVKPKAWLSRGPSPGPGRLLLVCHVSGFYPKPVWVKWMR
GEQEQQGTQPGDILPNADETWYLRATLDVVAGEAAGLSCRVKHSSLEGQDIVLYWGGSYTSMGLIALAVLACLLFLLIVG
FTSRFKRQTSYQGVL
...............................................................T................
................................................................................
.....................................................S..........................
................................................................................
...............
Name
Sequence
Sequence
Sequence
Etc etc
Sequence
Sequence
Sequence
Sequence
Sequence
Sequence
Residue
Thr
Thr
Thr
Name
Sequence
Sequence
Sequence
Sequence
Sequence
Sequence
Etc etc
Sequence
Sequence
Sequence
Sequence
Sequence
Sequence
Residue
Ser
Ser
Ser
Ser
Ser
Ser
Thr
Thr
Thr
Thr
Thr
Thr
Ser
Ser
Ser
Ser
Ser
Ser
No.
42
44
55
248
260
266
300
322
329
No.
18
34
35
39
40
57
284
285
298
301
323
330
80
160
240
320
80
160
240
320
Potential Threshold Assignment
0.0611
0.6493 .
0.0087
0.6573 .
0.0117
0.6491 .
0.0131
0.0089
0.0224
0.0147
0.0480
0.0639
0.5840 .
0.6578 .
0.6957 .
0.7357 .
0.7096 .
0.6021 .
Potential Threshold Assignment
0.0161
0.6211 .
0.0044
0.6673 .
0.0048
0.6717 .
0.0337
0.6118 .
0.0013
0.6521 .
0.0065
0.6484 .
0.0005
0.0082
0.0003
0.0007
0.0003
0.0052
0.6401 .
0.6389 .
0.6778 .
0.6924 .
0.6441 .
0.6277 .
Note: The new version of this server does not predict these sites. This is a good
lesson in the evolving nature of these servers and why validation at more than
one is a good idea.
51
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
6) Motifs and Domains
If you want to determine the function of a protein the first tool of choice is homology
searching (see day 2). Unless this finds you a match with a well characterized
protein comprehending the entire length of yours you should look for motifs and
domains in your protein. To determine if your protein sequence contains known
motifs or conserved domain structures you should search the protein against one of
the motif or profile databases. There are many of these available but we will discuss
ProfileScan (now called myHits), which allows you to search both the Prosite and
Pfam databases simultaneously. See the documentation for more details.
ProfileScan - http://hits.isb-sib.ch/cgi-bin/PFSCAN
Example: Human CFTR sp|P13569|CFTR_HUMAN
ƒ
Go to the URL above
ƒ
Paste your sequence in the box provided.
ƒ
The sequence must be written using the one letter amino acid code:
ƒ
Tick the motif databases you wish to search, other parameters should be OK.
ƒ
Press the “scan” button.
The output for this program is too large to show here, but it gives lots of detail about
motifs in the CFTR protein identifying potential: ABC transporters family signature;
ATP/GTP-binding site motif A (P-loop); Protein kinase C phosphorylation sites; Nglycosylation sites; Casein kinase II phosphorylation site; N-myristoylation sites;
cAMP- and cGMP-dependent protein kinase phosphorylation site; Bipartite nuclear
localization signal; NACHT-NTPase domain profile; Guanylate kinase domain profile
etc.
Remember that these programs only tell you are that there is a motif present and thus
there is the potential for these modifications and functions to occur. It is up to you to
determine experimentally which are real but at least you now know what to look for.
52
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
7) Secondary Structure Prediction
If protein structure, even secondary structure, can be accurately predicted from the
now abundantly available gene and protein sequences, such sequences become
immensely more valuable for the understanding of drug-design, the genetic basis of
disease, the role of protein structure in its enzymatic, structural, and signal
transduction functions, and basic physiology from molecular to cellular, to fully
systemic levels. In short, the solution of the protein structure prediction problem
(and the related protein folding problem) will bring on the second phase of the
molecular biology revolution (Munson et al., 1994).
JPRED - http://www.compbio.dundee.ac.uk/~www-jpred/submit.html
Jpred is an Internet web server that takes either a protein sequence or a multiple alignment of
protein sequences, and predicts secondary structure.
It works by combining a number of
modern, high quality prediction methods to form a consensus. Please be aware that secondary
structure prediction is an extremely complex problem that is under intensive research and we are
still at a relatively primitive stage. We cannot discuss the details of protein secondary structure
here but if you are interested in this area we recommend that you take a look at any major
biochemistry textbook. Essentially protein secondary structure consists of 3 major
conformations; the α Helix, the β pleated sheet and the coil conformation.
Example: Human alpha 1 hemoglobin. NP_000549.1
ƒ
At the ExPASy Æ “Secondary structure prediction”.
ƒ
Click on the link to JPRED.
ƒ
Click “Prediction”.
ƒ
Paste your sequence in the box provided.
ƒ
The defaults are OK.
ƒ
Click “Run secondary structure predictions!”
ƒ
Point 4 on the submission page allows you to deselect the BLAST search against
PDB (Protein Data Bank). If your sequence already has had its structure
predicted or experimentally determined it will be in here and you can follow the
link to PDB for information on the structure of your protein.
ƒ
If your protein is in PDB you can view your protein secondary structure using
RasMol (To download RasMol see the course website for a link).
53
M.Sc. in Molecular Medicine Bioinformatics Course
ƒ
Feb. 2005
Once you have RasMol running you can open your structure in it a view it using
a number of different options.
Otherwise continue with prediction
ƒ
The program may take a long time so you can save a bookmark and return to
your results later or choose to have your results e-mailed to you.
ƒ
There are a number of options to view the output, view your output in HTML
format (option 4).
ƒ
The complete output is too large to show here (see webpage).
ƒ
Scroll down through the output until you get to “Jpred” output. The line of
output beside this is the consensus secondary structure for your sequence. H=
Helices E= strands C= coils.
54
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
A Few Other Useful Tools at ExPASy
FindMod
Predicts potential protein post-translational modifications (PTM) and find potential
single amino acid substitutions in peptides. The experimentally measured peptide
masses are compared with the theoretical peptides calculated from a specified
SWISS-PROT/TrEMBL entry or from a user-entered sequence, and mass differences
are used to better characterise the protein of interest.
NetPhos:
The NetPhos WWW server produces neural network predictions for serine, threonine
and tyrosine phosphorylation sites in eukaryotic proteins.
Sulfinator:
Predicts tyrosine sulfation sites in protein sequences. Tyrosine sulfation is an
important post-translational modification of proteins that go through the secretory
pathway.
REP:
Searches a protein sequence for a collection of repeats such as leucine rich repeats
and many others.
Other Resources for Protein Sequence Analysis
1) Protein Prospector at UCSF - http://prospector.ucsf.edu/
MS-Digest: A protein digestion tool from the UCSF Mass Spectrometry Facility that
performs an in silico enzymatic digestion of a protein sequence, and calculates the
mass of each peptide.
MS-Product: A tool from the UCSF Mass Spectrometry Facility that calculates the
possible fragment ions resulting from fragmentation of a peptide in a mass
spectrometer. Fragmentation possibilities for post-source decay (PSD), high-energy
collision-induced dissociation (CID), and low-energy CID processes may be
calculated.
2) Pasteur Institute - http://bioweb.pasteur.fr/seqanal/protein/intro-uk.html
Antigenic: finds antigenic sites in proteins.
Helixturnhelix: reports nucleic acid binding motifs in your protein of interest.
DAY 2: Accessing Completed Genomes
55
M.Sc. in Molecular Medicine Bioinformatics Course
TOPICS:
1. UCSC Genome Bioinformatics.
2. Ensembl.
3. NCBI Genomic Biology.
Accessing Genomic Sequences:
56
Feb. 2005
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
There is no one resource available on the web that allows you to access all the
available genomes. In this course we will take a look at 3 excellent sites for
accessing most of the genomic information that is available out there – UCSC
Genome Bioinformatics; Ensembl & NCBI Genomic Biology. These sites often
contain similar information and it may be possible to get most of the information you
require from just one of these sites, however, to get the maximum amount of
information it is often worth having a look at all 3 of these sites. In this course we
will primarily concentrate on accessing the human genome, however, any of the
examples that we describe can easily be applied to any of the available species.
Remember that most of the genomes are still in a draft state and are subject to
change as more sequence becomes available.
http://genome.cse.ucsc.edu/
At this site the latest assembly of the human, chimp, dog, mouse, rat, opossum,
chicken, X.tropicalis, zebrafish, tetradon, fugu, C.elegans, C.briggsae, C.intestinalis,
A.mellifera, A.gambiae, a number of Drosophilae genomes, S.cerevisiae and the
SARS genomes can be accessed. You can choose which one you want to access by
using the pull down menus under “Genomes”.
Once you have decided what genome you want to access there are two major ways to
do so – 1) BLAT Search 2) Genome Browser.
BLAT Search:
Not to be confused with BLAST, a BLAT search is designed to quickly find
sequences of 95% and greater similarity of length 40 bases or more on the genome.
It may miss more divergent or shorter sequence alignments. It will find perfect
sequence matches of 33 bases, and sometimes find them down to 22 bases. You can
use this tool to locate any DNA/RNA sequence on the genome.
To do a BLAT search:
ƒ
Click on “Blat” in the left hand side menu.
ƒ
Paste your sequence in the box provided or upload a file containing your
sequence using the “Browse…” button.
57
M.Sc. in Molecular Medicine Bioinformatics Course
ƒ
Feb. 2005
Multiple sequences can be searched at once if separated by a line starting
with > and the sequence name. (Fasta format)
ƒ
Using the pull-down menus choose the genome and assembly you wish to
search (default is most recent assembly).
ƒ
You can leave the defaults in the other menus as they are (unless you want to
search a protein => change “Query type:” to protein) and “Submit”.
ƒ
This will take you to BLAT Search Results – There may be more than one hit
against the genome, but the best hit will be identified by its percentage
identity and the highest score.
Example - Homo sapiens corticotropin releasing hormone receptor 1 (CRHR1)
mRNA. [NM_004382.2]
ƒ
You can click on either “browser” (see next section) or “details”
ƒ
Details – alignment of the mRNA to the genomic sequence. Gives you the
intron-exon structure of your gene.
Browser:
The genome can also be accessed via the browser, which is a graphical display of the
genome where various features can be displayed at once.
To access the genome via the browser:
ƒ
Click on “Genome Browser” in the left hand side menu of the start page or
via BLAT as described above.
ƒ
This will bring you to the Genome Browser Gateway.
ƒ
Here again you can choose which genome and assembly you wish to access.
58
M.Sc. in Molecular Medicine Bioinformatics Course
ƒ
Feb. 2005
In the “position” box you can enter a number of terms to access a particular
region of the genome.
ƒ
You can also enter the accession number of a sequenced human genomic
clone, an mRNA or EST accession, the name of a fingerprint map contig, an
STS marker, a cytological band, a range of a chromosome, or words from the
Genbank description of an mRNA such as the gene name.
Example - Homo sapiens corticotropin releasing hormone receptor 1 (CRHR1)
mRNA. [NM_004382]
ƒ
One way to search for this gene is to type CRHR1 in the position box and
click “Submit”.
ƒ
Known Genes: Are just that, known genes that match search criteria. You
can see that there are a number of variants.
59
M.Sc. in Molecular Medicine Bioinformatics Course
ƒ
Feb. 2005
RefSeq Genes: CRHR1 is a known RefSeq gene (RefSeq is an NCBI
database of annotated genes with 1 reference sequence given for any 1 gene)
and is located on chromosome 17 at the position shown above.
ƒ
Human Aligned mRNA Search Results: Displays the known human
mRNAs for CRHR1.
ƒ
Non-Human Aligned mRNA Search Results: Displays the known nonhuman mRNAs for CRHR1.
ƒ
AceView Gene Models with Alt-Splicing: AceView attempts to find the
best alignment of each mRNA/EST against the genome, and clusters the
alignments into the least possible number of alternatively spliced transcripts.
ƒ
Click on one of the links to take you to a graphical display of the CRHR1 on
the genome (see below).
ƒ
You can use the zoom buttons to zoom in or out of the current location on the
genome enabling you to view a wider or more specific genomic context
around your gene. You can also use the move buttons to move along the
genome.
60
M.Sc. in Molecular Medicine Bioinformatics Course
ƒ
Feb. 2005
There are a number of features displayed – just to point out a few.
o Base position – the coordinates of the gene on the chromosome.
o Chromosome band i.e. 17q21.31
o RefSeq Genes: Known genes in this area – click on one of the links to
the left to get more details.
o AceView gene models with alternative splicing.
o Human mRNAs from Genbank.
o SNPs.
ƒ
Click on any of the links for more details.
ƒ
Below the graphical display there are a number of other items that you can
also choose to display on the browser.
ƒ
You can choose to hide these options or display them in various formats. The
full option displays each item on its own line on the browser.
ƒ
You can find out about any of the options by clicking on the blue hyperlinks.
ƒ
Once you have chosen which options you wish to display click the “refresh”
button”.
61
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
Obtaining Genomic Sequence from UCSC Genome Browser:
ƒ
Click anywhere on the Known Gene track. This takes you to a page with
information about your gene including links to RefSeq, OMIM, LocusLink,
PubMed, GeneLynx, GeneCards, Mouse Ortholog etc.
ƒ
You can follow any of these links to more information on your gene.
ƒ
To obtain sequence information click on the “Sequence” Link:
ƒ
Click on the link to “Genomic”.
ƒ
Tick any of options that you require and then click submit to obtain your
sequence.
For more detailed information on the UCSC Genome Browser click on the
link to the User Guide at the start page.
62
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
http://www.ensembl.org
Ensembl is a joint project between EMBL - EBI and the Sanger Institute to
develop a software system which produces and maintains automatic annotation
on eukaryotic genomes. A wide range of genomes are available.
ƒ
Click on one of the species to access the genomic information e.g. Mouse.
ƒ
To find your gene of interest you can enter in the empty box, the gene
symbol, gene accession number, mRNA accession number, SwissProt
accession number, EST accession number etc.
ƒ
You can also access the genome by chromosome number.
ƒ
Below this there are 3 buttons
63
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
o Export – If you have an Ensembl I.D. for your gene you can
download its sequence from here.
o BLAST/SSAHA- BLAST your sequence against the genome.
SSAHA is similar to BLAT.
o EnsMart – Allows the download a large datasets e.g. all the genes on
a chromosome, the entire genome etc.
Example: Mouse beta-defensin 4 (defB4).
ƒ
Use the pull-down “Anything” menu to select “Gene” and type the gene
RefSeq symbol in the empty box (defB4).
ƒ
Click “Lookup”.
ƒ
This will take you to a query results page. In this case there is only one hit
but sometimes you will have to look through a number of entries to find what
you are looking for.
ƒ
Click on the EnsEMBL Gene: ENSMUSG00000059230 link for
information on your gene such as its sequence, structure, domains that it
contains etc.
ƒ
Click on the link to “Genomic Location” to display the gene in the genome
similar to the UCSC browser.
ƒ
There are 4 major views displayed
o Chromosome – highlights position on the chromosome
o Overview – shows genes surrounding gene of interest on chromosome
band
o Detailed View – more detailed view of your gene.
o Basepair View – displays sequence, translated sequence and
restriction sites.
64
M.Sc. in Molecular Medicine Bioinformatics Course
ƒ
Feb. 2005
On the detailed view you can use the pull down menus to “Features”, “DAS
Sources” & decorations to choose what details you wish to display (similar to
UCSC)
ƒ
The “Export” menu can be used to download the genomic sequence or
features of this area in the genome (E.g. list of genes).
ƒ
As with the UCSC browser you can zoom in or out of this region of genome
or move along the genome using the “window” buttons.
65
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
http://www.ncbi.nlm.nih.gov/Genomes/index.html
Many of you will be familiar with the National Center for Biotechnology
Information (NCBI) website which has many very useful resources including;
Entrez, PubMed, Genbank, BLAST, OMIM. Today we will see how to use the NCBI
site to interrogate the genomic sequences that are available there.
The NCBI site provides a good starting point for accessing the widest range of
eukaryotic and microbial genomes. Many of these genomes will have their own
dedicated sites located at other websites, but the NCBI site will provide links to
them.
Accessing the Human Genome:
To access the human genome - go to the URL above and click on the link to
“Human”. This page provides a number of links such as a link to BLAST where you
can search your sequence against the human genome. You can also browse the
genome by chromosome by clicking on one of the chromosomes. The best way to
access the genome if you have a particular gene of interest is to search for your gene
in Entrez Gene. Entrez Gene provides a single query interface to curated sequence
and descriptive information about genetic loci. It presents information on official
nomenclature, aliases, sequence accessions, phenotypes, EC numbers, MIM
numbers, UniGene clusters, homology, map locations, and related web sites.
ƒ
Follow the “Gene Database” link on the Human Genome Resources page.
ƒ
At the top of the page search Entrez Gene by entering your gene name (full
name, abbreviation or accession number) in the box and “Go”. (Example:
BRCA2)
ƒ
This brings up a results page that matches the query for some reason.
ƒ
You can use the limits section to limit your search by various criteria such as
organism.
66
M.Sc. in Molecular Medicine Bioinformatics Course
ƒ
Feb. 2005
Click on BRCA2 (i.e. GeneID 675) to take you to the Entrez Gene page for
that gene.
Starting at the top of the page:
ƒ
A graphic of the BRCA2 transcript is shown including the intron-exon
structure. You can click on this graphic to obtain the sequence.
ƒ
This is followed by a graphic showing BRCA2 in its genomic context i.e.
what genes are located around it.
ƒ
This is followed by various information on the gene including;
Gene aliases – other names for the gene.
Summary - written by staff of the NCBI RefSeq group describing the function,
localization, and sequence properties of the gene and its products.
Bibliography – a detailed list of PubMed entries for the gene.
Interactions – What other genes/proteins are known to interact with BRCA2.
A General Gene Information Section – includes the official gene symbol and
name, gene ontology details, homology with mouse and rat, etc.
There is also a link to the NCBI Map Viewer (see below).
NCBI Reference Sequences (RefSeq) - All RefSeq records created for a given
locus are listed. Multiple records are distinguished by the brief description of the
transcript variant. This section provides links to:
RefSeq nucleotide record (genomic and mRNA accessions have 'NG_' and 'NM_'
prefixes, respectively).
RefSeq Product - protein record (the 'NP_' prefix).
Conserved domains found in the protein.
Related Sequences - A table of a subset of representative nucleotide and protein
accessions for the locus. EST accession numbers are provided if no other sequence
data are available to represent the locus.
67
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
Additional Links - This section names and provides links to additional sites that
may contain information related to this locus such as OMIM, UniGene, etc.
MAP VIEWER:
This is the NCBI graphical display tool, which you can use to display the genomic
context of your sequence. This tool is not as user friendly or as advanced as the
UCSC or Ensembl browsers and we recommend that you use these to view the
genome graphically where possible. Not all species are available at these sites so you
may need to use Map viewer.
ƒ
Click on “Maps & Options” to choose which features you wish to display.
ƒ
Click on any of the genes, RNAs or Unigenes to get more information.
ƒ
You can download genomic sequence for the region selected using the
“Download/View Sequence/Evidence” link.
68
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
Accessing the Other Genomes: http://www.ncbi.nlm.nih.gov/Genomes/index.html
Plant Genomes Central
The plant genomic effort has one technical hurdle relative to other genomic efforts.
The range of plant genome size is very large; extending from approximately the
same size as small animals to more than five times as large as human. At NCBI
resources for many plant species are available including;
ƒ
Arabidopsis thaliana (thale cress)
ƒ
Gossypium (cotton)
ƒ
Hordeum vulgare (barley)
ƒ
Lycopersicon esculentum (tomato)
ƒ
Medicago truncatula (barrel medic)
ƒ
Oryza sativa (rice)
ƒ
Solanum tuberosum (potato)
ƒ
Triticum aestivum (bread wheat)
ƒ
Zea mays (corn)
Malaria Parasite
This resource provides data and information relevant to malaria genetics and
genomics following the sequencing of the malaria parasite Plasmodium falciparum
and one of its major vectors Anopheles gambiae genomes. These resources include;
ƒ
Organism specific sequence BLAST databases.
ƒ
Genome maps & linkage markers.
ƒ
Information about genetic studies.
ƒ
Links to other malaria web sites.
ƒ
Genetic data on related apicomplexan parasites.
69
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
Microbial Genomes
This resource provides links to the 222 (as of 15/02/2005) completely sequenced
bacterial genomes (21 Archaea & 201 eubacteria). You can download information
on the genome in a number of different formats
[T] - All proteins of the complete genome were searched against "nr" database. The
detected homologs were classified into three taxonomic groups, Eukaryota,
Eubacteria and Archaea in TaxTable.
[P] – Download the protein sequences from ProtTable.
[C] – Functional classifications are located in COG Table.
[D] - 3-D neighbors (proteins with sequence similarity to proteins with known 3D
structure)
[L] – BLAST a sequence against the genome.
[S] - CDD search (list of conserved domains in proteins).
[F] - Genomic sequence in FASTA format.
For most of the genomes you can follow links to an organism-specific website
with even further details (usually hosted by the sequencing consortium)
70
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
Retroviruses
Collection of resources at NCBI specifically designed to support the research of
retroviruses. The resources include:
ƒ
Taxa-specific pages for HIV-1, HIV-2, SIV, HTLV, STLV.
ƒ
Genotyping tool - uses the BLAST algorithm to identify the genotype of a
query sequence
ƒ
Alignment tool - global alignment of multiple sequences
ƒ
HIV-1 automatic sequence annotation - generates a report in GenBank format
for one or more query sequences
ƒ
Genome maps - graphical representation of 50 retrovirus complete genomes
If you still can’t find what you are looking for at any of these sites try:
The Institute for Genomic Research (TIGR) - http://www.tigr.org/
The Sanger Institute - http://www.sanger.ac.uk/
71
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
Some Other NCBI Resources:
Unigene - http://www.ncbi.nlm.nih.gov/UniGene/
UniGene is an experimental system for automatically partitioning GenBank
sequences into a non-redundant set of gene-oriented clusters. Each UniGene cluster
contains sequences that represent a unique gene, as well as related information such
as the tissue types in which the gene has been expressed and map location. The
dataset is pretty comprehensive – for human there are:
52,888
sets total
In addition to sequences of well-characterized genes, hundreds of thousands novel
expressed sequence tag (EST) sequences have been included. Consequently, the
collection may be of use to the community as a resource for gene discovery.
UniGene has also been used by experimentalists to select reagents for gene mapping
projects and large-scale expression analysis.
It should also be noted that no attempt has been made to produce contigs or
consensus sequences. There are several reasons why the sequences of a set may not
actually form a single contig. For example, all of the splicing variants for a gene are
put into the same set. Moreover, EST-containing sets often contain 5' and 3' reads
from the same cDNA clone, but these sequences do not always overlap.
The NCBI genetic disease site - http://www.ncbi.nlm.nih.gov/disease/
This is rather a useful site, which classifies syndromes, diseases and conditions by
sort: immune system, muscle and bone, signals, transporters, nervous system etc.
You can browse through the hierarchy to find interesting diseases in your field of
interest.
72
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
OMIM - http://www.ncbi.nlm.nih.gov/Omim/
The On-line Mendelian Inheritance in Man is a remarkable resource for all aspects
of medical and clinical genetics. NCBI has a server that allows you to search this
database.
Questions and Exercises
1) What contribution has Kirk Douglas made to medical/genetic research ?
2) What is the map-position of the gene involved in PKU ?
3) What happens when you search for Huntingdon ?
4) Better try Huntington ?
5) Any other genes where a key molecular biological flag is poly CAG repeats ?
6) For a female role model in science look up Julia Bell.
7) In what proportion of OMIM entries is "mental retardation" involved ?
73
M.Sc. in Molecular Medicine Bioinformatics Course
Day 2 – Two Sequence Alignment
TOPICS
1. Dotplots
2. Two Sequence Alignment – global or local?
74
Feb. 2005
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
A really important aspect of bioinformatics is the concept of sequence alignment.
This is really important for homology searching – iteratively comparing a sequence
to each sequence in a database – but two sequence comparisons can also yield useful
information – you can find SNPs in this way or get clues about essential
residues/bases in two similar sequences.
Dotplots:
Paradoxically one of the most useful two sequence analyses you can do is to
compare a sequence to itself. One way to do this is looking for stem-loop and
inverted repeat structures with Mfold. A dot plot is the first thing to think of when
you want to look for repeats or other structural motifs in one sequence. Go to
http://bioweb.pasteur.fr/seqanal/interfaces/dottup.html
And paste in your sequence as both sequence a and sequence b. Two Xenopus
sequences from swissprot demonstrate the usefulness of this analysis. Dotplots work
by comparing a moving window of residues/bases across the whole length of the
sequence. Repeated units show clearly if you set the sensitivity of the dotplot
properly. If the repeated unit is short then a long window will not find the repeat
because it will be swamped by the random noise to either side of the repeat. On the
other hand a very short window will find hits all over the place. You should choose
several different window/word sizes to see which gives you the most convincing
picture.
About 8% of swissprot sequences have annotated repeats.
The dottup program, while it looks effectively for repeated windows, does not allow
a lapse in sensitivity – where e.g. 12/15 matches would be acceptable. GCG has a
pair of programs compare and dotplot that are recommended for this. Or use
LALIGN (below) to find the sub-optimal repeats.
Dotplots on two different sequences can show where common domains are, even if
their order has changed.
Two sequence alignment: global or local?
Having found repeated motifs in your sequence with this graphical method, you will
want to align the sequence itself. Sequences with known repeats are quite difficult
to align: global alignment program gets confused about which motifs to align with
75
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
which; local alignment programs, such as blast or Smith-Waterman, tend to align the
best pair of repeats only. So the program of choice is
Lalign:
http://www.ch.embnet.org/software/LALIGN_form.html
from Bill Pearson’s Fasta package. Otherwise you have to ask whether you want to
align (as much as possible of) the whole sequence or the best motif. With closely
related sequences you will get essentially the same picture with either local or global
methods. With more distant relatives you have to ask yourself what alignment
answers for you best. PISE has two options.
For local alignment WATER (Smith-Waterman algorithm):
http://bioweb.pasteur.fr/seqanal/interfaces/water.html
For global alignment NEEDLE (Needleman-Wunsch algorithm):
http://bioweb.pasteur.fr/seqanal/interfaces/needle.html
on the course home page there are alternatives for doing both sorts of alignment.
Suggestion
You might also like to use secondary structure (α-helix, β-sheet, turn, loop)
prediction software in conjunction with sequence alignment.
Further sequence comparison tools at PISE:
needle, stretcher: Needleman-Wunsch global alignment.
water, matcher: Smith-Waterman local alignment.
merger, megamerger: Merge two overlapping sequences.
stssearch: Searches DNA sequences for matches with a set of STS primers.
supermatcher: Finds a match of a large sequence against one or more sequences.
dotmatcher: Creates a dot plot of two sequences.
dottup: Displays a wordmatch dotplot of two sequences
est2genome: Align EST and genomic DNA sequences.
diffseq: Find differences (SNPs) between nearly identical sequences.
76
M.Sc. in Molecular Medicine Bioinformatics Course
Day 2– Homology Searching
TOPICS
1. Introduction to homology searching.
2. BLAST.
3. FASTA.
4. Smith-Waterman.
77
Feb. 2005
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
Introduction
Perhaps the most widely used bioinformatics protocol is to search a database for
sequences similar to a candidate sequence. Because of an implicit underlying
hypothesis that if sequences are similar at some statistically significant level they
share a common ancestor, this methodology is generally called homology searching.
It is a useful tool because, if two sequences are similar, then they are likely to have a
similar structure and if they have a similar structure they are likely to have a similar
function.
You can thus get important clues about the function of an as yet
uncharacterized sequence.
There are several different algorithms for implementing a homology search, and
each program will have a wide range of options and parameters to help you carry out
a more informative type of search. The de facto standard for homology searching is
the blast family of programs and this chapter will concentrate on them. You should
note, however, that for searches with DNA sequences against DNA databases, the
program Fasta is often more sensitive, if in general it will be a little slower. SmithWaterman searches are generally more informative than either Blast or Fasta but
very much slower.
Blast: http://www.ncbi.nlm.nih.gov/BLAST/
Blast is a finely tunable algorithm to search very large databases for homologues in a
managable/finite time. It may be helpful to think that the complete human genome
DNA comprises more than 3.2 * 109 bases. On a letter for letter basis this is the
equivalent of about 8 complete Encyclopedia Britannicas. So the task of finding a
sentence similar to the one you are now reading in such a forest of information is,
shall we say, daunting. It is a 5 step process:
1. break the query sequence into a number of 'words' (typically 4 protein residues).
2. search the database for matches to these words.
3. the program builds on the "hits" by extending the alignment out on either side of
the core word - these extended hits are called HSPs - high scoring segment pairs.
4. all the statistically significant segment pairs are sorted by some scoring criterion,
so that the 'best' matches are presented first.
78
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
5. the significant matches are formally aligned to show where the homologous
regions are.
Blast is not one program but a family of programs for carrying out different classes
of search:
blastn: searches a DNA sequence against a DNA database such as EMBL, Genbank,
or dbEST.
blastp: searches a protein sequence against a protein database such as Swissprot, or
trembl (conceptual translations of the EMBL DNA database) or genpept (ditto for
Genbank) or, most commonly, "nr" a non-redundant database which ideally contains
one copy of every available sequence.
Then you have:
blastx: searches a DNA sequence (translated in all six reading frames) against a
protein database.
tblastn: searches a protein sequence against a DNA database (translated in all six
reading frames) – essential for searching EST databases.
and in the interests of completeness there is:
tblastx: searches a DNA sequence (translated in all six reading frames) against a
DNA database (translated in all six reading frames).
See the Blast page at NCBI for details of other flavours of Blast programs.
Fasta:
The other widely used, although possibly not widely enough used, algorithm for
doing homology searches against databases is Fasta, maintained by Bill Pearson in
Virginia. You can carry out Fasta searches from: http://www.ebi.ac.uk/Tools/ this
introductory course will not cover Fasta except to note that it is a) a little slower than
blast b) it is the algorithm of choice if you have to search a DNA sequence against a
DNA database.
79
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
Smith-Waterman:
These searches are very much more sensitive than either blast or fasta, but
consequently take a much longer time to complete. Perhaps 20x slower than blast.
One
implementation
of
S-W
is
Blitz,
which
can
be
found
on
http://www.ebi.ac.uk/Tools/ the EBI homology server. In order to get S-W searches
down to sensible times it is often carried out on Massively Parallel Computers.
Because for many biological searches blast will give you results that are a) good
enough and b) returned in the shortest time, we will investigate that algorithm in
more detail.
Options in blast.
Masking/filtering of less informative sequence motifs.
If your query sequence is protein you can "mask" regions of the protein that may
give you confusing or biologically uninformative information. This masking can be
of two types, using two different algorithms. xnu masks repeated sequences while
seg masks regions of low-complexity - regions where there are "too many" serines
for example. Masking for low-complexity stops you hitting sequences that are
similar to your the query sequence only because they both have similar
compositional bias: proline-rich proteins for example. An example follows:
>P04729 Wheat gamma gliadin
MKTFLVFALIAVVATSAIAQMETSCISGLERPWQQQPLPPQQSFSQQPPFSQQQQQPLPQ
QPSFSQQQPPFSQQQPILSQQPPFSQQQQPVLPQQSPFSQQQQLVLPPQQQQQQLVQQQI
PIVQPSVLQQLNPCKVFLQQQCSPVAMPQRLARSQMWQQSSCHVMQQQCCQQLQQIPEQS
RYEAIRAIIYSIILQEQQQGFVQPQQQQPQQSGQGVSQSQQQSQQQLGQCSFQQPQQQLG
QQPQQQQQQQVLQGTFLQPHQIAHLEAVTSIALRTLPTMCSVNVPLYSATTSVPFGVGTG
VGAY*
and after low complexity masking:
>P04729 SEG low-complexity masked
MKTFLVFALIAVVATSAIAQMETSCISGLERPWXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXLNPCKVFLQQQCSPVAMPQRLARSQMWXXXXXXXXXXXXXXXXXXXXXXX
RYEAIRAIIYSIIXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXHQIAHLEAVTSIALRTLPTMCSVNVPLYSATTSVPFGVGTG
VGAY*
Similar filtering (another word for masking) can be carried out on DNA sequences
with a program called DUST. This will effectively erase such minimally informative
but very widely distributed sequences as polyA tails.
80
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
Scoring matrices.
Homology searching algorithms all look for the best matches between the query
sequence and database sequences. "best" is defined by a high score using one of
several alternative scoring matrices. One such matrix - blosum62 - is shown below.
This matrix is based on observed substitutions in a database of aligned sequences
where 62% of the residues are identical. The distribution of the remaining 38% is
analysed to yield:
# BLOSUM 62
A
R
N
D
C
Q
E
G
H
I
L
K
M
F
P
S
T
W
Y
V
A
4
-1
-2
-2
0
-1
-1
0
-2
-1
-1
-1
-1
-2
-1
1
0
-3
-2
0
R
-1
5
0
-2
-3
1
0
-2
0
-3
-2
2
-1
-3
-2
-1
-1
-3
-2
-3
N
-2
0
6
1
-3
0
0
0
1
-3
-3
0
-2
-3
-2
1
0
-4
-2
-3
D
-2
-2
1
6
-3
0
2
-1
-1
-3
-4
-1
-3
-3
-1
0
-1
-4
-3
-3
C
0
-3
-3
-3
9
-3
-4
-3
-3
-1
-1
-3
-1
-2
-3
-1
-1
-2
-2
-1
Q
-1
1
0
0
-3
5
2
-2
0
-3
-2
1
0
-3
-1
0
-1
-2
-1
-2
E
-1
0
0
2
-4
2
5
-2
0
-3
-3
1
-2
-3
-1
0
-1
-3
-2
-2
G
0
-2
0
-1
-3
-2
-2
6
-2
-4
-4
-2
-3
-3
-2
0
-2
-2
-3
-3
H
-2
0
1
-1
-3
0
0
-2
8
-3
-3
-1
-2
-1
-2
-1
-2
-2
2
-3
I
-1
-3
-3
-3
-1
-3
-3
-4
-3
4
2
-3
1
0
-3
-2
-1
-3
-1
3
L
-1
-2
-3
-4
-1
-2
-3
-4
-3
2
4
-2
2
0
-3
-2
-1
-2
-1
1
K
-1
2
0
-1
-3
1
1
-2
-1
-3
-2
5
-1
-3
-1
0
-1
-3
-2
-2
M
-1
-1
-2
-3
-1
0
-2
-3
-2
1
2
-1
5
0
-2
-1
-1
-1
-1
1
F
-2
-3
-3
-3
-2
-3
-3
-3
-1
0
0
-3
0
6
-4
-2
-2
1
3
-1
P
-1
-2
-2
-1
-3
-1
-1
-2
-2
-3
-3
-1
-2
-4
7
-1
-1
-4
-3
-2
S
1
-1
1
0
-1
0
0
0
-1
-2
-2
0
-1
-2
-1
4
1
-3
-2
-2
T
0
-1
0
-1
-1
-1
-1
-2
-2
-1
-1
-1
-1
-2
-1
1
5
-2
-2
0
W
-3
-3
-4
-4
-2
-2
-3
-2
-2
-3
-2
-3
-1
1
-4
-3
-2
11
2
-3
Y
-2
-2
-2
-3
-2
-1
-2
-3
2
-1
-1
-2
-1
3
-3
-2
-2
2
7
-1
V
0
-3
-3
-3
-1
-2
-2
-3
-3
3
1
-2
1
-1
-2
-2
0
-3
-1
4
Exercise:
Use the matrix to verify that the following sequence match clipped from a blast
homology search has the right score (the convention is that exact matches are echoed
on the middle line, "mismatches" have nothing, while "conservative substitutions",
such as the replacement of leucine by isoleucine below, are given a +):
Score = 28:
Query:
3 LKQSNTLL 10
L QSNT+L
Sbjct:
62 LYQSNTIL 69
Choosing a different scoring matrix will give you a different cohort of hits.
81
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
#BLOSUM 30
A R N
A 4 -1 0
R -1 8 -2
N 0 -2 8
D 0 -1 1
C -3 -2 -1
Q 1 3 -1
E 0 -1 -1
G 0 -2 0
H -2 -1 -1
I 0 -3 0
L -1 -2 -2
D
0
-1
1
9
-3
-1
1
-1
-2
-4
-1
C
-3
-2
-1
-3
17
-2
1
-4
-5
-2
0
Q
1
3
-1
-1
-2
8
2
-2
0
-2
-2
E
0
-1
-1
1
1
2
6
-2
0
-3
-1
G
0
-2
0
-1
-4
-2
-2
8
-3
-1
-2
H
-2
-1
-1
-2
-5
0
0
-3
14
-2
-1
I
0
-3
0
-4
-2
-2
-3
-1
-2
6
2
L
-1
-2
-2
-1
0
-2
-1
-2
-1
2
4
#BLOSUM
A R
A 5 -2
R -2 6
N -2 -1
D -3 -3
C -1 -5
Q -1 1
E -1 -1
G 0 -3
H -2 0
I -2 -4
L -2 -3
D
-3
-3
1
7
-5
-1
1
-2
-2
-5
-5
C
-1
-5
-4
-5
9
-4
-6
-4
-5
-2
-2
Q
-1
1
0
-1
-4
7
2
-3
1
-4
-3
E
-1
-1
-1
1
-6
2
6
-3
-1
-4
-4
G
0
-3
-1
-2
-4
-3
-3
6
-3
-5
-5
H
-2
0
0
-2
-5
1
-1
-3
8
-4
-4
I
-2
-4
-4
-5
-2
-4
-4
-5
-4
5
1
L
-2
-3
-4
-5
-2
-3
-4
-5
-4
1
5
90
N
-2
-1
7
1
-4
0
-1
-1
0
-4
-4
Compare the scores of following two alignments using blosum30 and blosum90
Alignment Score
Query: GHDEICI
39
GH + C
Sbjct: GHACNCG
5
Matrix
Blos30
Blos90
Score
Alignment
19 Query: HEQCRLEN
+E
LEN
24 Sbjct: QENAHLEN
In the examples above, Blosum 30 will give a higher score to and thus preferentially
find the GHDEICI match while Blosum 90 will find HEQCRLEN. In real database
searches changing the substitution matrix may change the order in which sequences
are scored and reported, in other cases it will identify totally different sequences as
having a relationship with the query sequence.
Expectation cutoff
The blast defaults are designed to suit most of the people most of the time. In order
to minimise the collection of marginal, statistically non-significant information, blast
sets an 'expectation cutoff' parameter to 10. Accepting this means that blast will not
report any match so common that you would expect to find 10 copies in the database
by chance alone. A search for a short protein motif, ELVIS for example, in
Swissprot with its 77,000 entries and 2 million residues will, by chance alone, find
82
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
several to many copies. If you are using blastp for such a short motif search then you
should crank up the expectation cutoff to the maximum of 1000. On the other hand,
if you are only interested in very precise homologues and do not wish to be
overwhelmed with a flood of marginal alignments, you might consider setting the E
value to 0.001
Limit search taxonomically
Most Blast servers now will allow you to choose a subset of the sequence universe to
search against. You should be able to search only human sequences or only
mammalian sequences for example.
Output delivery options.
While blast is a general workhorse for finding similar sequences, each researcher
will be asking a more or less specific question of their search. If you want to see if
your sequence is homologous with anything, then a single hit would be enough. If
you wanted to find all members of a protein family, perhaps to align them to find
conserved residues, then more then 200 hits might not be enough. The quantity of
information returned by a typical blast search can be substantial and will consume
large amounts of disk to store it and many trees to print it. Accordingly, you are
given the option to limit a) the number of hits and b) the number of alignments
reported. Good servers will give you the option of returning the output in HTML
with clickable links to the relevant database entries.
WWW access to Blast.
You can access blast in many different ways at many different sites. These are NOT
all equivalent! The default parameters may be significantly different, the databases
may not be updated on the same schedule and so may be significantly different in
size or level of redundancy. Three accessible, authoritative, alternatives are on the
WWW.
The Blast server at the NCBI in Bethesda, MD, USA:
http://www.ncbi.nlm.nih.gov/BLAST
The Blast server at the EBI in Hinxton, UK:
http://www.ebi.ac.uk/searches/searches.html
83
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
The SIB blast site is easily customizable
http://www.ch.embnet.org/
Blast guidelines.
When to use what algorithm
a. As a rule of thumb, if your DNA sequence is coding (i.e. not an intron, a structural
RNA, "junk" DNA or some upstream control region), you should translate it first and
use blastp search a protein database. It will be quicker, more sensitive and find more
distant relatives.
b. If your DNA sequence is not coding, use Fasta instead. You should, therefore,
rarely have to use blastn.
c. If you want to do a preliminary check for frameshift errors in your sequence, use
blastx to compare your sequence, translated in all six reading frames, against a
protein database. Why might this help you identify frameshift errors?
d. If you want to search for a particular protein sequence in a database of expressed
sequence tags (ESTs) you will have to use tblastn.
A widely applicable blast protocol
If you want to carry out a reasonably comprehensive search of a protein database to
find potential homologues to a query sequence you will have to carry out several
blastp searches. You will however, adjust your approach depending on the exact type
of information that will satisfy your quest. On any well designed blast server it
should be easy to determine what are the available options, but you should scrutinise
the page carefully to determine what are the default options and parameters. By all
means take the defaults, but, on its own this is unlikely to result in an adequate, let
alone comprehensive, search. The DNA databases are doubling in size every 12-14
months; so a fresh blast search just before submitting your paper has much to
recommend it.
On any reputable WWW homology server:
a. Paste in your sequence and do a search taking the default parameters.
84
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
b. Do the search again, with or without low-complexity masking, depending on what
option the server has chosen as the default in part a. If low complexity regions are
found the XXXed sequence should appear at the top of your results.
c. Do the search again using two different substitution scoring matrices. One based
on sequences that are evolutionarily "close" such as Blosum90 or PAM30 and
another based on sequences that are evolutionarily "distant" such as Blosum40 or
PAM250. The latter search is more likely to pick up a rather distant, diffuse weak
homologue.
d. If appropriate (sometimes your sequence will have no low-complexity regions) do
b x c to carry out, in all, six blast searches.
e. If your results indicate that the first 100s of best hits are members of a well
characterised protein family (a fact that you may already know), and that these hits
are all pointing to a particular domain of your query protein, you may have to edit
(by hand!) your sequence (XXXXing out the already identified region) to find more
distant and potentially interesting homologues which have been swamped out by a
deluge of higher scoring hits.
f. Scrutinise the results of all your searches taking into account not only the scores
but also the alignments. Pay particular attention to hits which are unexpected or
counter-intuitive.
g. You can eliminate a large number of useless but positive hits by only searching,
say, human sequences.
Interpreting output from blastp.
Output from a blast search is voluminous and in four or five parts.
1. The first part is administrative, and should include copyright information, the date,
references and most importantly a note of what database has been searched and what
size it was. With the DNA database doubling in size every year, you will not be able
to 'replicate your blast experiment' after an interval of as little as two weeks. You
should note down these details for your materials and methods section.
85
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
2. On some sites (NCBI) a very useful graphic showing the length and degree of
homology of all the hits follows. You can ‘mouse-over’ this to see which sequences
are homologous to (part of) your query.
2. There follows a list of "hits" with a) a database accession number or other
identifier b) a brief description c) a score and d) some information on the probability
of finding such a hit in the searched database. There will be a certain amount of
variation among servers in how this information is presented.
3. After this there are a number of alignments of the query sequence with the
significant hits.
4. Finally there is more administrative and statistical information including any
warnings or error messages.
The hit list should look like:
Blast server EBI:
Sequences producing significant alignments:
SW:GDB1_WHEAT
SW:GLTC_WHEAT
SW:GLTB_WHEAT
SW:GLTA_WHEAT
SW:GDB3_WHEAT
SW:HOR1_HORVU
SW:HOR3_HORVU
P04729
P16315
P10386
P10385
P04730
P06470
P06471
Score
(bits)
E
Value
GAMMA-GLIADIN B-I PRECURSOR.
GLUTENIN, LOW MOLECULAR WEIGHT SUBUNIT ...
GLUTENIN, LOW MOLECULAR WEIGHT SUBUNIT ...
GLUTENIN, LOW MOLECULAR WEIGHT SUBUNIT ...
GAMMA-GLIADIN (GLIADIN B-III) (FRAGMENT).
B1-HORDEIN PRECURSOR.
B3-HORDEIN (FRAGMENT).
616
510
480
343
329
323
310
e-176
e-144
e-135
3e-94
5e-90
3e-88
3e-84
SW:INVO_RAT P48998 INVOLUCRIN.
SW:SRY_MOUSE Q05738 SEX-DETERMINING REGION Y PROTEIN (TESTIS...
SW:FTSK_ECOLI P46889 CELL DIVISION PROTEIN FTSK.
SW:OVO_DROME P51521 OVO PROTEIN (SHAVEN BABY PROTEIN).
SW:FCA_ARATH O04425 FLOWERING TIME CONTROL PROTEIN FCA.
SW:CLOC_MOUSE O08785 CIRCADIAN LOCOMOTER OUTPUT CYCLES KAPUT...
SW:E75B_DROME P17672 ECDYSONE-INDUCIBLE PROTEIN E75-B.
61
61
59
58
57
56
52
4e-09
4e-09
2e-08
2e-08
7e-08
1e-07
1e-06
Then after a large number of ‘sensible’ hits, such reports as:
The 1e-06 on the last line of the output tells you that the probability of finding a
match as good as this by chance in the current database is 1 * e-06. For biologists
who are used to accepting probabilities of 0.05 or 0.001 as meaningful, this is highly
significant statistically, but may nevertheless mean little or nothing biologically.
The first three hits are the same when you use the blast server at the NCBI but,
because the implementation is different the probabilities are different. You’ll have to
be careful to record where, when and using what parameters you do your blast
searches if you want them to be reproducible.
Blast server NCBI:
86
M.Sc. in Molecular Medicine Bioinformatics Course
Sequences producing significant alignments:
gi|121100|sp|P04729|GDB1_WHEAT
gi|121459|sp|P16315|GLTC_WHEAT
gi|121102|sp|P04730|GDB3_WHEAT
gi|123458|sp|P06470|HOR1_HORVU
Feb. 2005
Score
(bits)
GAMMA-GLIADIN B-I PRECURSOR ...
GLUTENIN, LOW MOLECULAR WEIG...
GAMMA-GLIADIN (GLIADIN B-III...
B1-HORDEIN PRECURSOR >gi|100...
197
176
114
103
E
Value
2e-50
3e-44
2e-25
4e-22
To make an estimate of the biological significance, you will have to look further
down the output until you come to a listing of the alignments and scores of which the
"hit-list" is a summary:
>SW:DC11_DROME P18169 drosophila melanogaster (fruit fly). defective chorion-1 fc125
protein precursor. 2/91
Length = 1123
Score = 215 (80.7 bits), Expect = 7.7e-16, P = 7.7e-16
Identities = 73/233 (31%), Positives = 119/233 (51%)
Query: 34 QQQPLPPQQ-SFSQQPPFSQQQQQPLPQQPSFSQQQPPFSQQQPILSQQPPFSQQQQPVL 92
QQ P+ QQ +S++
QQ QQ + Q P
QQ+ +S++Q + QQ
QQ P++
Sbjct:570 QQNPMMMQQRQWSEEQAKIQQNQQQIQQNPMMVQQRQ-WSEEQAKI-QQNQQQIQQNPMM 627
...
Query:149 QRLARSQMWQQSSCHVMQQQCCQQLQQIPEQSRYEAIRAIIYSIILQEQQQGFVQPQQQQ 208
Q
R
W +
++QQ
QQ Q
+Q+R + +
+ ++Q+Q+Q
PQ Q
Sbjct:688 QMQQRQ--WTEDP-QMVQQM--QQRQWAEDQTRMQMAQQ---NPMMQQQRQMAENPQMMQ 739
Query:209 PQQSGQG---VSQSQQQSQQQLGQCSFQQPQQQLGQQPQ---QQQQQQVLQGT 255
+Q +
+ Q+QQ +QQ
Q
QQ QQ+
+ Q
QQQQ+Q++Q T
Sbjct:740 QRQWSEEQTKIEQAQQMAQQN--QMMMQQMQQRQWSEDQAQIQQQQRQMMQQT 790
You can see that almost all the matched residues are Q = Glutamine. It is doubtful if
this means anything more than that both genes happen to have a lot of CAG and
CAA codons! Certainly you'd want other independent information before concluding
that Wheat Gamma Gliadin and this Drosophila gene share a recent common
ancestor or a similar structure.
From the NCBI server, using low complexity masking, you find, among many other
hits, the following alignment:
sp|P06471|HOR3_HORVU B3-HORDEIN
Length = 264
Score = 62.5 bits (149), Expect = 1e-09
Identities = 32/63 (50%), Positives = 38/63 (59%)
Query: 131 LNPCKVFLQQQCSPVAMPQRLARSQMWXXXXXXXXXXXXXXXXXXXXXXXRYEAIRAIIY 190
LNPCKVFLQQQCSP+AM QR+ARSQM
R+EA+RAI+Y
Sbjct: 111 LNPCKVFLQQQCSPLAMSQRIARSQMLQQSSCHVLQQQCCQQLPQIPEQLRHEAVRAIVY 170
Query: 191 SII 193
SI+
Sbjct: 171 SIV 173
This is meaningful both statistically and biologically because it turns out the hordein
is a barley storage protein functionally equivalent to wheat gliadin.
Exercise:
87
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
1. Use SRS to find a mouse sequence in SwissProt. Try using any of the search
criteria used on day 1 or try the one of the following keywords from the course
homepage.
2. Carry out a blast search taking the default parameters to see if you can find a
human or a yeast homologue.
Try changing the substitution matrix or low
complexity masking to see if you can alter the order or composition of the 'hits'.
NB. Do NOT submit another search until the first result is
returned – especially at NCBI
88
M.Sc. in Molecular Medicine Bioinformatics Course
Day 2– Multiple Sequence Alignment
TOPICS
1. Introduction to multiple sequence alignment (MSA).
2. ClustalW
3. T-Coffee.
4. Multiple Sequence Alignment Editors.
89
Feb. 2005
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
Introduction
It is truism to say that there would be no genetics, and no very interesting
biology, but for the fact that there is variability between individuals and among
species. For years biological research depended on observable (bristle count, leaf
size, plumage colour, colony morphology) variations. Then it became possible to
document differences by using biochemical and other techniques (gram stain, lactose
metabolism, blood groups). Over the last two or three decades it has become possible
to get a rather direct measure of similarities and differences in the living world as
molecular biologists have succeeded in cloning and sequencing DNA from an
enormous variety of organisms. Notably a number of complete genomes have been
completely sequenced over the last eight or so years, ultimately giving us the genetic
and developmental blueprint for several living organisms. It is still many years
before we will collectively be able to make complete sense of, say, the 4 million base
pairs of the E. coli genome. Let alone the 1000x bigger human genome. One tool we
have already used for making sense of sequence is homology searching.
Another widely used bioinformatic technique is to try to align several related
sequences to find which residues/bases are conserved and which are variable. This
will help in the understanding of the constraints under which the sequences may
labour: conserved residues may be an essential part of the active site of an enzyme,
variable residues may be part of a 'generic' alpha-helix.
Multiple sequence alignment is also a vital prerequisite for trying to determine the
phylogenetic relationships among a group of related sequences - and by
extrapolation between the species or varieties that contain those sequences.
Multiple sequence alignment is very computationally intensive. The numbers
involved in evaluating all possible alignments between two sequences allowing gaps
in either is large. When 3 or more sequences are involved the numbers become so
large that the problem becomes incomputable. It requires an insight and a shortcut to
get biologically informative alignments in a finite time. One of the earliest
successful programs that could calculate a non-trivial multiple sequence alignment in
a reasonable time was invented in TCD in 1986 by Des Higgins. We will be using
web-based derivatives of the original clustal program that was written all those years
ago for incredibly primitive pre-windows PCs. The program is also freely and widely
available for PCs, Macs and Unix workstations. These standalone versions are
probably more sensitive and convenient than the WWW based version.
90
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
ClustalW
http://www2.ebi.ac.uk/clustalw/
http://www.ch.embnet.org/software/ClustalW.html
This web page allows you to make a multiple sequence alignment of any group (N
>= 2!) of sequences. The poor thing will attempt to align whatever sequences you
give it, but this may take a long time if the sequences are unrelated or numerous.
This is another example where a user-friendly program, which makes a lot of choices
for you by default, can be a poisoned chalice. There is a tendency among users to
believe that the computer or the program does the alignment and that this excuses the
humans involved from exercising judgment. There is even a widespread belief that
changing the options or particularly editing a delivered alignment is somehow
"unscientific" because it requires a subjective assessment of what is correct, sensible
and meaningful. This wrong-headed attitude is frequently compounded by loading
the computer generated multiple sequence alignment directly into a phylogenetic tree
drawing algorithm to determine the relationships amongst the included taxa. Such a
phylogeny program will, like ClustalW, try to do what it is asked to do and may
generate a tree that is, shall we say, fatuous.
Clustal is a program for computer-aided multiple sequence alignment. It takes
some of the grunt work out of the complex and time-consuming business of aligning
many sequences. It does this by the judicious insertion of gaps to represent the
insertions and deletions that have occurred over evolutionary time since the most
recent common ancestor of the sequences included. All users of the program are
morally and scientifically obliged to scrutinize critically the alignment and see how
it can be improved. There are numerous, colorful, multiple sequence alignment
editors available to help you do this.
The ClustalW home page is nicely designed because all the options and
parameters are visible on the one page as choice buttons. You can get a little help on
the effect of each of these choices by clicking on the hypertext link above the choice
button. Rather more information on the theory and practice of Clustal can be found
at:
http://www-igbmc.u-strasbg.fr/BioInfo/ClustalX/Top.html
91
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
The Clustal WWW servers invites you to "Enter or Paste a set of Sequences in any
Format", an invitation which should be treated with caution. FASTA format has
much to recommend it. In this format, each sequence is represented by a single title
line beginning with a ">" followed by the sequence itself on subsequent lines;
typically 60 residues or bases per line, thus:
>ACDRECAP.RECA
355
MDEPGGKIEFSPAFMQIEGQFGKGAVMRAGDKPGINDPDVKSTGSLGLDGALGQGGLPRG
RVVEIYGPESSGKTTLTLKAIASAQAEGATPAFTDAEHALDPGFASKLGVNVKRLLISQP
DTGEQALEIADMLFRSGAVDVIVKDSVAALTPKAEIEGEMGDSHQGLHARLMSQALRNKT
ANISRWNKLVIFKKQIRMKMGVYGRPETTTGGNALKFYASVRLDIRRMGAMKKSATKSYD
WSTRVKVVKNKVAPPFRQAELAIYYGEGIYRGSEPVDLGVKLENVEKSGGWYSYPGRRIG
QGKANARQYLRVKPEFPGIFEQGIRGAMAAPHPLGFGERRDVQQESGEPYGNNGX
>BRURECA.RECA
361
MSQNSLRLVEDNSVDKTKALDAALSQIERAFGKGSIMRLGQNDQVVEIETVSTGSLSLDI
ALGVGGLPKGRIVEIYGPESSGKTTLALHTIAEAQKKGGICAFVDAEHALDPVYARKLGV
HLENLLISQPITGEQALEITDTLVRSGAIDVLVVDSVAALTPRAEIEGEMGDSHGLQARL
MSQAVRKLTGSISRSNCMVIFINQIRMKIGVMFGSPETTTGGNALKFYASVRLDIRRIGS
IKERDEVVGNQTRVKVVKNKLAPPFKQVEFDIMYGAGVSKVGELVDLGVKAGVVEKSGAW
FSYNSQRLGQGRENAKQYLKDNPEVAREIETTLRQNAGLIAEQFLDDGGPEEDAAGAAMX
>NGRECAG.RECA
349
MSDDKSKALAAALAQIEKSFGKGAIMKMDGSQQEENLEVISTGSLGLDLALGVGGLRRGR
IVEIFGPESSGKTTLCLEAVAQCQKNGGVCAFVDAEHAFDPVYARKLGVKVEELYLSQPD
TGEQALEICDTLVRSGGIDMVVVDSVAALVPKAEIEGDMGDSHVGLQARLMSQALRKLTG
HIKKTNTLVVFINQIRMKIGVMFGSPETTTGGNALKFYSSVRLDIRRTGSIKKGEEVLGN
ETRVKVIKNKVAPPFRQAEFDILYGEGISWEGELIDIGVKNDIINKSGAWYSYNGAKIGQ
GKDNVRVWLKENPEISDEIDAKIRALNGVEMHITEGTQDETDGERPEEX
With a very highly conserved protein (histones or mammalian beta globins or recA
from gamma proteobacteria) it may well be possible to align sequences by hand and
eye and good judgement, using, say, Microsoft WORD. Nevertheless, this is likely to
be a time consuming process and becomes impossible if many gaps are required or if
the evolutionary relationship between the sequences is more tenuous.
Clustal works in a three step-process:
1) All sequences are aligned and compared to each other and a score or 'distance' is
calculated between each pair of sequences.
2) This matrix of distances between each pair of sequences is used to create a
'dendrogram' or phylogenetic tree among the included sequences. (This was
Des Higgins' key insight that cracked the problem open)
3) The dendrogram is used as the basis for constructing the real multiple sequence
alignment: basically the most closely related sequences or groups of sequences
are aligned first.
The quality of the alignment is determined by assigning a positive score to each pair
of identical residues which is aligned, and a lower or negative score to 'mismatches'.
92
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
The scores are read off from the substitution matrix which is in force (by default or
by choice). See Chapter on BLAST for more on substitution matrices.
The parameters most likely to affect the quality of the alignment are the gap
penalty (GAP OPEN), the gap-extension penalty (GAP EXTENSION) and, to a
lesser extent, the substitution matrix (MATRIX).
Gap Open Penalty.
If you attempt to align two sequences starting at the amino terminus or the 5'
end of the sequences and one of the sequences has a deletion, then the alignment is
likely to be very poor after the deletion unless a gap is inserted. This gap mimics the
biological reality that one sequence has lost one or more residues/bases. Usually we
don't know where the deletion has occurred or indeed if it is really an insertion in the
other sequence. Clustal attempts to estimate where such a deletion is most likely to
have happened. It does this with a Gap Penalty. The gap penalty is typically more
negative than the 'worst' mismatch. If the gap is correctly sited then the negative
score incurred by the gap penalty will be more than compensated for by enhanced
positive scores further down the alignment. A high gap penalty will discourage gaps,
while a very low gap penalty will allow gaps willy-nilly and so enable you to align
two completely unrelated sequences.
Gap Extension Penalty.
Most sequence alignment programs that work well use what are called affine
gap penalties, so that a gap of three bases/residues is not penalised three times more
heavily that a gap of one. This is taking account of the fact that a point deletion is
more or less as common as a longer one. So taking the default gap penalties from
the clustalWWW server (Open = 10, Ext=0.05) we get a score of -10 for a single
residue gap and -10.45 (10 + 9*0.05) for a gap of ten residues.
93
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
T-COFFEE
For distant or difficult alignments T-COFFEE is almost certain to give you a better
result than clustalW. It is freely available for download but is also available over the
web.
http://www.ch.embnet.org/software/TCoffee.html
Paste your PROTEIN sequences into the box on this page and click on the [run TCOFFEE] box. When the run is finished a
Here are your search results:
Will appear. There are a number of formats for outputting your alignment. You are
advised to choose phylip output if you plan to use that software suite for
constructing phylogenetic trees.
Exercise:
1) Choose any 5-10 sequences from the same family (defined by prosite?) or from
the results of a homologue search. Or from the list of mammalian sequences
which have more than one representative in SwissProt which are on the course
homepage.
2) Run them through the clustal WWW server taking the default parameters.
3) Critically evaluate the alignment:
a) if one sequence is much shorter than the others find out why - a partial
sequence ?
b) if one or two sequences seem to be distorting the alignment, consider
ejecting them and redoing the alignment.
c) can you improve the alignment by choosing different gap penalties ?
4) For a more difficult problem, fetch the following sequences and try to align them:
ftp://www.binf.org/pub/abc/casein.pep
or get them from the course website
5) If you can get a good alignment use the Jpred Predict Protein prediction server at
the EBI to see if the gaps appear in peptide loops (that might not be expected
to be essential to the structure and function of the enzyme).
6) Can you find the prosite motif that defines your family of proteins in your
multiple sequence alignment? Are the elements of that motif always
conserved?
94
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
7) Does T-COFFEE make a better fist of a “difficult” multiple sequence alignment
like the casein dataset?
Multiple sequence alignment editors
For reasons outlined at the beginning of this chapter it is important not to treat
multiple sequence alignment software as a black-box. You must scrutinize the
alignment created and almost certainly you will want to do some editing to align
motifs, cysteines, and hydrophobic residues. Each alignment will be different and
you can look up SwissProt or Pfam to discover structural information about, and
conserved residues peculiar to, your protein (family) of interest. Obviously, TCoffee and ClustalW can’t read PubMed, SwissProt – that’s your job. Try these
MSA editors:
On the WWW:
JalView: http://www.ebi.ac.uk/~michele/jalview/contents.html
See http://www.hgmp.mrc.ac.uk/embnet.news/vol5_4/embnet/jalview.html for a
description.
For MS-Windows:
Genedoc: http://www.psc.edu/biomed/genedoc
Note: There are many others
95
M.Sc. in Molecular Medicine Bioinformatics Course
DAY 2: Phylogenetic Trees
TOPICS:
1. Introduction to phylogenetics.
2. Using MEGA to construct phylogenetic trees.
3. Phylip.
4. WebPhylip.
5. A protocol for drawing trees on the WWW.
6. ClustalW for trees.
96
Feb. 2005
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
Introduction:
The introduction to the previous chapter indicates how multiple sequence alignment
can identify sites, regions and domains in your protein which are invariant, or
conserved, or hypervariable. MSA is also a prerequisite for constructing
phylogenetic trees. It is really important that you try to put your gene and protein of
interest in a correct evolutionary context – if you can determine where your gene
came from, and what its closest relatives are, you can get vital clues about the
structure, function and expression pattern of your gene. These clues may save you
months of work at the bench and thousands of euros in costs.
●
If you find that your human gene is most closely related to a constitutively
expressed mouse homologue, then your gene is less likely to be inducible.
●
If you find that your human gene is matched by two equally distant mouse
homologues, it may indicate that the functions of your gene have been divided
between the mouse genes (subfunctionalisation) or that one of the mouse genes
has acquired a new function (neofunctionalisation).
●
A comprehensive phylogenetic analysis may reveal that your mouse model has
more likely evolved independently from your human system of interest and so
will be a less appropriate or even wholly misleading.
●
Phylogenetic analysis of gene families can show that some genes are tissue
specific and form a closely-related grouping. Unknown genes in the same
group are perhaps more likely to share the same expression pattern.
●
A blast search against the mouse genome may find you the most closely related
mouse homologue to your gene. Reciprocal blast analysis may show that this
best hit is a poor model because it is more closely related to other human
genes. Effective phylogenetic analysis can sort the problem out.
●
As Multiple Sequence Alignment is an essential pre-requisite for phylogenetic
trees, so phylogenetic trees are an essential pre-requisite for an analysis of sites
undergoing positive selection, which are good likely targets for protein
interaction or drug-design.
97
M.Sc. in Molecular Medicine Bioinformatics Course
●
Feb. 2005
A good phylogenetic analysis with a clearly drawn tree can lubricate the
publication process, impress editors and over-awe referees.
A reasonable on-line introduction to the vocabulary and principles of phylogenetics
as well as to the resources available at the NCBI can be found at:
http://www.ncbi.nlm.nih.gov/About/primer/phylo.html
Phylogenetic tree construction is one of the most computationally intensive and
time-consuming applications in bioinformatics. There are, for example, in excess of
1,000,000 different trees that can be constructed from even as few as 10 taxa. Under
maximum likelihood and maximum parsimony algorithms each one of these trees
will be investigated and compared. Under such circumstances, it is unwise to rely on
a web-based resource; it is better to use a tree construction package locally.
Although you can run PHYLIP on the web for the course it is better for you to learn
how to access this package either via INCBI or on some other local server (PHYLIP
is available as free downloadable versions for PC and Mac). PAUP is also an
excellent general-purpose phylogenetics package, which is available for very little
money. In this course, we will make most use of the program MEGA, which is free
and user friendly. But we retain the sections on Phylip for completeness.
Methods for calculating trees are fairly controversial. Journal referees are likely to
have strong feelings on the matter of using maximum parsimony or maximum
likelihood. Neighbor joining tree may be acceptable to them only if your dataset is
so large that MP and ML will take a ludicrously long time to compute an answer. In
general, MP is losing ground to ML. And watch out for Bayesian methods that are
becoming increasingly fashionable. You should be able to a) use an appropriate
algorithm/program and b) justify your using it. In the time allotted in this course,
there will not be time to carry out a comprehensive investigation of the effects of
algorithm and parameter choice on phylogenetic tree construction. But we encourage
you to compare and contrast different methods, using a relatively small dataset in
your own time.
As elsewhere in the course, graphics are a problem in phylogenetics. A tree is
virtually impossible to interpret unless graphically displayed, yet it is difficult to get
98
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
satisfactory tree-display tools on the web. MEGA’s tree visualization is well
integrated into the package and this is one reason why we are using it as our primary
demonstration tool in the current course.
The online manual for Mega2 can be found:
http://www.megasoftware.net/WebHelp/mega2_help.htm
First catch your software:
http://www.megasoftware.net/
Installing MEGA
Sixth item on the left-hand menu takes you to Downloads:
http://www.megasoftware.net/text/downloads.sht
You must fill in
Last name: [_______________]
First Name: [_______________]
E-mail Address: [_______________]
(*)Autoinstall from web
Then click:
[Submit and Download]
Thereafter accept all defaults as you are walked through the installation process.
The following protocol will allow you to take a file of aligned sequences from
clustal, then construct and display a phylogenetic tree based on the alignment. In
addition, it uses a bootstrap approach to assess the degree of statistical confidence in
the various branches of the tree. It is largely mechanical in nature, a more thorough
treatment of the theory and practice appears later in this chapter.
Running MEGA
99
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
Having installed the software, a Mega2 icon should appear on your desktop.
1. To Begin -Click the Mega2 icon.
A main “Molecular Evolutionary Genetics Analysis, version 2.1” window
should appear with Windows-like Menu bar:
File
Phylogeny
Windows
Help
Then the following hypertext links:
Click me to activate a data file
Go to the MEGA2 web page
Citing MEGA2 in publications
2. Converting to MEGA format
As with almost all bioinformatic software, MEGA has its own idiosyncratic
format, so the first step is to convert your *.aln output from Clustal to *.meg
format:
File Æ Convert to MEGA Format
This will open a “Select File and Format” window that will a) let you browse
to find your .aln alignment file and b) convert files in a wide variety of formats
- including .aln (CLUSTAL) - to something MEGA can read.
Click [√ OK] to get:
A “MEGA2” window with
File conversion complete….
Click [OK]
And a .meg file should appear in the window, the top of which looks like:
#Mega
Title: act.aln
#ACT1_SCHCO
--MEDEVAALVIDNGSGMCKAGFAGDDAPRAVFPSIVGRPRHQGVMVGMGQKDSYVGDEA
QSKRGILTLKYPIEHGIVTNWDDMEKIWHHTFYNELRVAPEEHPVLLTEAPLNPKANREK
MTQIMFETFNAPAFYVAIQAVLSLYASGRTTGIVLDSGDGVTHTVPIYEGFALPHAILRL
DLAGRDLTDFLIKNLMERGYPFTTTAEREIVRDIKEKLCYVALDFEQELQTAAQSSALEK
SYELPDGQVITIGNERFRAPEALFQPAFLGLEAAGIHETTYNSIFKCDLDIRRDLYGNVV
LSGGTTMFP-GIADRMQKELTA
etc. etc.
3. Analyzing the data with MEGA
You can then return to the main “Molecular Evolutionary Genetics Analysis,
version 2.1” window and click the link:
Click me to activate a data file
100
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
In the “Choose a Data file to Analyze” window, select the .meg file you want
to analyze then click [Open].
In the “Input Data” window accept the default Protein Sequences then click
on [√ OK]
If the format is correct, the MEGA main menu should now have more items on
the Menu-bar:
File
Data
Distances
Phylogeny
Tests Windows
Help
4. Constructing a Neighbor-joining tree
Now do:
Phylogeny Æ Neighbor-joining (NJ)…
To create an “Analysis Preferences” window in which you can
Accept the default Model [Amino: Poisson correction] – not least because the
alternative Gamma Model requires you to estimate the Gamma parameter –
and then click on [√ OK]
A “Tree Explorer” window should appear with MEGA’s estimate of the
phylogenetic relationships among your sequences. Explore the buttons to see
how you can change the appearance of the tree using the Subtree and View
menus.
5. Statistical confidence in your tree.
A tree is only as good as the confidence you can put in it. This can be assessed
by bootstrapping your data. Return to the “Analysis Preferences” window,
then Test of Phylogeny Æ Change Test of Inferred Phylogeny from the default
(*) None
( ) Bootstrap
to
( ) None
(*) Bootstrap
then [√ OK]. The analysis will take appreciably longer (because it is being
bootstrap replicated 1000 times) and the “Tree Explorer” window will now
show numbers at each node. These are bootstrap values. By convention, you
can be reasonably confident in a clade (phylogenetic group) that has > 70
bootstrap support while 100 is very robust support for a grouping.
6. Other analysis with MEGA.
101
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
If your alignment is reasonable you can thus use Mega to generate a picture of
the phylogenetic relationships among your sequences and get a feel for its
statistical validity.
Neighbor-joining is widely seen to be an acceptable
method for inferring phylogeny. As you will have seen from the menu, Mega
will construct also UPGMA, Maximum Parsimony and Minimum Evolution
trees. Apart from the strong advice to NEVER use UPGMA to draw trees, you
will need more information to bring these other methods to bear on your data.
102
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
http://www.cbr.nrc.ca/cgi-bin/WebPhylip/index.html
http://biocore.unl.edu/WEBPHYLIP/
http://bioweb.pasteur.fr/seqanal/interfaces/neighbor-simple.html
PHYLIP is very widely used to construct phylogenetic trees. It is not so handy as
Mega, but does allow you to make Maximum Likelihood trees. We have left the
Phylip protocols in the manual to allow you to explore them in your own time.
Some of the compare and contrast exercises can be carried out within Mega if you
prefer. Note that the tree-drawing option in ClustalW has default PHYLIP output.
ClustalW trees can therefore be fed into PHYLIP programs such as “retree” to be
viewed.
This is a list of some of the programs in PHYLIP commonly used with sequence
data, together with a description of some of these programs :
Sequence type
Distance matrix
Max. parsimony
Max.likelihood
Protein
Protdist+Neighbor
Protpars
Protml
DNA
DNAdist+Neighbor
DNApars
fastDNAml
For bootstrapping (assessing the statistical significance of) your trees. You need
flanking steps:
Process
Phylip program
Randomly resample the data
Seqboot
Draw multiple trees
One protocol from prev. table with M
option
Consense
Calculate consensus among trees
Distance Matrix
103
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
PROTDIST/DNADIST calculates a distance matrix from aligned sequences. An
essential prerequisite for Neighbor.
NEIGHBOR. This is an implementation by Mary Kuhner and John Yamato of
Saitou and Nei's "Neighbor Joining Method," and of the UPGMA (Average Linkage
clustering) method. Neighbor Joining is a distance matrix method producing an
unrooted tree without the assumption of a clock. UPGMA does assume a clock.
There is NO reason why you should use UPGMA to draw a tree unless you are
reduced to a pencil and paper to calculate it. Neighbor-joining branch lengths are
not optimized by the least squares criterion but the method is very fast and thus can
handle much larger data sets. Note: Neighbor-joining is also incorporated into
ClustalW.
Maximum Parsimony
Maximum parsimony attempts to count the number of evolutionary steps (mutations)
that are necessary to construct trees of different topology. It tries to investigate all
possible trees and determine which is most parsimonious – which requires fewest
evolutionary steps. It has difficulty trying to determine exactly where a mutation has
occurred on the tree. One consequence of this is that, while MP tries to work out the
topology of trees (branching order) it gives up when trying to assign branch lengths.
PROTPARS.
This estimates phylogenies from protein sequences (input using
standard one-letter code for amino acids) using the maximum parsimony method, in
a variant which counts only those nucleotide changes that change the amino acid, on
the assumption that silent changes are more easily accomplished.
DNAPARS. This applies Maximum Parsimony to DNA datasets.
Maximum Likelihood.
Many journals or referees may now insist on Maximum Likelihood trees. PHYLIP
has a Maximum Likelihood algorithm for DNA sequence data, DNAML. A very
similar, not strictly PHYLIP program called FastDNAML is available as a
replacement on the PISE site. You should add this to your armory of software now.
For protein datasets, PROTML is a Phylip-like option for maximum likelihood trees.
The software packages PAUP and Mega also have options for this sort of analysis.
PROTML drawing Maximum Likelihood trees with aligned protein data.
DNAML and FastDNAML for drawing DNA Maximum Likelihood trees.
104
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
RETREE. This reads in a tree (with branch lengths if necessary) and allows you to
reroot the tree, to flip branches, to change species names and branch lengths, and
then write the result out. It can be used to convert between rooted and unrooted
trees.
DRAWTREE and DRAWGRAM These plot unrooted phylogenies, cladograms,
and phenograms in a wide variety of user-controllable formats. Neither calculates
trees; they merely draw them.
SEQBOOT bootstraps for trees
CONSENSE majority rule (consensus) trees used with SEQBOOT
PHYLIP has been implemented on the web in two fundamentally different ways. As
WebPhylip on the Canadian site and as part of the PISE, EMBOSS suite at Pasteur.
Trees from DNA sequences (a warning)
It is also important to realise that the phylogenetic trees drawn from protein
sequences may differ from the trees drawn from the DNA sequences of the same
gene. Obviously there is more information in DNA trees than their protein
equivalent (silent sites etc) but some of this information may be confounding or
confusing. Silent sites get 'saturated' beyond a certain evolutionary distance - they
become essentially random and without meaningful information content. On the
other hand spurious associations may appear merely because two organisms or
sequences have a similar aberrant G+C content.
An argument could be made that trees drawn from aligned DNA sequences
with an appropriate model using maximum likelihood are the best trees you can
currently present. Nevertheless you should, if your sequence is coding, translate your
DNA sequences into protein to construct the alignment, then use a copygaps
program to transfer the gaps to the DNA sequence. Any coding DNA alignment that
has gaps of one or two residues, breaking up codons is almost certain to be wrong.
Different data = different tree?
Using the same group of animals but a different protein, the phylogenetic
relationships will not always appear the same. You should not, therefore, assume
that the phylogenetic tree derived from a particular class of proteins is the definitive
phylogenetic relationship for the species. To demonstrate how the trees can be
105
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
different you will do more alignments and trees for a different protein from the same
species. We have a dataset of somatotropin protein sequences on the course website.
Exercises
There are three basic datasets to try out during the practicals today:
1.Casein proteins
2.Casein DNA, CDSs cognate with those proteins
3.Somatotropin proteins
In each case, four mammalian taxa are represented: lagomorphs, rodents, artiodactyls and
primates.
The object of the exercises is to see what effect
•
different algorithms (ML, MP, NJ)
•
different implementations of the same algorithm (Pise, WebPhylip)
•
different datasets (caseins, somatotropins)
•
different sorts of data (DNA, protein)
have on the problem of inferring the taxonomic relationships among these mammalian orders.
This is not a trivial issue as shown by 3 papers in Nature in 2001 attempting to give a definitive
answer to the problem. The subtext is to show that choice of program, options and parameters
can significantly affect your attempts to explain relationships among YOUR taxa/genes/proteins.
You are advised to construct a series of controlled experiments, keeping everything the same
except one variable and comparing the results. Examples might be:
PISE protpars, default parameters: casein vs somatotropin.
PISE protpars vs Webphylip Protpars: casein protein dataset.
106
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
A step by step PHYLIP protocol
With the unreliability of the web we have decided to use Windows Phylip installed
here in UCD. You can either install Phylip on your own PC (easy, but make sure
you have a reasonably powerful machine) or transfer the following protocol to any of
the WWW implementations. There follows an illustrated protocol for calculating a
bootstrapped neighbor joining distance matrix based phylogenetic tree. Phylip has a
particular format for its input multiple sequence alignment. It is recognizable by
having two numbers (the number of taxa and number of sites) on the first line:
BRU
RLR
NGR
ECO
YPR
PSE
TTH
ACD
8
370
MSQNSLRLVE
------------------------------------M
----------------------------
DNSV-DKTKA
---V-DKSKA
-MSD-DKSKA
AIDE-NKQKA
AIDE-NKQKA
-MDD-NKKRA
-MEE-NKRKS
-MDEPGGKIE
LDAALSQIER
LEAALSQIER
LAAALAQIEK
LAAALGQIEK
LAAALGQIEK
LAAALGQIER
LENALKTIEK
FSPAFMQIEG
AFGKGSIMRL
SFGKGSIMKL
SFGKGAIMKM
QFGKGSIMRL
QFGKGSIMRL
QFGKGAVMRM
EFGKGAVMRL
QFGKGAVMRA
GQNDQVVEIE
GSNENVVEIE
DGSQQEENLE
GEDRS-MDVE
GEDRS-MDVE
GDHER-QAIP
GEMPK-LQVD
GDKPGINDPD
TVSTGSLSLD
TISTGSLGLD
VISTGSLGLD
TISTGSLSLD
TISTGSLSLD
AISTGSLGLD
VIPTGSLGLD
VKSTGSLGLD
IALGVGGLPK
IALGVGGLPR
LALGVGGLRR
IALGAGGLPM
IALGAGGLPM
IALGIGGLPK
LALGIGGIPR
GALGQGGLPR
GRIVEIYGPE
GRIIEIYGPE
GRIVEIFGPE
GRIVEIYGPE
GRIVEIYGPE
GRIVEIYGPE
GRVTEIFGPE
GRVVEIYGPE
SSGKTTLALH
SSGKTTLALQ
SSGKTTLCLE
SSGKTTLTLQ
SSGKTTLTLQ
SSGKTTLTLS
SGGKTTLALT
SSGKTTLTLK
TIAEAQKKGG
TIAEAQKKGG
AVAQCQKNGG
VIAAAQREGK
VIAAAQREGK
VIAEAQKNGA
IIAQAQKGGG
AIASAQAEGA
You can create this from within clustalW from the multiple sequence alignment (2)
menu; output formats (9) option choosing format (4) Toggle PHYLIP format
output.
This sequence needs to go through the following steps but care needs to be taken
with the input and output file names:
1. Protdist – expects a phylip format sequence alignment file called infile, if it
cannot find a file with that name it asks for input filename:
protdist:
can't read infile
Please enter a new filename> rec8.phy (or whatever your file is called)
The next menu will then appear:
Protein distance algorithm, version 3.573c
Settings for this run:
P
Use PAM, Kimura or categories model?
M
Analyze multiple data sets?
No
I
Input sequences interleaved?
Yes
0
Terminal type (IBM PC, VT52, ANSI)?
1
Print out the data at start of run
107
Dayhoff PAM matrix
ANSI
No
M.Sc. in Molecular Medicine Bioinformatics Course
2
Feb. 2005
Print indications of progress of run
Yes
Are these settings correct? (type Y or the letter for one to change)
(Type P to change to a Kimura substitution model?) then Y to accept settings. Which
starts the program:
Computing distances:
BRU
RLR
.
NGR
..
ECO
...
YPR
....
PSE
.....
TTH
......
ACD
.......
Output written to output file
Creates file called outfile – rename this as, say, file.dst
2. Neighbor – expects a distance matrix called infile, if it cannot find a file with that
name it asks for input filename:
neighbor:
can't read infile
Please enter a new filename> file.dst
An acceptable/readable filename will give you this menu:
Neighbor-Joining/UPGMA method version 3.5
Settings for this run:
N
Neighbor-joining or UPGMA tree? Neighbor-joining
O
Outgroup root? No, use as outgroup
species 1
L
Lower-triangular data matrix? No
R
Upper-triangular data matrix? No
S
Subreplicates? No
J
Randomize input order of species? No. Use input order
M
Analyze multiple data sets? No
0
Terminal type (IBM PC, VT52, ANSI)? ANSI
1
Print out the data at start of run No
2 Print indications of progress of run Yes
3
Print out tree Yes
4
Write out trees onto tree file? Yes
Are these settings correct? (type Y or the letter for one to change)
Y
For this run you can accept all the settings by typing Y.
Which gives you the following on the screen:
CYCLE
CYCLE
CYCLE
CYCLE
5:
4:
3:
2:
OTU
OTU
OTU
NODE
1
4
7
1
(
(
(
(
0.07129)
0.04916)
0.31996)
0.13391)
JOINS
JOINS
JOINS
JOINS
108
OTU
OTU
OTU
OTU
2
5
8
3
(
(
(
(
0.06745)
0.07220)
0.37166)
0.20854)
M.Sc. in Molecular Medicine Bioinformatics Course
CYCLE
1: NODE 4 (
0.12581) JOINS OTU
LAST CYCLE:
NODE 1 (
0.02671) JOINS NODE 4 (
0.03301)
Feb. 2005
6 (
0.17116)
0.02300) JOINS NODE
7
(
Output written on output file
Tree written on tree file
The output file called treefile shows the topology of the tree:
(((ECO:0.04916,YPR:0.07220):0.12581,PSE:0.17116):0.02300,
(TTH:0.31996,ACD:0.37166):0.03301,((BRU:0.07129,RLR:0.06745):0.13391
,NGR:0.20854):0.02671);
This is the answer! The hierarchy of brackets tells you the relationships amongst the
taxa and the numbers the relative branch lengths. The format of this file is called
Newick or NewHampshire format and can be used by DRAWGRAM, DRAWTREE,
TreeView or GeneDoc to print a picture of the tree.
The output file outfile gives some further details of tree construction.
Bootstrapping your tree.
As a biologist you will want to have some idea of how confident you can be in this
tree. The standard way of determining this confidence is to do a bootstrap analysis.
The mechanics of bootstrapping within PHYLIP are laborious but necessary. It is a
few extra steps anyway.
1. Seqboot – which creates a number (default =100) of random resamplings of the
sequence input dataset. It expects a phylip format sequence alignment file called
infile, if it cannot find a file with that name it asks for input filename:
seqboot: can't read infile
Please enter a new filename>rec8.phy
Random number seed (must be odd)?
59
Bootstrapped sequences algorithm, version 3.573c
Settings for this run:
D
Sequence, Morph, Rest., Gene Freqs?
J
Bootstrap, Jackknife, or Permute?
R
How many replicates?
I
Input sequences interleaved?
0
Terminal type (IBM PC, VT52, ANSI)?
1
Print out the data at start of run
2 Print indications of progress of run
Molecular sequences
Bootstrap
100
Yes
ANSI
No
Yes
Are these settings correct? (type Y or the letter for one to change)
Y
completed replicate number
10
109
M.Sc. in Molecular Medicine Bioinformatics Course
completed
completed
completed
completed
completed
completed
completed
completed
completed
replicate
replicate
replicate
replicate
replicate
replicate
replicate
replicate
replicate
number
number
number
number
number
number
number
number
number
Feb. 2005
20
30
40
50
60
70
80
90
100
Output written to output file
Rename outfile to something like rec8boot.phy
2. Protdist - expects a phylip format sequence alignment file called infile, if it
cannot find a file with that name it asks for input filename:
protdist: can't read infile
Please enter a new filename>rec8boot.phy
Protein distance algorithm, version 3.573c
Settings for this run:
P Use PAM, Kimura or categories model?
M
Analyze multiple data sets?
I
Input sequences interleaved?
0
Terminal type (IBM PC, VT52, ANSI)?
1
Print out the data at start of run
2 Print indications of progress of run
Dayhoff PAM matrix
No
Yes
ANSI
No
Yes
Are these settings correct? (type Y or the letter for one to change)
M
How many datasets?
100
When bootstrapping you must toggle M for multiple datasets.
The settings are confirmed:
Settings for this run:
P Use PAM, Kimura or categories model?
M
Analyze multiple data sets?
I
Input sequences interleaved?
0
Terminal type (IBM PC, VT52, ANSI)?
1
Print out the data at start of run
2 Print indications of progress of run
Dayhoff PAM matrix
Yes, 100 sets
Yes
ANSI
No
Yes
Are these settings correct? (type Y or the letter for one to change)
Y
Output written to output file
The outfile looks like this:
Data set # 1:
Computing distances:
BRU
RLR
.
110
M.Sc. in Molecular Medicine Bioinformatics Course
NGR
ECO
YPR
PSE
TTH
ACD
Feb. 2005
..
...
....
.....
......
.......
Output written to output file
etc etc 100 times until
Data set # 100:
Computing distances:
BRU
RLR
.
NGR
..
ECO
...
YPR
....
PSE
.....
TTH
......
ACD
.......
Output written to output file
Rename outfile as, say, rec8boot.dst
3. Neighbor - expects a distance matrix called infile, if it cannot find a file with that
name it asks for input filename. As with protdist you have to toggle M for multiple
datasets:
neighbor: can't read infile
Please enter a new filename>rec8boot.dst
Neighbor-Joining/UPGMA method version 3.5
Settings for this run:
N
Neighbor-joining or UPGMA tree? Neighbor-joining
O
Outgroup root? No, use as outgroup species 1
L
Lower-triangular data matrix? No
R
Upper-triangular data matrix? No
S
Subreplicates? No
J
Randomize input order of species? No. Use input order
M
Analyze multiple data sets? No
0
Terminal type (IBM PC, VT52, ANSI)? ANSI
1
Print out the data at start of run No
2 Print indications of progress of run Yes
3
Print out tree Yes
4
Write out trees onto tree file? Yes
Are these settings correct? (type Y or the letter for one to change)
M
How many data sets?
100
The settings are confirmed:
Neighbor-Joining/UPGMA method version 3.5
Settings for this run:
111
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
N
Neighbor-joining or UPGMA tree? Neighbor-joining
O
Outgroup root? No, use as outgroup
species 1
L
Lower-triangular data matrix? No
R
Upper-triangular data matrix? No
S
Subreplicates? No
J
Randomize input order of species? No. Use input order
M
Analyze multiple data sets? Yes, 100 sets
0
Terminal type (IBM PC, VT52, ANSI)? ANSI
1
Print out the data at start of run No
2 Print indications of progress of run Yes
3
Print out tree Yes
4
Write out trees onto tree file? Yes
Are these settings correct? (type Y or the letter for one to change)
Y
Gazinnng! There is a lot of screendump ending with:
Data set # 100:
CYCLE
5: OTU
4 (
CYCLE
4: OTU
1 (
CYCLE
3: OTU
7 (
CYCLE
2: OTU
6 (
CYCLE
1: NODE 1 (
LAST CYCLE:
NODE 1 (
0.02042)
0.03249)
0.05709)
0.04365)
0.28861)
0.13853)
0.11455)
JOINS
JOINS
JOINS
JOINS
JOINS
JOINS NODE
4
OTU
OTU
OTU
NODE
OTU
(
5
2
8
7
3
(
(
(
(
(
0.07509)
0.07456)
0.31413)
0.05314)
0.21747)
0.13815) JOINS NODE
6
(
Output written on output file
Tree written on tree file
Rename treefile as rec8boot.tree
Rename outfile as rec8boot.out
4. Consense: run consense on treefile or what ever you renamed it as. Consense
sorts through the multiple trees (one for each resampling of the original dataset) and
decides what is the consensus tree.
consense: can't read infile
Please enter a new filename>rec8boot.tree
Majority-rule and strict consensus tree program, version 3.573c
Settings for this run:
O
Outgroup root?
species 1
R
Trees to be treated as Rooted? No
0
Terminal type (IBM PC, VT52, ANSI)? ANSI
1
Print out the sets of species Yes
2 Print indications of progress of run Yes
3
Print out tree Yes
4
Write out trees onto tree file? Yes
No, use as outgroup
Are these settings correct? (type Y or the letter for one to change)
Y
Output written to output file
112
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
Tree also written onto file
The outfile looks like this:
Majority-rule and strict consensus tree program, version 3.573c
Species in order:
ECO
YPR
PSE
TTH
ACD
BRU
RLR
NGR
Sets included in the consensus tree
Set (species in order)
How many times out of 100.00
.....**.
..******
.....***
...*****
...**...
100.00
100.00
85.00
79.00
67.00
Sets NOT included in consensus tree:
Set (species in order)
How many times out of 100.00
...*.***
..***...
..*.*...
...*.**.
...****.
..***..*
...*...*
...**..*
..**.***
29.00
13.00
11.00
5.00
4.00
3.00
2.00
1.00
1.00
CONSENSUS TREE:
the numbers at the forks indicate the number
of times the group consisting of the species
which are to the right of that fork occurred
among the trees, out of 100.00 trees
+----RLR
+100.0
+-85.0
+----BRU
!
!
+-79.0
+---------NGR
!
!
!
!
+----TTH
+100.0
+------67.0
!
!
+----ACD
+100.0
!
!
!
+-------------------PSE
!
!
!
+------------------------YPR
113
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
!
+-----------------------------ECO
remember: this is an unrooted tree!
While the treefile looks like this:
((((((RLR:100.0,BRU:100.0):100.0,NGR:100.0):85.0,(TTH:100.0,ACD:100.
0):67.0):79.0,PSE:100.0):100.0,YPR:100.0):100.0,ECO:100.0);
Here the numbers are not branch lengths but the number of times OTUs group
together when their data is resampled with replacement. You can take this treefile
(but rename it to something more memorable) and draw the Phylogenetic tree which
has been calculated with Drawgra, Drawtree, TreeView or Phylodendron on the web.
WebPhylip
This is implemented in three windows. Top right has the blurb and documentation;
lower right has the applications; while the left side is devoted to a hierarchical menu
of applications, which looks like:
1. Seq. Data
Conversion
This module will convert clustalw alignments into Phylip format suitable for the
programs.
2. Distance
Computation
This module will construct distance matrices from sequence data.
3. Data
Sampling
Very important module for getting access to bootstraps and other methods for
assessing the statistical significance of your best tree(s).
4. Phylogeny Methods for:
DNA
Protein
Res. Sites
Gene Freq.
0-1 Data
114
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
Dist. Matrix
Various methods available for different sorts of input data. We will only be dealing
with DNA and Protein input on this course.
5. Tree Consen.
6. Tree Editor
7. Plot Trees
This module might be used to get pictures of the trees you construct.
PISE PHYLIP
The Pasteur Phylip site contains a more straight forward list of programs that are
available. These include the following, which deal with protein and DNA sequences.
In most cases the default form is pretty unsatisfactory for anything except a taster.
The advanced forms give you the opportunity to change parameters. The *.doc links
will get you to the relevant page of the Phylip manual.
Programs for molecular sequence data [ sequence.doc ]
DNA
dnadist [ advanced form ] [ dnadist.doc ]
Distances from DNA sequences.
dnapars [ advanced form ] [ dnapars.doc ]
Parsimony method for DNA.
dnaml
(Maximum likelihood method) has been removed ; please use
rather fastDNAMLwhich is much faster and equivalent.
Proteins
protdist [ advanced form ] [ protdist.doc ]
Distances from protein sequences.
protpars [advanced form ] [ protpars.doc ]
115
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
Parsimony method for protein sequences.
Programs for distance matrix data [ distance.doc ]
neighbor[ advanced form ] [ neighbor.doc ]
Neighbor-joining and UPGMA methods
fitch [ advanced form ] [ fitch.doc ]
Fitch-Margoliash and least-squares methods
kitsch [ advanced form ] [ kitsch.doc ]
Fitch-Margoliash and least-squares methods with molecular clock
Programs for trees
drawtree [ advanced form ] [ drawtree.doc ]
Draw a tree (as computed before by phylogenetic analysis methods).
drawgram [ advanced form ] [ drawgram.doc ]
Draw a phenogram (see comment for drawtree).
If you now run DRAWTREE as before this file is automatically read into the
program and you can print out the tree as before. Note, however that in contrast with
the neighbor-joining tree this tree does not have branch lengths (all are the same
arbitrary length). This is a convention in maximum parsimony.
A protocol for drawing trees using the WWW.
First make your alignment
Take one of the protein sequence datasets from the course homepage. In this
example I have used the recA proteins. Open the page Æ crtl+A Æ ctrl+C to copy
the data.
Go from the course homepage to the ClustalW SIB, CH website and paste
ctrl+V your sequence in, alter any parameters or take the defaults, click [Run
ClustalW].
When the analysis is done, you will have a number of formats to choose from:
Here are your search results:
116
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
Multiple alignments
ClustalW (aln)
GCG/MSF
PIR
GDE
phylip
Thank you for using ClustalW
Click on phylip and you should see a page looking like (note the crucial first line
which tells phylip programs how many taxa/sequences there are and how long each
one is):
7
369
Bordatella
Ecoli
Neisseria
Pseudomona
Rhizobium
Thiobacill
Yersinia
MSQNSLRLVE
------------------------------------------------------M
DNSVDKTKAL
AIDENKQKAL
-MSDDKSKAL
-MDDNKKRAL
---VDKSKAL
-MEENKRKSL
AIDENKQKAL
DAALSQIERA
AAALGQIEKQ
AAALAQIEKS
AAALGQIERQ
EAALSQIERS
ENALKTIEKE
AAALGQIEKQ
FGKGSIMRLG
FGKGSIMRLFGKGAIMKMD
FGKGAVMRMFGKGSIMKLG
FGKGAVMRLFGKGSIMRL-
QNDQVVEIET
GEDRSMDVET
GSQQEENLEV
GDHERQAIPA
SNENVVEIET
GEMPKLQVDV
GEDRSMDVET
VSTGSLSLDI
ISTGSLSLDI
ISTGSLGLDL
ISTGSLGLDI
ISTGSLGLDI
IPTGSLGLDL
ISTGSLSLDI
ALGVGGLPKG
ALGAGGLPMG
ALGVGGLRRG
ALGIGGLPKG
ALGVGGLPRG
ALGIGGIPRG
ALGAGGLPMG
RIVEIYGPES
RIVEIYGPES
RIVEIFGPES
RIVEIYGPES
RIIEIYGPES
RVTEIFGPES
RIVEIYGPES
SGKTTLALHT
SGKTTLTLQV
SGKTTLCLEA
SGKTTLTLSV
SGKTTLALQT
GGKTTLALTI
SGKTTLTLQV
IAEAQKKGGI
IAAAQREGKT
VAQCQKNGGV
IAEAQKNGAT
IAEAQKKGGI
IAQAQKGGGV
IAAAQREGKT
crtl+A Æ ctrl+C to copy the data.
If you are happy with the quality of the alignment you might want to save a local
copy of this Phylip format alignment locally on your desktop.
PISE Phylip at the Pasteur.
Then go to the Pasteur page choose the protpars link and paste in your phylip format
sequences then click the [Run Protpars] button.
When the analysis is complete you should get
Results:
outfile
treefile
[Run the selected program on treefile]
params
protpars.out
standard error file
117
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
Click on the outfile link to see the a crude most parsimonious tree and a report on
how many evolutionary steps it requires (694 in this case):
Protein parsimony algorithm, version 3.573c
One most parsimonious tree found:
+--------------Rhizobium
!
+--4
+--Thiobacill
! ! +--------5
! ! !
+--Neisseria
! +--2
!
!
+-----Pseudomona
--1
+-----3
!
! +--Yersinia
!
+--6
!
+--Ecoli
!
+-----------------Bordatella
remember: this is an unrooted tree!
requires a total of
694.000
and treefile which looks like a NH format tree:
((Rhizobium,((Thiobacill,Neisseria),(Pseudomona,(Yersinia,Ecoli)))), Bordatella);
Or directly choose drawgram and click on [Run the selected program on treefile]
From the list of output options choose X Bitmap
L
M
R
J
K
H
D
B
E
C
O
T
P
X
F
Apple Laserwriter (with Postscript)
MacDraw PICT format
Rayshade 3D rendering program file
Hewlett-Packard Laserjet
TeKtronix 4010 graphics terminal
Hewlett-Packard 7470 plotter
DEC ReGIS graphics (VT240 terminal)
Houston Instruments plotter
Epson MX-80 dot-matrix printer
Prowriter/Imagewriter dot-matrix printer
Okidata dot-matrix printer
Toshiba 24-pin dot-matrix printer
PC Paintbrush monochrome PCX file format
X Bitmap format
FIG 2.0 format
Then click [Run Drawgram]
Click on plotfile link to view your tree.
If your browser does not automatically open this file, Save As tree.xbm or tree.bmp
and then open it from the desktop or Temp folder.
WebPhylip, at U.Nebraska,Lincoln or CBR at Halifax, Nova Scotia
From the left-hand menu click on 4. Phylogeny Methods for: Protein.
118
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
Then click on
1. Parsimony Help Run
Paste in your sequence, change any parameters, or accept defaults then [Submit]
Your outfile (a crude tree and the number of steps) and treefile (NH format tree)
should appear one above the other in the top right hand box.
From the left hand menu click on draw trees then
1. Draw Cladograms Help Run
in the lower right box change
Output Format:[postscript] Style of tree [phenogram] to [X bitmap]
and
Use tree file from last stage? [no] if no, type below. to [Yes]
Then [Submit]
Your tree should appear in a separate window. Internet explorer asks if you want to
save the file or open it in its current location – you want to open it.
ClustalW for trees
ClustalW also constructs neighbour-joining trees. Assuming that you already have a
multiple sequence alignment (from the last session), start ClustalW locally and
choose option 4 to get the “Phylogenetic Tree Menu”.
****** PHYLOGENETIC TREE MENU ******
1. Input an alignment
119
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
2. Exclude positions with gaps?
= OFF
3. Correct for multiple substitutions? = OFF
4. Draw tree now
5. Bootstrap tree
6. Output format options
S. Execute a system command
H. HELP
or press [RETURN] to go back to main menu
Your choice:
Choose 1 to input your alignment
Sequences should all be in 1 file.
6 formats accepted:
NBRF/PIR, EMBL/SwissProt, Pearson (Fasta), GDE, Clustal, GCG/MSF.
Enter the name of the sequence file: cas1.aln
Sequence format is Clustal
Sequences assumed to be PROTEIN
Sequence
Sequence
Sequence
Sequence
Sequence
Sequence
Sequence
1:
2:
3:
4:
5:
6:
7:
CAS1_BOVIN
CAS1_HUMAN
CAS1_MOUSE
CAS1_PIG_I
CAS1_RABIT
CAS1_RAT_I
CAS1_SHEEP
328
328
328
328
328
328
328
aa
aa
aa
aa
aa
aa
aa
You should now have returned to the Main Menu. Chose 4 to enter the Phylogenetic
Tree Menu. Unless you have compelling reasons to do otherwise, choose options 2
and 3 to correct for multiple hits and to exclude all gaps. Choose 4 again to draw a
neighbour-joining tree, hit <return> to accept the default filename.
Then hit
<return> again to return to the main menu.
Using clustalw as a format converter.
Later in this exercise you will use the PHYLIP tree drawing program PROTPARS to
draw a tree using the parsimony method. PHYLIP does not align sequences it only
draws trees, so to do this you will need to input an alignment in PHYLIP format.
The alignment just done is in ClustalW format so to obtain an alignment in PHYLIP
format you will need to change the output options (You can do this now or later, if
you do it later after exiting ClustalW you will need to feed in the input file cas1.aln
again ).
120
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
To do this choose option 2 (multiple alignments) from the main menu to give you the
Multiple Alignment Menu again. This time choose option 9 to change the output
format, the following menu will appear on your screen :
********* Format of Alignment Output *********
1. Toggle CLUSTAL format output
= ON
2. Toggle NBRF/PIR format output = OFF
3. Toggle GCG/MSF format output
= OFF
4. Toggle PHYLIP format output
= OFF
5. Toggle GDE format output
= OFF
6. Toggle GDE output case
7. Toggle CLUSTALW seq numbers
8. Toggle output order
=
=
=
LOWER
OFF
ALIGNED
9. Create alignment output file(s) now?
0. Toggle parameter output
= OFF
H. HELP
Enter number (or [RETURN] to exit):
Choose options 1 then 4 to turn Clustal format off and PHYLIP format on. Now
choose option 9 and then hit <return> to accept the default filename (cas1.phy).
When this has finished, exit from ClustalW by returning to the main menu and
entering x.
Graphics: drawing the tree you just calculated.
You can view the neighbour-joining tree you have just drawn in ClustalW using the
PHYLIP program RETREE, or you can get a hard copy with DRAWGRAM or
DRAWTREE or phylodendron http://iubio.bio.indiana.edu/treeapp/ to display it
in 2 dimensions.. All of these programs are for displaying the tree rather than
determining its topology. Or use Mega2.
121
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
Printed sources about BioInformatics & the InterNet.
Briefings in Bioinformatics - a journal aimed at users rather than developers with
useful review and how-to articles.
Books:
Bioinformatics : A Practical Guide to the Analysis of Genes and Proteins. Andreas
Baxevanis & B.F.Francis Ouellette (Eds). John Wiley & Sons 2nd Ed 2001; ISBN:
0471-38390-2 The Course text book!
Fundamentals of Molecular Evolution. W-H Li and D Graur. Sinauer 1991. ISBN
0 87893 452 9
Fundamentals of Molecular Evolution. D Graur and W-H Li . Sinauer 2000. ISBN
0-87893-266-6
PAUP 4.0 Phylogenetic Analysis Using Parsimony (and other methods) Manual.
David L Swofford. Sinauer 1999. 0 87893 801 X
Introduction to Bioinformatics. TK Attwood & DJ Parry-Smith. Addison Wesley
Longman 1999. ISBN 0582 32788 1.
Molecular Evolution: a phylogenetic approach. RDM Page and EC Holmes.
Blackwell 1998. ISBN: 0-86542-889-1
Bioinformatics for Dummies. Notredame and Claverie. 2003
Articles:
Baldauf, SL (2003) Phylogeny for the faint of heart: a tutorial. TIG 19(6): 345-351.
122
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
APPENDIX I
SEQUENCE SYMBOLS
Nucleotides
IUB code
MEANING
COMPLEMENT
A
C
G
T/U
M
R
W
S
Y
K
V
H
D
B
X/N
.
A
C
G
T
A or C
A or G
A or T
C or G
C or T
G or T
A or C or G
A or C or T
A or G or T
C or G or T
G or A or T or C
not G or A or T or C
T
G
C
A
K
Y
W
S
R
M
B
D
H
V
X
.
Amino Acids
SYMBOL
MEANING
A
B
C
D
E
F
G
H
I
K
L
M
N
P
Q
R
S
T
V
W
X
Y
Z
*
Ala
GCT, GCC, GCA, GCG
Asp, Asn GAT, GAC, AAT, AAC
Cys
TGT, TGC
Asp
GAT, GAC
Glu
GAA, GAG
Phe
TTT, TTC
Gly
GGT, GGC, GGA, GGG
His
CAT, CAC
Ile
ATT, ATC, ATA
Lys
AAA, AAG
Leu
TTG, TTA, CTT, CTC, CTA, CTG
Met
ATG
Asn
AAT, AAC
Pro
CCT, CCC, CCA, CCG
Gln
CAA, CAG
Arg
CGT, CGC, CGA, CGG, AGA, AGG
Ser
TCT, TCC, TCA, TCG, AGT, AGC
Thr
ACT, ACC, ACA, ACG
Val
GTT, GTC, GTA, GTG
Trp
TGG
Unknown
Tyr
TAT, TAC
Glu, Gln GAA, GAG, CAA, CAG
Terminator TAA, TAG, TGA
CODONS
IUB code
123
!GCX
!RAY
!TGY
!GAY
!GAR
!TTY
!GGX
!CAY
!ATH
!AAR
!TTR, CTX, YTR
!ATG
!AAY
!CCX
!CAR
!CGX, AGR, MGR
!TCX, AGY
!ACX
!GTX
!TGG
!XXX
!TAY
!SAR
!TAR, TRA
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
APPENDIX II
The Universal Genetic Code.
Phe UUU
UUC
Leu UUA
UUG
Ser UCU
UCC
UCA
UCG
Tyr UAU
UAC
ter UAA
ter UAG
Cys UGU
UGC
ter UGA
Trp UGG
Leu CUU
CUC
CUA
CUG
Pro CCU
CCC
CCA
CCG
His CAU
CAC
Gln CAA
CAG
Arg CGU
CGC
CGA
CGG
Ile AUU
AUC
AUA
Met AUG
Thr ACU
ACC
ACA
ACG
Asn AAU
AAC
Lys AAA
AAG
Ser AGU
AGC
Arg AGA
AGG
Val GUU
GUC
GUA
GUG
Ala GCU
GCC
GCA
GCG
Asp GAU
GAC
Glu GAA
GAG
Gly GGU
GGC
GGA
GGG
Exceptions to the Universal Code so far discovered:
#1: Yeast Mitochondrial Code: CUN=T AUA=M UGA=W
#2: Mitochondrial Code of Vertebrates: AGR=* AUA=M UGA=W
#3: Mitochondrial Code of Filamentous fungi: UGA=W
#4: Mitochondrial Code of Insects and platyhelminths: AUA=M UGA=W AGR=S
#5: Nuclear Code of Candida cylindracea (see nature 341:164): CUG=S
#6: Nuclear Code of Ciliata: UAR = Q
#7: Nuclear Code of Euplotes: UGA=C
#8: Mitochondrial Code of Echinoderms: UGA=W AGR=S AAA=N
#9: Mitochondrial Code of Ascidaceae: UGA=W AGR=G AUA=M
#10: Mitochondrial Code of Platyhelminthes: UGA=W AGR=S UAA=Y AAA=N
#11: Nuclear Code of Blepharisma: UAG=Q
124
M.Sc. in Molecular Medicine Bioinformatics Course
Feb. 2005
APPENDIX III
Biochemically meaningful grouping of Amino Acids
(Taken from Willie Taylor's (1986) paper in J Theor Biol 119: 205-218.)
Conservative substitutions:
Marked with a : in clustalW
"strong groups"
STA
NEQK
NHQK
NDEQ
QHRK
MILV
MILF
HY
FYW
Marked with a . in clustalW "weak
groups"
CSA
ATV
SAG
STNK
STPA
SGND
SNDEQK
NDEQHK
NEQHRK
FVLIM
HFY
125
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement