Partek Express
®
™
Copyright
Copyright 1993-2010 by Partek Incorporated.
The software described in this document is furnished under a license agreement. The software may be used or copied only
under the terms of the license agreement. No part of this manual may be reproduced in any form without prior written
consent from Partek Incorporated. Licensee shall prevent and not permit any third parties, persons, or entities from copying,
reproducing, duplicating, examining, inspecting, studying, and/or reviewing the Software, Documentation, and/or
Information.
Partek, Partek Pro, Partek Express, Partek Analytical Spreadsheet, and Pattern Visualization System are registered
trademarks of Partek Incorporated. All other brand and product names mentioned are trademarks owned by their respective
companies or organizations.
Copyright © 2010 Partek Incorporated. St. Louis, MO.
Partek® Express™: Get Started with Partek® Express™
Partek Express
Partek® Express™ is a stand-alone software package that is produced by Partek
Incorporated and is distributed by Affymetrix Incorporated. This chapter will
briefly introduce each chapter in the Partek Express documentation.
Chapter 1: Get Started with Partek® Express™
Chapter 1 explains the chapters in Partek® Express™ Users Manual.
Chapter 2: Installation
Chapter 2 explains system recommendations, installation instructions, and library
file directory set-up for Partek® Express™.
Chapter 3: The Guided Workflow
Chapter 3 explains the intuitive guided workflow in Partek® Express™.
Chapter 4: Create a New Study
Chapter 4 describes how to create a new study in Partek® Express™.
Chapter 5: Edit Sample Information
Chapter 5 describes how to edit sample information with Partek® Express™.
Chapter 6: Data Import
Chapter 6 explains the algorithm, library files, and annotation files used when
importing .CEL or .CHP files.
Chapter 7: Quality Assessment
Chapter 7 describes quality assessment procedures in Partek® Express™, such as
generating quality control metrics, invoking quality control plots, and applying
quality control checks. Detailed information about the quality control metrics is also
explained in this chapter.
Chapter 8: Principle Components Analysis
Chapter 8 explains the principle components analysis plot, as well as its function, in
Partek® Express™.
Partek® Express™: Get Started with Partek Express
1
Chapter 9: Effect Sizes
Chapter 9 explains the use and function of effect size in experimental design.
Partek® Express™ contains plots to give a visual representation of effect size.
Chapter 10: Gene Significance
Chapter 10 explains how to view gene significance results in Partek® Express™.
Chapter 11: Power Analysis
Chapter 11 describes the benefits and how to use power analysis in Partek®
Express™.
Chapter 12: Report
Chapter 12 breaks down the workflow report in Partek® Express™.
Chapter 13: Pathway Analysis
Chapter 13 describes how to invoke external Pathway Analysis software from
Partek® Express™.
Chapter 14: Menu
Chapter 14 describes the menu shortcuts and functions in Partek® Express™.
Partek® Express™: Get Started with Partek Express
2
Partek® Express™: Installation
Partek® Express™ is a stand-alone software package. It can be installed on a Windows®
computer with or without other software packages like Affymetrix® GCOS™, Affymetrix®
Expression Console™, and/or Partek® Genomic Suite™.
System Recommendations










Operating System: Windows® XP SP3, Vista, or Windows 7; Linux; Mac
Operating System Language: English*
CPU: 2 GHz or higher
Memory (RAM): 1 GB or higher
Free disk space: 20 GB or higher
Monitor: 800x600 resolution or higher
Mouse: 2 button with scroll wheel
Internet connection for the purpose of downloading library and annotation files
Web browser with the Adobe® PDF plug-in for the purpose of viewing the User’s
Manual
*Partek Express can run on non-English versions of Windows, but there are
restrictions. For example, you should be able to finish the on-line tutorial
(http://www.partek.com/Tutorials), but CEL file names and sample attributions
need to be in English.
Installation Instructions
The Partek Express download page is located at:
http://www.partek.com/html/updates.html.

Download and run the installer. The web browser may pop up a Security
Warning dialog like the following:
Figure 2. 1: Download and Run the Installer from a Web Browser
Partek Express: Installation
3


Select Run to download and run the installer
After the downloading is finished, the Partek Express Setup Wizard will appear
(Figure 2. 2). Select Next >
Figure 2. 2: Install Partek Express

In the Choose Installation Directory step, you can select the install location
(Figure 2. 3). By default, Partek Express will be installed in C:\Program
Files\Partek Express. To change the default location, select Browse…. Select
Next >
Figure 2. 3: Choose Installation Directory
Partek Express: Installation
4



To import Affymetrix .CEL/.CHP files, Partek Express will need to download and
store library/annotation files, so in the Choose Microarray Library Directory step
(Figure 2. 4), you will need to select the location for downloading and storing
those files. The default folder is C:/Microarray Libraries, to change the default
folder, select Browse…. Make sure you have enough disk space and have read
and write permission to the directory you select.
Note: if the same library directory is also specified in Affymetrix® Expression
Console™ or Partek® Genomic Suite™, Partek Express will share the library and
annotation files with these software packages. Please refer to the corresponding
user’s guide on how to specify library directory in those software packages.
Select Next >
Figure 2. 4: Choose Microarray Library Directory

Review the installation options (Figure 2. 5). Select Next >
Partek Express: Installation
5
Figure 2. 5: Installation Options Review
Figure 2. 6: Installation Complete

Select the Finish button to complete the installation
Node-Locked License Setup
Upon successful installation, double-click the Partek Express icon on the desktop to launch
the software. If a license could not be obtained, a Partek Express Initialization Error dialog
will show up (Figure 2. 7).
Partek Express: Installation
6
Figure 2. 7: Setup License Error
To request a node-locked license:
 Select the Windows Start menu > All Programs > Partek > Partek Express >
lmtools > select the System Settings tab (Figure 2. 8)
Figure 2. 8: LMTOOLS System Settings











Select the Save HOSTID Info to a File button to save the system information to
a file, such as mySystemInfo.txt
Send an email to licensing@partek.com with the following information:
Your name
Your institution
Institution address
Phone number
Partek product name with version number: Partek Express x.xx.xxxx
Note: You can find this information from the Updates page:
http://www.partek.com/html/updates.html
Attach the mySystemInfo.txt saved by LMTOOLS
The licensing request should be from an official email address (not gmail,
hotmail, etc.)
After getting a license file from Partek, save the file to C:\Program Files\Partek
Express\license\license.dat
Partek Express: Installation
7


If you specified a different installation directory (Figure 2. 3), right click the
Partek Express icon on the desktop then choose Properties to find your
installation directory
Launch Partek Express again
Library Directory Setup
This section is only necessary if the installer has failed to create a Microarray Library
Directory as shown in Figure 2. 4. Otherwise, upon the first launch of the software, Partek
Express will prompt you to specify a library directory:
Figure 2. 9: Library File Directory

Use Windows Explorer to create a directory with read and write permissions (e.g.
C:\Microarray Libraries)
Figure 2. 10: Creating a directory with read and write permissions


Return to Partek Express and select OK (Figure 2. 9)
Specify the newly created directory (e.g., C:\Microarray Libraries) and select OK
Partek Express: Installation
8
Figure 2. 11: Specify the library directory
Note: if the same library directory is also specified in Affymetrix® Expression Console™
or Partek® Genomic Suite™, Partek Express will share the library and annotation files with
these software packages. Please refer to the corresponding user’s guides on how to specify
library directory in those software packages.
Partek Express: Installation
9
Partek® Express™: The Guided Workflow
Partek® Express™ Workflow
Partek® Express™ provides a guided workflow (Figure 3. 1), which is divided
into individual steps for study definition, data import, quality control, principle
data analysis, and effect sizes estimate. The workflow is designed to be intuitive
through the use of dialogs. Simply follow the instructions in the current dialog to
get to the next step. You can choose to stop at any step, save your study, and exit.
When the saved study is loaded back into Partek Express, you can resume from
your last step.
Figure 3. 1: Partek Express Work Flow Diagram
Partek Express: The Guided Workflow
10
The example study shown in Figure 3. 2 has reached the Effect Sizes step.
The main screen parts are:
1.
2.
3.
4.
The current study name
The Menu bar
The Tools bar
The Result and Report tabs. Each step produces its own result tab. You
can select a tab to view the result from a previous step. The Report tab
shows the information of the current study
5. The results of current step
6. The Tell Me More… button; selecting it will bring up the corresponding
documentation chapter for the current step
7. Description of the next step
8. Optional workflow steps
9. The Back button; selecting it will delete the current step’s result and go
back one step. The difference between the Back button and the Results tab
is that the Back button will destroy the current step’s result and can only
go back one step at a time; however, the Results tab can be used to view
results (including the Report) from other steps.
10. The Next button; selecting it will proceed to the next step and eventually
finish the workflow
11. The Progress Bar
12. Finished steps are colored in light red
13. The current step is highlighted in Bold
14. Future steps are colored in grey
Figure 3. 2: An Example Study at the Effect Sizes Step
Partek Express: The Guided Workflow
11
Partek® Express™: Creating a New Study
Introduction
This chapter describes how to create a new study in Partek® Express™.
Specify the Study File
A new study can be created by doing any of the following:



Selecting File > New Study
Selecting the New Study icon “ ”
Selecting the Create Study button at the bottom right of the main
screen
A file browser will appear to let you specify a location and file name for the new
study (Figure 4. 1). Give the File name, then select the Save button.
Figure 4. 1: Specify the Study File
Select Samples
After the study file is created, a sample selector window will appear so that you
can navigate through folders to select samples for the study (Figure 4. 2). The
main screen parts are:
Partek Express: Creating a New Study
12
1. The current directory; you can change the directory by typing in or pasting
a new address
2. The Go Back icon; selecting this will take you to the previous directory
3. The Go Up icon; selecting this will take you up one level in the directory
4. The Browse… button; selecting this will bring up the system directory
browser
5. The main directory browser; from here, select the directory where the
sample .CEL/.CHP files are located
6. .CEL/.CHP files in the current directory will show in each row. If there are
associated .ARR files, the sample attributes will also be shown here.
7. Selecting the column header’s white box will Select/Deselect all samples
8. Selecting the white box will Select/Deselect individual or highlighted
samples
9. <Click>, <Control+Click>, or <Shift+Click> to highlight samples
10. Selecting the corner square will de-highlight all samples
11. The Add Samples button will add selected samples and close the window
Figure 4. 2: Viewing the sample selector
Only files from the same folder can be added at the same time. After adding
samples from a folder, select the Add More Samples icon “ ” on the Sample
Information Editor tool bar to add samples from a different folder.
Partek Express: Creating a New Study
13
Partek® Express™: Edit Sample Information
Partek® Express™ Sample Editor
This chapter describes how to edit sample information using Partek® Express™.
The sample editor shows sample attributes in a tabular format (Figure 5. 1). Each
row corresponds to one sample. Each column corresponds to one sample
attribute. If the ARR files are located for intensity or summarization files, the
sample attribute information stored in them will be automatically extracted and
filled into the sample editing table. You can then use the sample editor to add
more attributes, delete unnecessary attributes, rearrange samples or sample
attribute order. The rest of this chapter will provide a detailed guide on how this
can be done.
Figure 5. 1: Viewing the Sample Information editor
Getting Started
Column Types
The three types of columns in the Partek Express sample editor are listed below.
Partek Express: Edit Sample Information
14
Type
text
numeric
Description
variable length string
double precision floating point (8 bytes) (-1.7E308 to
1.7E308)
categorical Variable length nominal
Table 5. 1: Reviewing the column types in Partek Express
Different column types are displayed in different colors. Text columns are shown
in gray; numeric columns are shown in blue; categorical columns with random
effect are shown in red; and categorical columns with fixed effect are shown in
black (see below for more details on random and fixed effects).
Random Effect vs. Fixed Effect
In a study, if an effect has all possible categories of interest, then it's usually a
fixed effect. Effects like gender, disease state, tissue type, and treatment are
usually fixed effects. Effects like subject, batch, and operator are usually random
effects. In addition, if an effect has a subset of all possible categories of interest,
then it's usually a random effect.
For example, if a study has 10 patients. They represent only a random sample of
the global subjects of which an inference is being made. This patient effect is a
random effect.
Here is another way to tell if a factor is random or fixed: imagine repeating the
study. Would the same categories of each effect be used again?
Gender - Yes, the same genders would be used again - a fixed effect
Subject - No, the samples would be taken from other subjects - a random effect
You can specify an effect to be random or fixed by right-clicking on the column
header then selecting Properties.
Note: Specifying an effect as a fixed or random effect will affect future statistical
results. Please consult a statistician if you are not sure whether an effect is fixed
or random.
View Sample Distribution for a Categorical Column
To view the sample attribute distribution for a column, single click on the header
of a categorical column. The pane at the bottom of the sample editor will show a
bar chart of sample attribute values (Figure 5. 2). Note: the sample distribution
bar chart is only available for categorical columns, when clicking on the text or a
numeric column, the bar chart will be blank. The main screen parts are:
1. Column header; selecting it will show the sample distribution
Partek Express: Edit Sample Information
15
2.
3.
4.
5.
The category name (column header as in 1.)
The total number of categories in a column
The total number of samples
The individual category and the frequency in parentheses
Figure 5. 2: View Sample Distribution for a Categorical Column
Read-only Columns
In order to maintain data integrity for your study, some columns are read only.
For example, in Figure 5. 2, the column CEL File Name and Scan Date are readonly columns. If you attempt to modify the values in these columns, a dialog will
appear to notify you that the change cannot be made (Figure 5. 3).
Figure 5. 3: Viewing the Read-only Column warning
Attribute Editing Limitations
The sample attributes need to be in English. It is recommended that you limit the
maximum number of characters to 32 in a cell. Long or non-English attribute
names will decrease the readability of visualizations.
Partek Express: Edit Sample Information
16
Sample Attribute (column) Operation
To Select a Column
To select a column, click on the column header. The column will be highlighted.
To Add Column(s)
There are four ways to add columns to the table.



End of the Table: Select the Add Attribute… button and choose a
predefined attribute from the popup menu. The selected attribute will be
added to the end of the table
End of the Table: Select Add Attribute… > Other…. This will bring up
the add column dialog (Figure 5. 4). Specify the desired column label and
choose a column type using the radio buttons and select OK. The column
will be added to the end of the table
Before the Current Column: Right click on a column header and select
Insert Attribute. This will bring up the Add Column dialog (Figure 5. 4).
Specify the desired column label and choose a column type using the radio
buttons and select OK. The column will be inserted before the column
where the menu was invoked from
Figure 5. 4: Configuring the Add an Attribute dialog

After the Current Column: Right click on a text column and select Split
Column. This will bring up the Text to Column Splitter dialog (Figure 5.
5). Two options are provided to split the column – by delimiters or by
fixed width. After specifying splitting parameters, select the Update
button. A preview of the result will be shown at the bottom of the dialog.
Specify column labels and properties using the entry box and dropdown
menu. If column labels are not specified, default labels will be assigned
automatically. To skip a column, choose Skip from the drop down menu
under that column. Select OK when done. The resulting column(s) will
be inserted after the column being split
Partek Express: Edit Sample Information
17
Figure 5. 5: Configuring the Sample Information Creation dialog
To Delete a Column
To delete a column, right click on the column header and select Delete Attribute.
To Edit Column Properties
Right clicking on a column header and selecting Properties will bring up the
column properties dialog (Figure 5. 6). Specify the desired attribute label and
attribute type and select OK. Note that the sample information editor won’t allow
two columns to have the same column label. If the new column label specified
already exists in the table, the OK button will gray out.
Figure 5. 6: Column properties dialog
To Reposition a Column
Repositioning a column can be done by drag and drop. Simply select the header
of the column to be repositioned and start dragging, when the column is in the
desired spot, release the mouse button.
Partek Express: Edit Sample Information
18
To Sort Samples Based on One Column
To sort samples based on one column, right click on a column header and select
Sort Ascending or Sort Descending.
To Sort Samples Based on Multiple Columns
To sort samples based on multiple columns, sort samples in the reverse order of
the columns desired to be sorted. For example, if you want to sort samples first
based on scan date then based on type, sort them based on type first and then scan
date.
Sample (row) Operation
To Select Row(s)
To select rows, click on the row header. Multiple rows can be selected using the
Control and /or Shift key. Row selection can also be done by clicking on a bar in
the sample bar chart. All rows that fall in the category will be selected.
To Delete Row(s)
To delete row, select the rows to be deleted. After selecting the rows, right click
on the row header and select Delete Sample(s).
Edit Sample Attribute/Table Cell Operation
To Select Cells
To select a cell, click on the cell. Multiple cells can be selected by holding the
mouse, dragging, and using the Control and /or Shift key. Note that only cells
within a single column can be selected at the same time.
To Edit Multiple Cells
There are four ways to edit multiple cells.




Select multiple cells and start typing. This will change the value of all the
selected cells simultaneously
Select one or more (consecutive) cells, select the bottom edge of the
selected area and drag down
Use copy, cut, and paste (see below)
Right click on a bar in the sample bar chart and select Change Value. Edit
the value in the Edit Sample Attribute dialog (Figure 5. 7) and select OK.
The value of the category will be changed to the new value
Partek Express: Edit Sample Information
19
Figure 5. 7: Edit Sample Attribute Dialog
To Copy, Cut, and Paste
 To copy, select the cells and select Copy from right mouse menu, or press
Control+c.
 To cut, select the cells and select Cut from right mouse menu, or press
Control+x.
 To paste, select the cells and select Paste from right mouse menu, or press
Control+v. Note that if multiple cells are selected, paste can only be
performed if they are consecutive.
To Drag and Fill Data Automatically in Multiple Cells
Instead of entering sample information manually, you can use the auto fill feature
provided by the Partek Express sample editor. The auto fill feature can identify
certain data patterns and automatically fill in multiple cells. To start, select the
cell(s) that contain the data that you want to fill into adjacent cells.
Note: these cells have to be in the same column and they must be in consecutive
order.

Place the mouse at the bottom edge of the selected cells. If auto fill is
possible, the mouse cursor will turn into a down arrow. Hold the mouse
down and drag down across the cells you want to fill.
The following patterns are supported for auto fill:



Arithmetic series (integers, floating point numbers or integers with same
prefix or suffix)
Days of the week
Months of the year
Undo and Redo Editing
You can undo and redo up to 100 actions per session in the Partek Express sample
editor.
To Undo
To undo, select the Undo button or press Control+z.
To Redo
To redo, select the Redo button or press Control+y.
Partek Express: Edit Sample Information
20
Note that certain actions cannot be undone, such as saving a study. If you can’t
undo an action, the Undo button will be grayed out.
Formatting the Table
To Best Fit Column(s)
There are two ways to best fit column width(s).
1. Select the Fit Columns button. This will best fit all columns in the table
2. Double click on the right edge of a column header. This will best fit the
column left to it
To Manually Specify a Column Width
Right clicking on a column header and selecting Column Width will bring up the
Column Width dialog (Figure 5. 8). Specify a positive integer pixel value as the
column width and select OK.
Figure 5. 8: Configuring the Column Width dialog
Partek Express: Edit Sample Information
21
Partek® Express™: Data Import
Introduction
For most studies, Partek® Express™ will automatically download library/annotation
files and import the data. This chapter provides more details on Partek Express’
data import algorithm and on situations when automatic downloading cannot be
achieved.
Import Algorithm
Import .CEL Files
When importing .CEL files, Partek Express uses RMA, which includes background
correction, quantile normalization, and median polish summarization. For more
information about RMA, refer to the following:



Bolstad, B.M., Irizarry R. A., Astrand, M., & Speed, T.P. (2003), A
Comparison of Normalization Methods for High Density Oligonucleotide
Array Data Based on Bias and Variance. Bioinformatics 19(2):185-193.
Irizarry, R.A., Bolstad, B.M., Collin, F., Cope, L.M., Hobbs, B., Terence, P.,
&Speed, T.P. (2003), Summaries of Affymetrix GeneChip probe level data
Nucleic Acids Research 31(4):e15.
Irizarry, R.A., Hobbs, B., Collin, F., Beazer-Barclay, Y.D., Antonellis, K.J.,
Scherf, U., & Speed, T.P. (2002) Exploration, Normalization, and
Summaries of High Density Oligonucleotide Array Probe Level Data.
Biostatistics 4 (2):249-64.
Note: Median polish summarization may produce the same value for each sample
when there are very few samples. For this reason, when there are <= 4 samples,
Partek Express will use mean summarization.
Import .CHP Files
Intensities in .CHP files are already normalized and summarized. Partek Express
directly reads the intensities without further processing the data. For more
information, please refer to the Affymetrix software that was used to create the
.CHP files.
Import Agilent Files
When importing Agilent two color files, Partek Express uses mean as the
summarization method and does loess normalization to get the log ratio of two color
data. When importing Agilent one color files, Partek Express does log base 2
transformation.
Partek Express: Data Import
22
Import Illumina Files
Data from Illumina’s BeadStudio software package can be exported in a custom
Partek report file (.ppj file) for seamless importation of your Illumina data into
Partek Express. When importing Illumina files, browse to select the .ppj file and
select open.
Import NimbleGen Files
When importing NimbleGen files, Partek Express also uses RMA, which includes
background correction, quantile normalization, and median polish summarization.
Library and Annotation Files
Affymetrix Library and Annotation Files
When importing .CEL files, Partek Express will automatically download the
corresponding library files (Figure 6. 1).
Figure 6. 1: Downloading the Library File during data import
If the library files are stored on a local drive, select the abort button
to manually
specify the file location. As shown in Figure 6. 2, the Specify File Location dialog
will appear.
Figure 6. 2: Specifying the File Location
Partek Express: Data Import
23
There are several options available to specify the library file:
1. Select the Browse… button to select the library file
2. Select the Search and Copy… button to let Partek Express find the correct
library file by searching a directory and all its sub-directories
3. Select the Download button to continue the downloading process
After manually specifying the file location, select OK.
Agilent Annotation Files
When importing Agilent files, Partek Express will automatically generate an
annotation file, which is extracted from the Agilent data files.
Illumina Annotation Files
Data from Illumina’s BeadStudio software package can be exported into two files.
One is a custom Partek report file (.ppj file) and the other is an annotation file
(.annotation.txt). Place the annotation file into the directory that holds the .ppj file,
so that Partek Express will be able to find the annotation file during import.
Figure 6. 3: Viewing the directory that holds the Illumina .ppj file and Annotation
files
NimbleGen Annotation Files
The NimbleGen importer will invoke a window to specify a NimbleGen annotation
file (.ngd file). After importing, this annotation file will be converted into a Partek
recognized format (.annotation.txt).
Partek Express: Data Import
24
Figure 6. 4: Specifying the NimbleGen Annotation file
Partek Express: Data Import
25
Partek® Express™: Quality Assessment
Introduction
The quality assessment procedures in Partek® Express™ are explained in this
chapter. The procedures include generating QC metrics, invoking QC plots, and
applying QC checking. The detailed algorithm of QC metrics also will be
explained.
Quality Assessment Procedures
Generate QC Metrics and Graphs
QC metrics are generated from importing CEL files or CHP files; however, the
CEL file importer invokes the QC metrics calculation during importing, but the
CHP file importer just extracts QC metrics from the CHP files.
Figure 7. 1: Generating QC metrics during importing
Upon finishing the import, the quality assessment results will be shown in the QC
Metrics tab. The QC Metrics tab provides quality control information from the
control and experimental probes on the Affymetrix chips to provide confidence in
the quality of the microarray data or to identify samples that do not pass the QC
metrics. The QC metrics result can be viewed either in line graph format or in
Partek Express: Quality Assessment
26
spreadsheet format by selecting the corresponding radio buttons at the top of QC
Metrics tab.
The main screen parts are explained below and pictured in Figure 7. 2.
1. The QC Limits button; selecting this will perform QC limits checking
2. The QC Graphs radio button; selecting this will allow you to view QC
graphs including line graph, sample box plot and sample MA plot by
selecting the corresponding tab
3. The QC Metrics radio button; selecting this will allow you to view the QC
metrics table
4. QC metrics grouping categories, which includes Hybridization, Labeling,
3’/5’ and Other. By default, the Hybridization tab is selected; to view the
other categories, select the corresponding tab
5. The Select All and Clear All buttons; selecting will either check or
uncheck QC metrics
6. The Hybridization and Labeling controls’ metrics are listed in the
expected order from high to low. In this example, AFFX-rs-P1-cre-avg
should be higher than AFFX-r2-Ec-bioD-avg, which should be higher than
AFFX-r2-Ec-bioC-avg, which should be higher than AFFX-r2-Ec-bioBavg. For the controls on 3’/5’ and Other tab, their QC metrics will not be
listed in any particular order.
7. Samples are on X axis. In this example, there are 25 samples. Their order
is the same as on the Sample Information tab.
8. For Hybridization and Labeling controls, their metrics are in Log 2
intensity scale. For 3’/5’ controls, their metrics are ratios*. For metrics
under the Other tab, the Y axis may be Intensity or Value.
9. The sample selection lines; <click> or <control-click> to select one or
more samples. The sample selection is synchronized with the Sample
Information tab and sample selection boxes in the box plot
10. The mouse-over view in the Line Graph; dragging the mouse over points
in the graph will show the .CEL file name of a sample
11. The Log Expression Signal radio button; selecting this will allow you to
view a box plot on the log expression signal. This box plot is generated
from the probe set signal values that have been normalized and
summarized
12. The Log Probe Cell Intensity radio button; selecting this will allow you to
view a box plot on the log probe cell intensity. This box plot is generated
from the probe cell intensity values prior to normalization and
summarization. This box plot is only available for CEL file importer but
not CHP file importer, because the data from CHP file are already
normalized and summarized
13. The Relative Log Expression Signal radio button; selecting this will allow
you to view a box plot on the relative log expression signal. This box plot
is a summarization of the RLE values which are obtained by calculating
Partek Express: Quality Assessment
27
the log base 2 difference between the probeset signal estimate and the
median of probeset signal estimates across all of the arrays
14. The sample selection boxes; <click> or <control-click> to select one or
more samples. The sample selection is synchronized with the Sample
Information tab and sample selection lines in the line graph
15. The mouse-over view in the Box Plot; dragging the mouse over points in
the graph will show the five-number summaries of each sample. The fivenumber summaries are the smallest observation, 25th quantile (Q1),
median, 75th quantile (Q3), and the largest observation
16. Sample MA Plot; selecting sample 1 and sample 2 will invoke a sample
MA plot, which compares the intensities between these two samples
(rows). Each point represents one column in the spreadsheet. For the two
selected samples, the average is on the X axis, and the difference on the Y
axis. The points are expected to be centered along Y=0 for all values of X
* When importing .CEL files, Partek Express transforms the intensity with log base
2. Suppose the log 2 intensity of 3’ control is A3 and the log 2 intensity of 5’
control is A5, Partek Express first does base 2 anti-log A3 and A5 to a3 and a5 then
calculates the ratio as a3/a5. When importing .CHP files, Partek Express directly
reads the QC metrics from the header of .CHP files. Please note some .CHP files
may have calculated 3’/5’ ratio from the logged data. For example, the intensity of
normalized 3’ control is A3 and 5’ control is A5. They have already been logged.
Some .CHP files may directly give the ratio as A3/A5.
Partek Express: Quality Assessment
28
Figure 7. 2: Viewing the QC metrics visualizations
Partek Express: Quality Assessment
29
Apply QC Checking
Introduction
When the QC metrics data is generated, the QC data is automatically tested against
several predefined criteria. If any of the QC data fail any of the criteria, the failing
QC metrics will be highlighted in the QC metrics spreadsheet at which point a
determination must be made to either continue the analysis, omit the samples that
failed the QC criteria, or to rerun the failed samples to generate new data that passes
the QC criteria.
The main screen parts of the QC Metrics table are explained below and pictured in
Figure 7. 3.
1. The QC Metrics radio button; selecting this will allow you to view the QC
Metrics table
2. Metrics that didn’t pass the QC Limits checking will be highlighted
3. The sample selection is synchronized with the Sample Information tab
Figure 7. 3: Showing QC metrics table in Partek Express
Configuring the QC Check Dialog
Open the Apply QC Checking dialog by selecting the QC Limits button.
Pre-defined Thresholds
There are eight pre-defined thresholds within the Apply QC Checking dialog. These
pre-defined thresholds are grouped according to their probe set properties, which
Partek Express: Quality Assessment
30
include Pre-defined Hybridization, Pre-defined Labeling, Pre-defined 3’/5’, and
Custom.
Figure 7. 4: Configuring the Applying QC Checking dialog
User Specified Thresholds
Selecting the Add Custom Criteria button at the bottom of the Apply QC Checking
dialog will add user specified thresholds. The user specified thresholds will be
shown on the Custom tab.
Partek Express: Quality Assessment
31
Figure 7. 5: Adding user specified thresholds
The options in the Custom tab include the following:
1. Comparison type:
1. Compare metric 1 to metric 2
2. Compare metric 1 to threshold value
3. Compare metric 1 to average of metric 1 across arrays by
standard deviation. The in boundary range for metric 1 is
(mean - range multiplier * standard deviation, mean + range
multiplier * standard deviation)
2. Comparison operator (if it is):
1. Less than
2. Less than or equal to
3. Greater than
4. Greater than or equal to
5. Equals
6. Not equals
Select the metric for Metric 2 or input the threshold value or the range multiplier.
The option to be used here is decided by the comparison type selected above.
will delete the user specified threshold. Pre-defined thresholds cannot be deleted.
Partek Express: Quality Assessment
32
Save Thresholds
All the user specified thresholds combined with the pre-defined thresholds can be
saved as a criteria file by selecting the Save Criteria File button. The entry box for
the criteria file name is located on the top of the dialog. The Default thresholds are
created by Partek Express and cannot be overwritten. The thresholds that were
applied last time will be automatically saved as a criterion file named Last_used,
which is also not able to be modified.
Delete Thresholds
Delete Criteria File can delete any criteria file you save. The criteria file Default
and Last_used are created by Partek Express and cannot be deleted.
Load Thresholds
The previously saved thresholds can be loaded into Partek Express by selecting the
criteria file name from the dropdown list box, which is also used as an entry box for
saving thresholds. All the criteria files that have been saved will be shown in the
dropdown list box. Last_used criteria will always be loaded by default.
Running the QC checking
Apply will check if the QC metrics results are within the boundaries defined by
those thresholds. Any result that is out of the boundary will be colored on the QC
metrics table (Figure 2).
Cancel will close the QC Check dialog without doing any checking.
Quality Assessment Metrics
Hybridization Metrics
Four exogeneous (E. coli derived) pre-labeled molecules are spiked into the
hybridization cocktail before hybridization, but after sample labeling. The spikes
test to ensure that hybridization correctly occurred on the array. These molecules
are spiked in at increasing concentrations: BioB < BioC < BioD < Cre. A graph of
the values is automatically created and displayed in the Hybridization tab within the
QC Metrics section. Make sure to ensure that each of the spikes has the correct
relative abundance in the samples as displayed in Figure 7. 6.
Partek Express: Quality Assessment
33
Figure 7. 6: Viewing the line graph of Hybridization Spikes. In each sample, the
four hyb spikes have increasing concentrations from BioB as the lowest to Cre as
the highest.
Labeling Metrics
Up to five unlabeled polyA control spikes are available to spike into the samples to
control for the sample labeling reaction. The spikes are inserted into the sample
prior to labeling and their resulting detection is dependent on the labeling reaction
that labels the biological sample. They are derived from B. subtilis, and are
typically spiked in at increasing concentrations of Lys < Phe < Thr < Dap. Make
sure to confirm that these spikes were used in the samples and confirm that the
correct concentrations were used. This Down’s syndrome experiment was run
before these spikes were commercialized and they show a different intensity
pattern. The graph of these spikes (Figure 7. 7) is displayed in Partek Express in
the QC Metrics section under the Labeling tab. Partek Express only extracts the
Dap, Phe, and Lys spikes.
Partek Express: Quality Assessment
34
Figure 7. 7: Labeling spikes of Dap, Phe, and Lys. This experiment shows DAP <
LYS < PHE
3’ / 5’ Ratio Metrics
Partek Express will calculate and plot the 3’ / 5’ ratio of GAPDH. It is displayed
under the QC Metrics section in the 3’/5’ tab. GAPDH has separate probe sets at
the 3’ and 5’ end of the gene. In high-quality samples, reverse transcriptase should
process from the 3’ through towards the 5’ end. The 3’ / 5’ ratio compares the
abundance of the signal at the 3’ end over the abundance at the 5’ end. A ratio of 3
or less is considered acceptable.
Figure 7. 8: Viewing the 3’ / 5’ Ratio for Human GAPDH across all samples in the
experiment; all values are less than
Other Quality Control Metrics
Three additional quality control metrics are displayed in the Other tab within the
QC Metrics section: PM Mean, Mad_Residual Mean, and RLE Mean. For more
information on these values consult the Quick Reference Card from Affymetrix
entitled, “QC Metrics for Exon and Gene Design Expression Arrays”.
PM Mean
PM Mean is the mean raw probe intensity from a sample. It is a measure of how
bright or dim an array is. Samples within an experiment should have roughly
similar PM Means. There are not any default criteria regarding PM Mean. Samples
should be scanned for “outlier” values as determined by the user through visual
inspection.
MAD Residual Mean
MAD Residual Mean is a bit of a complex measurement. It is the mean across all
probe sets of the Median Absolute Deviation (MAD) of the residuals between the
predicted and actual probe values. During signal estimation, a model is created
Partek Express: Quality Assessment
35
based on the trends for each probe across the whole experiment. This model can be
used to “predict” how a probe will respond. The residual is the difference between
the predicted and actual values. When examined at a sample level (across all probe
sets) the MAD Residual Mean value is a measure of how well the individual sample
fits the model for the experiment. Samples with higher values fit less well.
RLE Mean
RLE Mean is the mean of the absolute relative log expression (RLE) across all
probe sets on each array. Consult Chapter 6 of the Partek Express User Manual for
more information on its calculation. RLE Mean compares the signal each probe set
(gene) in a sample compared to the median gene-level signal value across the
experiment (all samples). If a sample has a high RLE Mean that implies that that
sample isn’t quite as similar to all of the samples. High RLE Mean values will flag
outliers. Affymetrix states that RLE Mean values across a diverse tissue panel
range from 0.27 to 0.61, while values across an experiment of only technical
replicates range of 0.1 to 0.23. Remember that if you have a collection of diverse
samples in the experiment the RLE Mean values will be higher than if the samples
were very similar.
References
Affymetrix White Paper: Quality Assessment of Exon and Gene Arrays. Revision
1.1, published April 06, 2007.
http://www.affymetrix.com/support/technical/whitepapers/exon_gene_arra
ys_qa_whitepaper.pdf.
Affymetrix Quick Reference Card, “QC Metrics for Exon and Gene Design
Expression Arrays”
http://www.affymetrix.com/Auth/support/downloads/quick_reference_car
ds/qc_metrics_exon_gene_qrc.pdf
Partek Express: Quality Assessment
36
Partek® Express™: Principal Components Analysis
Introduction
This section describes the principal components analysis (PCA) plot and its
function within Partek® Express™.
Invoking the PCA Plot

Select the Next button after completing the Data Import & QC Check
step. The PCA plot will be brought up and shown in another tab named
PCA (Figure 8. 1 )
Figure 8. 1: Viewing the PCA Plot
The main screen parts are:
1. The Shape by elements; these can be shaped as spheres, cubes, and/or
pyramids, etc. They represent individual samples. For example, in this
study there are 25 samples, represented by the 25 elements in the PCA
plot. The grouping of, and the relative distance between, those objects
visually reveal the relation of those samples
2. The X, Y, and Z axes are PC #1, PC #2, and PC #3, respectively. The
percentages shown in the axis labels display the amount of data variation
accounted for by the respective PC. Please refer to the Details About PCA
below for more information. Since the PC’s are ordered from greatest to
Partek Express: Principle Components Analysis
37
3.
4.
5.
6.
smallest amount of explained variation, PC #1 will be greater or equal to
PC #2, which is greater than PC #3
The percentage shown in the title is the sum of the percentage on those
axes. It cannot be greater than 100%
The Legends; these describe the use of symbols, colors, etc., used in the
plot
The Tools bar; selecting options within the Tools bar will change the PCA
plot style. See Configuring the PCA Plot below for more details
The Mode bar; options located in the Mode bar will control the PCA plot.
See Using the Viewer Modes in the PCA Plot below for more details
Viewing the PCA Plot
PCA is an excellent method for visualizing high dimensional data by reducing the
variation across all of the many thousands of probes being interrogated on the
chip into a two or three dimensional representation. In a PCA plot, each point
represents a sample (microarray) and corresponds to a row in the Sample
Information tab. The positions of the dots are relative to each other. Those dots
that are closer to each other represent samples in which the transcriptome
measurements over the whole chip are similar. Those dots that are further away
from each other represent samples in which the transcriptome measurements over
the whole chip are more dissimilar. Samples that have similar overall gene
expression levels will group together into clusters. Identifying separate clusters in
a PCA provides valuable information, such as, which of the phenotypic variables
are driving the major sources of variation within the experiment.
One example would be if an experiment only had one factor, treated and
untreated. Assuming that all the samples in the data set are the same except for
this one factor, it is possible to quickly identify if the treatment had a significant
effect on the overall gene expression. If all of the samples clustered together into
one group with the two colors mixed equally among the cluster, then there is no
distinctive difference between the gene expressions over the samples based upon
treatment. However, if the samples cluster into two distinct groups, one cluster
containing only treated samples and the other cluster containing only untreated
samples, then there is a difference in the gene expression profiles between the
treated and untreated samples.
Details About PCA
PCA is an exploratory technique that is used to describe the structure of high
dimensional data by reducing its dimensionality (Jolliffe, 1986). It is a linear
transformation that converts n original variables into n new variables (“PC’s”),
which have three important properties:


The new variables (PC’s) are ordered by the amount of variance explained
The new variables (PC’s) are uncorrelated
Partek Express: Principle Components Analysis
38

The new variables (PC’s) explain all variation in the data
PCA is a Principal Axis Rotation of the original variables that preserves the
variation in the data. Therefore, the total variance of the original variables is
equal to the total variance of the principal components. The eigenvectors and
eigenvalues define the rotation and variation and are described as follows:


The eigenvalues are the variances of the principal components
The eigenvectors are the direction cosines of the new axes (PCs) relative
to the old (original variables), thus they define the rotations of the original
axes
The method of PCA dates back to Harold Hotelling’s 1933 paper “Analysis of a
complex of statistical variables into principal components”.
Configuring the PCA Plot
Style
Color, size, shape and connecting lines can be configured on the toolbar on the
top of the PCA viewer.
Color By
By default, the color of the points is determined by the first categorical attribute in
the sample information spreadsheet. If the sample information spreadsheet does
not have a categorical variable, then the points will be colored by the first
attribute. By choosing from the Color drop down menu (Figure 8. 2), the points
can be colored by any column or by all the same color. If a categorical attribute is
selected, each category will be assigned a distinct color. If a numeric attribute is
selected, the color assigned to points will be based on a continuous color palette.
Figure 8. 2: Viewing the options in the Color By Configuration Menu
Size By
By default, all points in the plot are of the same size. Sizes of points can be
configured by choosing from the Size drop down menu (Figure 8. 3). If size is set
to “Auto”, the size of the points will be based on the number of points in the plot.
The size of a point can also be configured to correspond to the value in a specific
Partek Express: Principle Components Analysis
39
column in the sample information spreadsheet. When sizing by a categorical
attribute, each category will be assigned a distinct size. When sizing by a numeric
column, the size will be based on the order of points, and the legend lists the
minimum, middle, and maximum values.
Figure 8. 3: Viewing the options in the Size By Configuration Menu
Shape By
By default, all points in the plot will be the same (point). Shapes of points can be
configured by choosing from the Shape drop down menu (Figure 8. 4). When an
attribute is selected, the shape of a point is determined by the attribute value of the
corresponding sample.
Figure 8. 4: Viewing the options in the Shape By Configuration Menu
There are five possible shapes to use. They are sphere, tetrahedron, cube,
octahedron, and icosahedron. If the specified attribute is categorical and has 4 or
fewer categories, each category will be assigned a distinct shape. If more than 5
categories are present in a categorical column, it will result in using the same
shape for more than one category. If the specified attribute is numeric, the range
of the values in that column is divided into 5 groups of equal range. The points
will be shaped according to the group into which they fall.
Connect By
A line can be drawn among points that have the same value for the specified
attribute. This is useful when looking at samples from the same subject (Figure 8.
5).
Partek Express: Principle Components Analysis
40
Figure 8. 5: Viewing the Connect By Configuration Menu
Dimension
When the dimension is set as 3D, all X, Y, and Z axes will contain one PC from
the dataset. If the dimension is set to be 2D, only X and Y will be drawn and the
viewer perspective will be turned off. The default setting for PCA dimension is
3D.
Using the Viewer Modes in the PCA Plot
This section will explain how to use the Mode Tool Bar in the viewer (Figure 8.
6). The Mode Toolbar is the vertical toolbar at the left side of the viewer.
Figure 8. 6: Viewing the Mode Toolbar
Changing Modes
Most of the icons in the vertical mode toolbar have multiple options that can be
accessed by clicking and briefly holding down the left mouse button on the mode
icon, upon doing that, a mode option menu will pop-up to the right. To select an
option from the menu, drag the mouse cursor over to the desired mode option and
then release the mouse button.
Selection Mode
To invoke selection mode, click on the Selection Mode icon on the Mode Toolbar.
When in selection mode, the following operations can be performed:



Select an individual item – click the left mouse button
Create a bounding box - hold down the left mouse button and drag the
mouse creates a box to select items inside the box
Add the item under the mouse cursor (or in the box) to the list of
selections - <Ctrl> + left click
Partek Express: Principle Components Analysis
41
Figure 8. 7: Viewing the Selection Mode icon on the Mode Toolbar
Zoom Mode
To invoke Zoom Mode, click on the Zoom Mode icon on the Mode Toolbar.
When in Zoom Mode, left click to incrementally zoom in, <Ctrl>-click to zoom
out.
Figure 8. 8: Viewing the Zoom Mode icon on the Mode Toolbar
Pan Mode
This icon is enabled only when the data is zoomed in on. Hold down the left
mouse button while dragging the mouse to interactively move the data (pan).
Figure 8. 9: Viewing the Pan Mode icon on the Mode Toolbar
Rotate Mode
There are two rotation modes: Manual Rotation Mode (one circle) and
Continuous Rotation Mode (two circles, one is on top of another).
Partek Express: Principle Components Analysis
42
Figure 8. 10: Viewing the Rotation Mode icon on the Mode Toolbar
Figure 8. 11: Viewing the Manual Rotation Mode and the Continuous Rotation
Mode
Manual Rotation Mode
The Manual Rotation Mode has the same functionality of the middle-mouse
button. Hold down the left mouse button while dragging the mouse to
interactively rotate the view.
Continuous Rotation Mode
Click the left mouse button to start and stop rotation. Selecting any other mode
also stops continuous rotation.
Common in All Modes
Mouse over
Place the mouse cursor over a data item (without clicking) to see information
about the item.
Reset
The <Home> key resets Rotation, Zoom and Pan back to their default values. The
same can be achieved by clicking the Reset icon on the mode toolbar (Figure 8.
12).
Figure 8. 12: Viewing the Reset Mode icon on the Mode Toolbar
References
Hotelling, H. “Analysis of a complex of statistical variables into principal
components”. J. Educ. Psych 1933, 26: 417-441.
Jolliffe, I.T. Principal Component Analysis, Springer-Verlag, New York, 1986.
Partek Express: Principle Components Analysis
43
Partek® Express™: Effect Sizes
Introduction
Effect size provides information on the importance of each experimental factor to
the transcriptome. Partek® Express™ uses Analysis of Variance (ANOVA) to test
for the difference in means of a response variable between different groups.
Configuring the ANOVA Dialog
Before estimating effect sizes, sample information has to be created first. Please
refer to the Edit Sample Information chapter to get more information about how to
create sample information.
The Partek Express workflow will introduce you to the Effect Sizes step
automatically by continually selecting Next > on the main window. The Effect Sizes
step is after this sequence: Start > Study Definition > Data Import & QC Check
> PCA.
Selecting Factor(s) of Interest
Main Effects
All the factors will be shown on the left panel of Estimate Effect Sizes dialog
(Figure 9. 1), however only categorical (fixed) factors or numeric factors can be
selected as the effect of interest. At most, two factors can be selected as the effect of
interest. Drag and drop the factor from the left panel to the top or the middle panel
on the right to set the effects of interest. The second effect of interest is optional.
Partek Express will automatically detect whether a factor is estimable or not based
on the current ANOVA model configuration. Non-Estimable will be inserted before
the factor name once the factor is detected as non-estimable (Figure 9. 2).
Partek Express: Effect Sizes
44
Figure 9. 1: Estimate Effect Sizes, Dialog 1
Figure 9. 2: Estimate Effect Sizes, Non-Estimable effects
Interactions
An interaction of the main effects will be automatically added to the ANOVA
model if more than one main effect is selected. An interaction is the variation
among the differences between means for the different levels of a factor over
different levels of the other factor. However if the interaction is detected as nonestimable or it will cause some other main effect to be non-estimable, it will not be
added to the ANOVA model.
Partek Express: Effect Sizes
45
Selecting the Grouping Factor
If there are multiple samples from the same specimen, the factor that holds the
specimen is typically specified as the grouping factor. Drag and drop the factor
from the left panel to bottom panel on the right to set the grouping effect.
Creating Comparisons


Selecting Next > will bring up the Create Comparison dialog (Figure 9.
3), which allows you to perform a linear contrast between two specific
groups within the context of ANOVA
Specify the factor or interaction to perform the comparison by selecting
one of the radio buttons in the Model Terms frame
Figure 9. 3: Create Comparison Dialog – Step 1
Configuring Comparison
The Next > button will lead you to configure the comparison (Figure 9. 4). You
must specify two groups to compare. The left panel in the Define Groups lists all
the levels (subgroups) of the selected factor or interaction.
Drag the levels in the left panel to Group1 or Group2. Figure 9. 4 shows two brain
tissues grouped together to compare to the heart tissue.
Using the Comparison Builder
You can drag and drop levels to assign levels to comparison groups. To do this,
select one or more items from one group, hold the left mouse button down and drag
them to the desired location, and then release the mouse. This will move the
selected items from the original location to the new location.
Partek Express: Effect Sizes
46
When first brought up, all levels will be shown in the Unassigned group. To select
multiple items, use <CTRL>, <Shift> and mouse click. Multiple selections can also
be done by pressing down the left mouse button over a blank area (not over the text
of an item) and dragging a bounding box (Figure 9. 5). All items that overlap with
the bounding box will be selected. Clicking on the column header will select all
items in the group. After the selection is made, start dragging.
Double clicking on an item in either Group 1 or Group 2 will move the item back to
the Unassigned group. Double clicking on the column header of Group 1 or Group
2 will move all items currently in the group back to the Unassigned group.
Figure 9. 4: Creating Comparison Dialog – Step 2 (Configuring comparison)
Figure 9. 5: Creating Comparison Dialog – Multiple Selection
Partek Express: Effect Sizes
47
Selecting Apply will add the specified contrast and stay on the same screen for
more action.
Selecting OK will add the specified contrast and will send it to the Comparison list
dialog (Figure 9. 6). Non-Estimable will be attached to the contrast name once the
contrast is detected as non-estimable.
Figure 9. 6: Viewing the Comparison List Dialog
Add Comparison
Selecting Add Comparison will add more comparisons.
Edit Comparison
will edit the corresponding comparison.
Remove Comparison
Selecting Remove Comparison will remove the corresponding comparison.
Rename Comparison
Editing the entry boxes will rename the corresponding comparison.
Running the Computations
Selecting Next > will perform the computation.
Selecting Cancel or the <Esc> key will close the dialog without doing any
computation.
Partek Express: Effect Sizes
48
Selecting < Back will go back to the previous dialog to do more ANOVA
configuration.
Running the computation includes two steps.
1. Assigning the batch effects
2. Running the ANOVA computation
Assigning the Batch Effects
The Partek Express Estimate Gene Significance will automatically assign batch
effects for you based on the correlation and significance test.
The Estimate Gene Significance uses Cramer’s V to do the correlation test between
the batch and the categorical main factors and uses Pearson correlation coefficient
to test the correlation between batch and numeric main factors. Experience shows
that the batch probably needs to be excluded from the model if it has a quite strong
correlation with the main effect or interaction since the strong correlation might
indicate a confounding or nesting-nested relationship between the batch and the
main factor or interaction. In this case, the effect of main factor will be stolen by the
batch if the batch is included in the ANOVA model. When the computed value is
larger than 0.8498, the batch will be excluded from the model.
The Significance test is used to test whether those batches that passed the
correlation test are significant or not. Estimate Gene Significance uses model
selection techniques to pick those batches that improve the model’s adjusted RSquares most.
Finally, Partek Express gives a summary page (Figure 9. 7) that shows what factors
that have been selected as interest and what factors are recommended to be included
as nuisance effects. Uncheck the batch effect if you do not want to include it in the
model.
Partek Express: Effect Sizes
49
Figure 9. 7: Estimate gene significance summary page
Setting the False Discovery Rate
A step-up multiple test correction will be automatically done for all the p-values
produced by ANOVA. Setting the false discovery rate (Figure 9. 7) will produce a
False Discovery Rate (FDR) report to the report tab when ANOVA is done. Please
refer the Multiple Test Correction for P-Values section below to get more
implementation details about step-up FDR.
Running the ANOVA computation
Selecting OK will perform the configured ANOVA computation.
Selecting Cancel or the <Esc> key will close the dialog without doing any
computation.
Implementation Details
Sir Ronald Fisher first developed ANOVA* in 1925. Many intermediate statistical
textbooks serve as an introduction to ANOVA (e.g. Steel and Torrie (1980) or
Snedecor and Cochran (1980)). Scheffé (1959) is also a classic reference.
* ANOVA is a parametric test, it makes certain assumptions about the distribution
of the response variable. The most important assumptions are that the data is
normally distributed and that the variance is approximately equal between the
groups (homogeneity of variance). Although ANOVA is most powerful when these
assumptions are met, in many cases ANOVA is very robust to violations of these
assumptions.
Partek Express: Effect Sizes
50
The Partek Express ANOVA can handle:




a balanced and an unbalanced design
random and fixed effects (mixed-model ANOVA), nested factors
multi-number of categorical effects (multi-way ANOVA)
numeric covariates (multi-way Analysis of Covariance, or ANCOVA)
Examples of each are provided below.
Example of a Balanced Experimental Design
A design is balanced when the number of samples is the same for each factor level.
Consider the two factors, Treatment and Time. This is referred to as a 2X6
experiment design because Treatment has 2 levels and Time has 6 levels.
Factor
Treatment
Time
Levels
Control, Treated
T1, T2, T3, T4, T5, T6
Figure 9. 8: An example of a balanced experimental design
Figure 9. 8 shows a balanced experimental design; in this case, a two-way crossed
ANOVA. Every level of the factor Time occurs with every level of the factor
Treatment. This is a balanced design because all of the levels of the two factors
have the same number of samples (3).
An Example of Unbalanced Experimental Design
A design is unbalanced when the number of samples is not the same for each factor
level. Below, in the Time and Treatment example, when a subject died at T6 (Time
6), the experiment became unbalanced (Figure 9. 9).
Partek Express: Effect Sizes
51
Figure 9. 9: An example of an unbalanced experimental design
Missing Values & Missing Treatment Combinations
If the levels of all the factors are completely crossed, Type III sums of squares is
used. However, if missing treatment combinations occur in any interaction, Type IV
sums of squares is used.
A missing treatment combination occurs when one of the cells in the multi-way
ANOVA table has no entries. If all three treated samples at time T6 are not
available; then the Treated X T6 combination is missing (Figure 9. 10).
Figure 9. 10: An example of a missing treatment combination
If an interaction corresponding to a treatment combination has no replication,
Partek Express will automatically remove that interaction from the model.
Therefore, when testing multiple response variables, the p-values of the removed
interactions will be represented by question marks (“?”) to indicate that the value
could not be computed.
Mixed Model ANOVA
To obtain estimates of variance components for mixed models, Partek Express uses
the method of moments estimation (Eisenhart, 1947).
The method of moments is used to equate analysis of variance mean sum of squares
to their expected values (s=Cσ²). S is a vector of the mean sum of squares, C is a
matrix, and σ² is a vector of variance components. The estimates of σ² are C-1s.
However, the method of moments method can produce negative estimates.
Partek Express: Effect Sizes
52
Nested Factors
Most of the time, the grouping factor is nested with at least one of the main effects.
In a two-way or a multi-way ANOVA, if each level of one factor occurs with each
level of another factor, the factors are said to be “crossed”. If, however, the levels of
one factor only occur within a single level of another factor, then one factor is said
to be “nested in” the other factor.
In the example below, there are two factors: Type and Subject ID. Type has 2 levels
and Subject has 10 levels.
Factor
Type
Subject
Levels
Normal, TS21
1218, 1389, 1390, 1411, 1478, 1479, 1521, 1565, 748, 847
Notice in the Crosstabulations table (Figure 9. 11) that each level of Subject
occurred within one level of Type. Therefore, Subject is nested within Type.
Figure 9. 11: An example of a nested factor
Random vs. Fixed Effects
The grouping factor should be categorical (random). If it is not random, the
Estimate Gene Significance procedure will automatically set it to random to do
ANOVA and set it back after ANOVA is done.
Most factors in an ANOVA are fixed factors, i.e. the levels of that factor represent
all the levels of interest. Examples of fixed factors include gender, race, strain, etc.
However, in experiments that are more sophisticated, a factor can be a random
effect, meaning the levels of the factor only represent a random sample of all of the
levels of interest. Examples of random effects include subject and batch. Consider
the example where one factor is type (with levels normal, diseased), and another
Partek Express: Effect Sizes
53
factor is subject (the subjects selected for the experiment). In this example, type is a
fixed factor since the levels normal and diseased represent all conditions of interest.
Subject, on the other hand, is a random effect since the subjects are only a random
sample of all the levels of that factor.
Equations of the ANOVA Results
Contrast Equations
When contrasting the average of treatments A and B versus treatment C, the
contrast equation is
1
1
A  B  1C  0
2
2
The ratio is calculated using the least square mean (LS Mean) of each term, thus the
ratio for the contrast is given by:
1
1
LSMean ( A)  LSMean ( B)
2
2
LSMean (C )
A second example contrasts the average of treatments A, B, and C versus the
average of treatments C and D. The contrast equation is
1
1
1
1
1
A B C  D E  0
3
3
3
2
2
The ratio for the contrast is given by:
1
1
1
LSMean ( A)  LSMean ( B)  LSMean (C )
3
3
3
1
1
LSMean ( D)  LSMean ( D)
2
2
Fold changes are calculated in a similar fashion using LSMeans.
LS Mean and Geometric Mean
The LS Mean (Least Squares Mean) is calculated as the linear combination (sum) of
the estimated means from a linear model (e.g. ANOVA, regression, etc). The LS
mean is based on the factors specified in the model, thus, the LS mean is “model
dependent” whereas arithmetic mean is “model independent”. When the data results
from a balanced experiment (same number of treatment combinations in each
group), the arithmetic mean and LS mean are identical. In unbalanced data, the
arithmetic mean and LS mean are different. In an unbalanced experiment, the LS
means are preferred because they reflect the model being fit to the data.
Consider a simple unbalanced two-factor experiment containing a control group and
a treated group, with unequal number of male and female animals in each group
Partek Express: Effect Sizes
54
(Figure 9. 12). The control group contains 4 females and 2 males, and the treated
group contains 2 females and 5 males.
Figure 9. 12: Unbalanced two-factor experiment crosstabulation
Using the arithmetic mean to estimate the means of the control and treated groups
ignores the imbalance of male and female in the two groups and may be biased. For
example, if you are estimating the effects of a gene’s expression which lies on Y
chromosome (females don’t have a Y chromosome, thus they will have lower
expression than males on this gene), the arithmetic mean would overestimate the
mean in treated group since the treated group contains more males, and
underestimate the control group since the control group contains more females
(Figure 9. 13).
Arithmetic Mean
Least Squares Mean
Figure 9. 13: Comparison of the arithmetic mean and LS mean in the control and
treated group of gene’s expression that lies on Y chromosome
The LS mean uses the estimates for both factors in the design, treatment, and
gender, and adjusts the means for the treated and control groups to account for the
imbalance in gender between the groups. The LS mean would produce a more
accurate, unbiased estimate of the mean of the treated and control groups in this
example (Figure 9. 13).
Data is often log transformed prior to doing statistical analysis in order to transform
a multiplicative effect into an additive effect. However, scientists sometimes want
to interpret effects as ratios, in which case log transformed data is inappropriate,
since it has been converted from a multiplicative effect to an additive effect. Simply
anti-logging the mean of logged data does not produce the mean of the un-logged
data; however, it does produce the geometric mean of the un-logged data. Antilogging a least squares mean produces a value that we call a “least squares
geometric mean”.
When a ratio is calculated based on LS means, the ratio of Group1 vs. Group2 is:
LSMean (Group1)
LSMean (Group 2)
Partek Express: Effect Sizes
55
When a ratio is calculated based on least squares geometric means, the estimate of
the LS means on logged data is first calculated for each group, and the difference of
the LS means is then anti-logged using the same base:
a LSMean(Group1) LSMean(Group2)
“a” is the base the data is log transformed on. This is equivalent to calculating the
ratio of the least squares geometric means for the two groups:
a LSMean(Group1)
a LSMean(Group2)
This ratio is more appropriate than the simple ratio of LS means in the case that
analyses have been performed on logged data.
Multiple Test Correction for P-Values
A p-value is the probability that the observed values could have occurred by
chance. It indicates the probability that one could obtain a test statistic that is as
extreme as or more extreme than the observed one if the null hypothesis is true. Pvalues provide a sense of the strength of the evidence against the null hypothesis.
The lower a p-value is, the stronger the evidence to reject the null hypothesis.
When multiple tests are performed, the probability of incorrectly rejecting a single
null hypothesis (“false positive” or “Type I error”) increases. There are several
methods to correct Type I error for multiple tests. Partek Express only uses the most
general one -- step up (Benjamini & Hochberg, 1995) method.
False Discovery Rate (FDR) – Step Up
False Discovery Rate is the proportion of false positives among all positives. In the
step up method, there are n number of p-values; they are sorted by ascending order,
and m represents the rank of a p-value. The calculation compares p-value*(n/m)
with the specified significance level, and the cut-off p-value is the one that
generates the last product that is less than the significance level.
Viewing the Effect Sizes Plots
The effect sizes tab presents either a bar chart or pie chart visualization to help
display the importance and significance of experimental factors included in the
analysis.
When using ANOVA to calculate p-values for effect sizes, an intermediate value,
called the F-Ratio, is also calculated. F-ratio is a measure of the variance of the
data explained by a factor relative to the unexplained variance or error.
Partek Express: Effect Sizes
56
To get an impression of the importance of each factor to the transcriptome overall,
you can look at the mean of the factor’s F-ratio across all genes on the bar chart or
on the pie chart.
To switch between bar chart and pie chart views, select the appropriate sub tab
within the effect sizes tab. Larger effect sizes indicate that the factor is more
significant to the data.
In the bar chart visualization, each factor is represented by a vertical bar and is
labeled with both the name of the factor and the height of the bar (Figure 9. 14).
The vertical axis is the mean F-ratio available from the ANOVA table as described
above. The error bar in the chart is always “1” so that the F-ratios are easier to
interpret. Relative to this, taller bars represent more significant factors. Bars at, or
near, error represent factors which are not significant to the transcriptome overall.
Figure 9. 14: Viewing the Effect Sizes Bar Chart
In the pie chart visualization, the magnitude of the F-ratios is used to generate a pie
chart simplifying comparison of importance or effect between different factors
(Figure 9. 15). Each section of the pie chart is labeled with the name of the factor
and a percentage of the pie contained. Larger pieces of the pie are more significant,
while factors at or near the size of the error slice represent factors which are not
significant to the transcriptome overall.
Partek Express: Effect Sizes
57
Figure 9. 15: Viewing the Effect Sizes Pie Chart
Configuring the Effect Sizes Plots
The title or axis labels of the effect size visualizations can be set on the left of the
visualization. To make changes, type in the desired title and axis labels and select
Apply.
References
Eisenhart, C. (1947). The assumptions underlying the analysis of variance.
Biometrics, 3: 1-21.
Tamhane, Ajit C., & Dunlop, Dorothy D. (2000). Statistics and Data Analysis from
Elementary to Intermediate. Prentice Hall. Pages 473-474.
Thompson, W.A., Jr (1962). The Problem of Negative Estimates of Variance
Components. Ann. Math. Stat. 33: 273-289.
Benjamini, Y., Hochberg, Y. (1995). Controlling the false discovery rate: a practical
and powerful approach to multiple testing, JRSS, B, 57, 289-300.
Partek Express: Effect Sizes
58
Partek® Express™: Viewing Gene Significance
Introduction
Partek® Express™ estimates Gene Significance using Analysis of Variance
(ANOVA). For more information about ANOVA, please refer to the Effect Sizes
chapter.
Viewing Gene Significance Results
The ANOVA results will be summarized in a spreadsheet and a dot plot (Figure 10.
1) which are located in the Gene Significance Estimates tab.
Figure 10. 1: Estimate Significance Gene Results
Searching for Genes
On top of the ANOVA Result Spreadsheet, a gene name search panel is provided. To
search for a gene name, type it into the entry box and choose Next or Previous. If
Next is selected, the search will begin at the currently active cell and go downwards.
If Previous is selected, the search will begin at the currently active cell and go
upwards. Two options, Match case and Match whole cell are available for search.
To enable an option, click the checkbox next to it.
If a match is found, the matched cell will be activated and the matching row will be
highlighted. You can continue searching down or up by clicking Next or Previous.
Result Spreadsheet
The analysis is performed on imported intensities for each gene. Each row of the
Results Spreadsheet corresponds to one gene.
Partek Express: Estimating Gene Significance
59
One of the most critical pieces of information contained in the Gene Significance
table is the p-value per gene per categorical variable. A p-value is a test statistic
(between zero and one) used to rank significance of results of starting with the null
hypothesis that a gene is similarly expressed across conditions. Stated slightly
differently and somewhat simplistically, the smaller the p-value for a given gene,
the more likely that the gene shows differential express across the given categorical
variables.
Each biological factor included in the ANOVA model will produce one additional
column in the Gene Significance table. Each pair-wise comparison included in the
ANOVA will add three additional columns into the table.
By default, the genes are sorted by the p-values of the first factor of interest. The
most significant gene is on the first row. To sort by a different column, simply right
click on the column heading and select Sort Ascending or Sort Descending in the
pop-up menu.
Dot Plot
Dot plot can be viewed gene by gene just by clicking the row label corresponding to
the response variable. It shows original intensities grouped by the factor in the
Group by dropdown list and colored by the factor in the Color by dropdown list.
In the dot plot, each dot is an individual sample data point. The X-Axis represents
the different types and the Y-Axis displays the log2 expression level of the gene.
When data is in log2 space, it is important to remember that the scale is typically
between zero and 16, and that any increment change of one represents a twofold
change in abundance. So if a gene changes from 6 in one condition to 8 in another
condition, that represents a fourfold change between the two conditions.
References
Eisenhart, C. (1947). The assumptions underlying the analysis of variance.
Biometrics, 3: 1-21.
Tamhane, Ajit C., & Dunlop, Dorothy D. (2000). Statistics and Data Analysis from
Elementary to Intermediate. Prentice Hall. Pages 473-474.
Thompson, W.A., Jr (1962). The Problem of Negative Estimates of Variance
Components. Ann. Math. Stat. 33: 273-289.
Benjamini, Y., & Hochberg, Y. (1995). Controlling the false discovery rate: a
practical and powerful approach to multiple testing, JRSS, B, 57, 289-300.
Partek Express: Estimating Gene Significance
60
Partek® Express™: Power Analysis
Introduction
The Partek® Express™ Power Analysis procedure conducts prospective analysis,
which is used to:


Determine the minimum sample size to achieve adequate power on a
given fold change
Determine what fold change could be acquired on the given sample size to
achieve the specified power
Implementation Details
Input for Power Analysis includes:







Experimental design
Statistical model (ANOVA)
Comparison (contrast) on which to do power analysis
Effect size (fold change)
Sample size
Significance level (alpha)
Power (1-beta)
Partek Express Power Analysis obtains the experimental design, statistical model
(ANOVA) and comparison (contrast) from the current study. These three
parameters are already decided by previous steps in Partek Express before doing
power analysis. Please refer the Estimate Gene Significance chapter to get more
information about how to configure an ANOVA model, how to set-up comparisons,
as well as implementation details.
Let Y be the response vector, X be the design matrix, β be the model parameter
vector, so the underlying function for the ANOVA model can be written in the form
of Y  X   where ε is the error term which is normally and independently
distributed with mean 0 and standard deviation σ. Comparison (contrast) was set in
the Estimate Gene Significance (ANOVA) step to test the null hypothesis
H 0 : L   where L is the contrast matrix.
For the four parameters like effect size, sample size, significance level and power,
each can be obtained by solving the following power analysis formulas when fixing
the other three.
Partek Express: Power Analysis
61
Power Analysis Formulas
power  P( F (rL , N  rx ,  )  F1 (rL , N  rx )) (Muller and Peterson 1984)
Where rL is the rank of contrast L, rx is the rank of design matrix X, N is the total
sample size, α is the significance level and λ is the non-central parameter of F
statistic under alternative hypothesis H A : L  0 .
  N ( L )( L( X diag ( w) X ) 1 L) 1 ( L ) 2
Where X is composed of the unique rows of design matrix X, w is a vector of
weights which reflect the proportion of each unique row in the whole design matrix
X. σ is the ANOVA model standard deviation.
Configuring the Power Analysis Dialog
The Partek Express embedded workflow will introduce you to the Power Analysis
step automatically by consistently selecting the Next button on the main window.
The Power Analysis step is after the View Effect Sizes step: Start > Study
Definition > Data Import & QC Check > PCA > Effect Sizes > Power Analysis.
Figure 11. 1: Configuring Power Analysis
Selecting Comparison
All the comparisons that were set in the Estimate Gene Significance step will be
shown in the Power Analysis dialog (Figure 11. 1); however, only one comparison
can be selected to do power analysis at a time. To specify one comparison, just click
one of the radio buttons in the Comparison frame. If no comparison was set in
Estimate Gene Significance step, the Power Analysis dialog will not be invoked.
Configuring the Effect Size
Selecting the Advanced... button in the Power Analysis frame will open the Power
Analysis Configuration dialog (Figure 11. 2) to configure the parameters of effect
size, sample size, significance, and power.
Partek Express: Power Analysis
62
Specify the range and step size for effect size in this dialog so that the Power
Analysis will produce the minimum sample sizes (the newly produced sample size
is supposed to be assigned to each comparison group with the same proportion as
the original dataset) required to achieve each of the specified effect sizes,
respectively. Effect size (fold change here) must be greater than or equal to one.
Decreasing the effect size will probably require more samples. For better viewing,
10 points of effect size can be accommodated in the specified range by the specified
step size.
Figure 11. 2: Configuring Power Analysis
Configuring the Sample Size
Specify the range and step size for the sample size so that the Power Analysis will
produce a fold change that is set by the given sample sizes. Sample size should be
larger than model’s degree of freedom. For better viewing, 10 points of sample size
can be accommodated in the specified range by the specified step size.
Configuring the Significance
The significance level is the probability to reject the null hypothesis when the null
hypothesis is actually true. A commonly used significance level of 0.1 is set as the
default. The range for significance level is between 0 and 1. Decreasing the
significance level will probably require more samples to achieve the same fold
change.
Configuring the Power
The power level is the probability to reject the null hypothesis when the null
hypothesis is actually false. A commonly used power of 0.8 is set as the default.
The range for power is between 0 and 1. Increasing the power will probably require
more samples to achieve the same fold change.
Partek Express: Power Analysis
63
Saving the Power Analysis Configuration
Selecting OK in the Power Analysis Configuration dialog (Figure 11. 2) will save
all the parameters configured and dismiss the dialog; selecting Cancel will close the
dialog without saving.
Running the Power Analysis
Selecting OK in the Power Analysis dialog (Figure 11. 1) will perform the
configured power analysis and dismiss the dialog; selecting Cancel will close the
dialog without doing any computation.
Visualizing the Data: Box plot
The box plot provides a way to graphically view the numeric data through five
numbers in summary. The five numbers, 10th percentile, 25th percentile, 50th
percentile, 75th percentile and 90th percentile of the power analysis, result in the
gene level. Partek Express Power Analysis will generate two box plots, Fold
Change to Sample Size and Sample Size to Fold Change. These two box plots can
be invoked by selecting the radio button on the Power Analysis tab in the Partek
Express main window.
Box Plot: Fold Change to Sample Size
The Fold Change to Sample Size box plot indicates the sample size (in Y axis) to
achieve the adequate power of the given fold change (in X axis).
Figure 11. 3: Box plot of Fold Change to Sample Size
Note: Y axis tick marks are in log (base 2) scale. The current study sample size is
marked with a blue reference line. Moving the mouse over a box-whisker will show
Partek Express: Power Analysis
64
a more detailed sample size report. In the examples shown in figure 3, to detect
50% of genes with a fold change of at least 1.75 would require 20.85 (round up to
21) samples.
Note: Power Analysis for a specific fold change assumes the proportion of samples
in each category is similar to that of the existing samples. Table 11. 1shows the
number of samples needed for a fold change of at least 1.75.
# of Samples
Percent of Genes
Fold Change
18.70
10%
1.75
19.44
25%
1.75
20.85
50%
1.75
24.17
75%
1.75
32.32
90%
1.75
Table 11. 1: Viewing the number of samples needed to achieve a fold change of at
least 1.75
Box Plot: Sample Size to Fold Change
The Sample Size to Fold Change box plot shows what fold change (in X axis) could
be acquired on the given sample size (in Y axis).
Figure 11. 4: Box plot of Sample Size to Fold Change
The blue line marks the number of samples used in the analysis of the study.
Moving the mouse over a box-whisker will bring up the detailed fold change report
for the respective samples size. In the example shown in Figure 11. 4, using 50
samples in the study would detect 50% of genes with a fold change of at least 1.28.
Partek Express: Power Analysis
65
Note: Power Analysis for a specific sample size assumes the proportion of samples
in each category is similar to that of the existing samples. Table 11. 2 shows the
fold change of 50 samples at varying percentages.
# of Samples
Percent of Genes
Fold Change
50
10%
1.16
50
25%
1.21
50
50%
1.28
50
75%
1.38
50
90%
1.53
Table 11. 2: Viewing the fold change of 50 samples at differing percentages
References
Muller, K.E. and Peterson, B. L. (1984), Practical methods for computing power in
testing the multivariate general linear hypothesis. Computational
Statistics and Data Analysis, 2: 143-158.
Muller, K.E. and Benignus, V.A. (1992), Increasing scientific power with statistical
power. Neurotoxicology and Teratology, 14: 211-219.
Muller, K.E., LaVange, L.M., Ramey, S.L. and Ramey, C.T. (1992), Power
calculations for general multivariate models including repeated measures
applications. Journal of the American Statistical Association, 87: 12091226.
Partek Express: Power Analysis
66
Partek® Express™: Report
Introduction
The Partek® Express™ Report tab records every step that has been done in the
current study. It can be used to:



Determine the data set used in the study
Determine what analysis was done and how it was done
Find the reference paper for each analysis performed
Description of Report
The report provides step by step information about the actions taken in the Partek
Express workflow. The main steps recorded are: Create Study, Import, PCA,
ANOVA and Power Analysis. The reference papers that are related to each of the
above steps are all recorded in the References section.
Create Study
From the Report tab, you can find the name of the study, the time on which the
study was created, and the user who created the study.
Import
The report of the Import step will tell you which files were imported, where to find
the library files, and what algorithm was used to do the importing. The algorithm
Partek Express uses to import CEL or CHP files is RMA.
RMA
The Partek Express implementation of RMA is tuned for speed and decreased
memory usage. There are four steps involved in the RMA importing method; only
perfect match (PM) probe values are used in this method:




Background correction is used on the PM values
Quantile normalization is used across all the chips in the experiment
Log (base 2) transforms the data, and if the data values are <= 0, then they
will be marked as missing
Median polish summarization is used to give a robust signal for each gene.
Note: Median polish might give the same summarized values for all/most
samples if your sample size is very small. When the sample size is <=4,
mean summarization is used
67
Partek Express: Report
PCA
The correlation method used to draw a principal components analysis (PCA) plot is
recorded in the PCA section.
ANOVA
The report for the ANOVA step will help you find the file on which ANOVA was
performed. It will tell you what method was used as well as what model was created
to do the ANOVA. A Step Up false discovery rate (FDR) report will be produced
after the ANOVA is done.
Power Analysis
The report for the Power Analysis step remembers all the parameters configured in
the Power Analysis dialog. The parameters include the comparison on which the
Power Analysis was performed, effect size, sample size, significance level and the
power specified to do power analysis.
References
All reference papers related to the performed steps, above, will be shown in the
References section.
68
Partek Express: Report
Partek® Express™: Pathway Analysis
Introduction
The Partek Express Pathway Analysis is used to:
 Dump useful information from the ANOVA results
 Provide an interface to launch Ariadne Pathway Studio® Explore or
Ingenuity® IPA® to do the Pathway Analysis
Please note: you need to have Ariadne Pathway Studio® Explore or Ingenuity®
IPA® installed on the same computer as Partek® Express™ is to do the Pathway
Analysis step.
Configuring the Pathway Analysis Dialog for Ariadne Pathway Studio
Explore
The Partek Express embedded workflow will introduce you to the Pathway Analysis
step automatically by selecting the Next button on the main window. The Pathway
Analysis step is after the View Effect Sizes step: Start > Study Definition > Data
Import & QC Check > PCA > Effect Sizes > Pathway Analysis.
The Ariadne tab in the Pathway Analysis dialog will configure and launch Ariadne
Pathway Studio Explore (Figure 13. 1).
Figure 13. 1: Configuring Ariadne Pathway Analysis
Selecting Ratio
Estimate Gene Significance step generated a ratio column for each comparison in
the result spreadsheet. All those comparisons’ name will be shown in the Pathway
Partek Express: Pathway Analysis
69
Analysis dialog (Figure 13. 1) except that whose ratio column is all composed of
question marks (non estimable values). To select one ratio, just check the check
button in front of the desired comparison name.
Selecting Numeric Value Associated
The p-value column with the same comparison name as ratio is set as the default
Numeric value associated. If this column is all composed of question marks, any
numeric column with the same comparison name as ratio will be set as default. To
change the Numeric value associated, click the dropdown list box and select one
from the dropdown list.
Running the Pathway Analysis with Ariadne Pathway Studio Explore



Selecting Launch Ariadne Pathway Studio® Explore will output the
ratio and Numeric value associated for each comparison selected in the
Pathway Analysis dialog (Figure 13. 1) and launch Explore to do pathway
analysis
Selecting Cancel will close the dialog without doing pathway analysis
Selecting Tell Me More… will lead you to this Pathway Analysis manual
Configuring the Pathway Analysis Dialog for Ingenuity IPA
The Ingenuity tab in the Pathway Analysis dialog will launch Ingenuity Pathway
Analysis (Figure 13. 2).
Figure 13. 2: Configuring Ingenuity Pathway Analysis
Partek Express: Pathway Analysis
70
Running the Pathway Analysis with Ingenuity IPA



Selecting the Launch Ingenuity® Pathways Analysis will invoke the
default browser. Selecting Send to IPA will launch Ingenuity IPA for the
pathway analysis (Figure 13. 3)
Selecting Cancel will close the dialog without doing pathway analysis
Selecting Tell Me More… will lead you to this Pathway Analysis manual
Figure 13. 3: Launching Ingenuity Pathway Analysis
Partek Express: Pathway Analysis
71
Partek® Express™: The Main Menu
In this chapter, you will find information about the Partek® Express™ menus.
The File Menu
New Study
Creates a new study. If another study is currently open in Partek Express, it will
close the current study first. You will be prompted to save the current study if
there are pending unsaved changes.
Load Study
Allows you to choose a study file to open. If another study is currently open in
Partek Express, it will close the current study first. You will be prompted to save
the current study if there are pending unsaved changes.
Save Study
Saves the current study.
Save Study As…
Saves the current study with a different name.
Zip Study
Allows you to specify a .zip file to package the current study and associated data
files (.CEL/.CHP and .ARR files) into (Figure 14. 1). Check the appropriate
checkboxes to include the data files
Figure 14. 1: Viewing the Zip Study Dialog
If there are pending unsaved changes, you will be prompted to save the study
before continuing.
Close Study
Closes the current study.
Partek Express: The Main Menu
72
Recent Studies
Opens a recent study.
Save Image As…
Saves the current image in the viewer. This menu is only available for some
plots/charts, such as the QC Metrics Plot, PCA Plot, Effect Sizes Bar Chart,
Effect Sizes Pie Chart, Gene Significance Estimates Dot Plot, and Power Analysis
Box Whisker Plot.
Export
Under the Export cascade menu there are the following items.
Sample Information (Tab Delimited)
Saves the sample information as a tab delimited text file.
Sample Information (.ARR)
Exports the sample information into Affymetrix .ARR file(s). This operation
requires that the .CEL or .CHP files specified in the sample information exist.
The .ARR files will be exported to the same folder where the data files are
located. If .ARR files already exist, you will be prompted to overwrite.
Intensities (Genes on Rows, Excel compatible)
Exports sample intensity values as a tab delimited file. Each column of the file
will contain one sample. The file format is compatible with Microsoft Excel.
This operation is only available after Data Import & QC Check.
Intensities (Samples on Rows, Partek GS compatible)
Saves the sample intensities to two files (e.g. xyz and xyz.fmt). xyz will have the
intensities, and xyz.fmt will have the format information. You can use Partek®
Genomics Suite™ (Partek GS), to open xyz.fmt. Note: you will need both files
(xyz and xyz.fmt) in order for Partek GS to open.
Gene significance Estimates (Tab Delimited, Genes on Rows)
Exports the gene significance estimates spreadsheet as tab delimited text files
with one row representing one gene. This file is Excel compatible.
Report
Exports the study information under the Report tab to a text file.
Manage Library Path
Brings up a folder browser (Figure 14. 2). Here the user specified folder will be
used to search for library files when importing data; this will also be the directory
Partek Express automatically downloads library files to.
Partek Express: The Main Menu
73
Figure 14. 2: Viewing the Browse for Folder Dialog
Manage Library Files
Invokes the file manager (Figure 14. 3), which allows for the update of annotation
and library files. For example, in Figure 14. 3, checking the Update Available
button, and then selecting the Download button will get the latest HG-U133A
Probeset Annotation file.
Figure 14. 3: Viewing the File Manager Dialog
Partek Express: The Main Menu
74
Exit
Closes the application.
The Edit Menu
Plot Fonts Configuration
Invokes the Plot Fonts Configuration dialog (Figure 14. 4). Here you can specify
the font size for the Title, Axis, Axis Title, Legend, and Label, and specify the Plot
Font.
.
Figure 14. 4: Plot Fonts Configuration Dialog
The Help Menu
On-line Tutorial
Launches a web browser, which will show the tutorial at:
http://www.partek.com/~devel/PartekExpressDownSyndrome_tutorial.pdf
User’s Manual
Invokes the Partek Express User’s Manual.
License Information…
Displays license information (Figure 14. 5).
Partek Express: The Main Menu
75
Figure 14. 5: Viewing License Information
Graphics Information…
Displays graphics information.
Check for Updates
Displays the Partek Express update webpage in a web browser.
About Partek Express
Displays information about the software.
Copyright  2010 by Partek Incorporated. All Rights Reserved. Reproduction of this material without express written consent from
Partek Incorporated is strictly prohibited.
Partek Express: The Main Menu
76