Affymetrix Expression Console Software v1.0 - User Guide

manual.book Page i Thursday, October 5, 2006 12:21 PM

Affymetrix Expression Console™ Software Version 1.0 — User Guide

P/N 702387 Rev. 1.0

ii

manual.book Page ii Thursday, October 5, 2006 12:21 PM

For research use only.

Not for use in diagnostic procedures.

Trademarks

Affymetrix®, ®, GeneChip®, HuSNP®, GenFlex®, Flying Objective™, CustomExpress®, CustomSeq®, NetAffx™, Tools To Take You As Far As Your Vision®, The Way Ahead™, Powered by Affymetrix™, GeneChip-compatible™, and Command Console™ are trademarks of Affymetrix, Inc.

All other trademarks are the property of their respective owners.

Limited License Notice

Limited License. Subject to the Affymetrix terms and conditions that govern your use of Affymetrix products, Affymetrix grants you a non-exclusive, non-transferable, non-sublicensable license to use this Affymetrix product only in accordance with the manual and written instructions provided by Affymetrix. You understand and agree that except as expressly set forth in the Affymetrix terms and conditions, that no right or license to any patent or other intellectual property owned or licensable by Affymetrix is conveyed or implied by this Affymetrix product. In particular, no right or license is conveyed or implied to use this Affymetrix product in combination with a product not provided, licensed or specifically recommended by Affymetrix for such use.

Patents

Software products may be covered by one or more of the following patents: U.S. Patent No's. 5,733,729; 5,795,716; 5,974,164; 6,066,454; 6,090,555, 6,185,561 6,188,783, 6,223,127; 6,228,593; 6,229,911; 6,242,180; 6,308,170; 6,361,937; 6,420,108; 6,484,183; 6,505,125; 6,510,391; 6,532,462; 6,546,340; 6,687,692; 6,607,887; and other U.S. or foreign patents.

Copyright

©2006 Affymetrix, Inc. All rights reserved.

iii

manual.book Page iii Thursday, October 5, 2006 12:21 PM

Table of Contents

CHAPTER 1 Welcome 3

INTRODUCTION 3

WORKFLOW DIAGRAM 5

ABOUT THIS MANUAL 6

FAQS 7

CONVENTIONS USED IN THIS GUIDE 7

TECHNICAL SUPPORT 10

CHAPTER 2 Installation and Setup 13

SOFTWARE REQUIREMENTS 13

MINIMUM HARDWARE RECOMMENDATIONS 13

INSTALLATION INSTRUCTIONS 14

SETUP PROFILE 14

LIBRARY FILES 17

CHAPTER 3 Creating a Study 25

GETTING STARTED 25

ANALYZING FILES 32

iv Affymetrix® Expression Console™ Software v.1.0 – User Guide

manual.book Page iv Thursday, October 5, 2006 12:21 PM

CHAPTER 4 3’ Expression Array Analysis 35

ANALYSIS CONTROLS 35

ANALYSIS ALGORITHMS 35

ANALYZE DATA 36

RESULTS 38

CHAPTER 5 Exon Array Analysis 43

SPECIFY CONTROLS 43

CHAPTER 6 QC Tables and Graphs 53

REPORTS 54

GRAPHS 58

CHAPTER 7 Exporting Data 81

EXPORTING 81

SAVING IMAGES 86

CHAPTER 8 Controls and Thresholds 89

REPORT CONTROLS 89

REPORT THRESHOLDS 95

Contents v

manual.book Page v Thursday, October 5, 2006 12:21 PM

CHAPTER 9 Advanced Analysis 101

3' EXPRESSION ARRAY CONFIGURATIONS 101

MASK FILES 109

MAS 5.0 CHP FILES IN THE COMMAND CONSOLE FORMAT 113

MAS 5.0 CHP FILES IN THE GCOS FORMAT 115

ADVANCED CONFIGURATION EXON ANALYSIS 116

CHAPTER 10 Analysis Scripts 123

ANALYSIS SCRIPT CREATION 123

ANALYSIS SCRIPT DELETION 125

ANALYSIS SCRIPT EXECUTION 126

Appendix A Algorithms 129

MAS 5.0 ALGORITHM 129

RMA ALGORITHM 130

PLIER ALGORITHM 131

COMPARISON OF ALGORITHMS 132

Appendix B Algorithm Parameters and Outputs 135

MAS 5.0 COLUMN HEADINGS 135

vi Affymetrix® Expression Console™ Software v.1.0 – User Guide

manual.book Page vi Thursday, October 5, 2006 12:21 PM

COLUMN HEADINGS FOR RMA AND PLIER 138

PROBE SET SUFFIXES 140

Index 143

manual.book Page 1 Thursday, October 5, 2006 12:21 PM

Chapter 1
Welcome


Chapter 1

3


Welcome

Welcome to the Affymetrix® Expression Console™ software (v.1.0) User Guide. The Expression Console application provides signal estimation and QC functionality for the GeneChip® Expression Arrays (3’ Expression Arrays and Exon Arrays). The Expression Console software allows users to:

• Generate probe set summarization (CHP) files from feature intensity (CEL) files for both 3’ Expression Arrays and Exon Arrays

• Capture a standard set of metrics for evaluating the success of the individual hybridizations for both 3' Expression Arrays and Exon Arrays

• Identify outlier samples in the data set

The Expression Console software is targeted for research personnel (such as laboratory technicians, research associates, and scientists) analyzing Affymetrix GeneChip® data.

Introduction

The Affymetrix® Expression Console™ software provides an easy way to create summarized expression values (CHP files) for individual or collections of 3' Expression Array and Exon Array feature intensity (CEL) files. In addition to CHP writing, the Expression Console application also produces a collection of QC metrics for evaluating the success of hybridizations. The user defines thresholds for these metrics and the software highlights the metrics that do not meet the defined thresholds. In addition, the CHP files are highlighted in the study table, if they have any metrics outside of the defined thresholds. Individual QC metrics for each labeling technique are discussed in Chapter 6, QC Tables and Graphs.

The Expression Console application contains graphic capabilities for visual inspection of the hybridization results. To identify outliers, the application is designed to display:

• Line graphs for individual or collections of metrics or probe sets

• Box plots for signal distributions before or after normalization

4 Affymetrix® Expression Console™ Software v.1.0 – User Guide


• MvA plots for signal distributions

• Heat maps for correlation matrices

The Expression Console application is not a secondary analysis

package. However, it does create the CHP files required for

secondary analysis packages from the Affymetrix GeneChip®

Compatible Program.

Chapter 1 | Welcome 5


Workflow Diagram

Figure 1.1

Affymetrix® Expression Console™ software workflow



About this Manual

This manual presents information about the Affymetrix Expression Console™ software in the following chapters and appendices:

• Chapter 2, Installation and Setup: Describes how to install and configure the Affymetrix® Expression Console™ v.1.0 software.

• Chapter 3, Creating a Study: Describes how to create a study to analyze the array data.

• Chapter 4, 3’ Expression Array Analysis: Describes how to create CHP files from CEL files for 3' Expression Arrays using either the MAS5, RMA, or PLIER algorithm.

• Chapter 5, Exon Array Analysis: Describes how to create CHP files from CEL files for Exon Arrays by applying either RMA or PLIER algorithms.

• Chapter 6, QC Tables and Graphs: Describes how to run reports and apply graphs for data interpretation.

• Chapter 7, Exporting Data: Describes how to export data using PDF, TXT, and PNG file options.

• Chapter 8, Controls and Thresholds: Describes how to identify, define, modify and/or remove controls and thresholds.

• Chapter 9, Advanced Analysis: Describes how to modify the default algorithm parameters for either the 3' Expression Array or Exon Array.

• Chapter 10, Analysis Scripts: Describes how to create a script that automates the process of running the selected analysis algorithm and creating a standard set of QC graphs and tables.

• Appendix A, Algorithms: Briefly describes the algorithms offered in the Expression Console™ software with links to reference material for further reading.



• Appendix B, Algorithm Parameters and Outputs: Gives definitions for report column headings, which represent MAS5, RMA, and PLIER output data.

FAQS

A list of Frequently Asked Questions (FAQS) about the Affymetrix® Expression Console™ software can be found on the Affymetrix website at www.affymetrix.com; then go to /Support/Product/Software/Expression Console Software.

Conventions Used in This Guide

This manual provides a detailed outline for all tasks associated with Affymetrix® Expression Console™ software. Various conventions are used throughout the manual to help illustrate the procedures described. Explanations of these conventions are provided below.

STEPS

Instructions for procedures are written in a numbered step format. Immediately following the step number is the action to be performed. Following the response, additional information pertaining to the step may be found and is presented in paragraph format. For example:

1. Click Yes to continue.The Delete task proceeds. In the lower right pane the status is displayed.

FONT STYLES

Bold fonts indicate names of commands, buttons, options or titles within a dialog box. When asked to enter specific information, the input is displayed in italics within the procedure being outlined.For example:

1. Click the Find toolbar button ; or

http://www.affymetrix.com



Select Edit Find from the menu bar.The Find dialog box opens.

2. Enter AFFX-BioB-5_at in the Find what box, then click Find Next to view the first search result.

3. Continue to click Find Next to view each successive search result.

SCREEN CAPTURES

The steps outlining procedures are frequently supplemented with screen captures to further illustrate the instructions given.

ADDITIONAL COMMENTS

Throughout the manual, text and procedures are occasionally accompanied by special notes. These additional comments and their meanings are described as follows:

The screen captures depicted in this manual may not exactly match

the windows displayed on your screen.

Information presented in Tips provide helpful advice or shortcuts

for completing a task.

The Note format presents supplemental information pertaining to

the text or procedure being outlined.

The Important format presents important information that may

affect the accuracy of your results.



Caution notes advise you that the consequence(s) of an action may

be irreversible and/or result in lost data.

Warnings alert you to situations where physical harm to person or

damage to hardware is possible.



Technical Support

Affymetrix provides technical support to all licensed users via phone or E-mail. To contact Affymetrix® Technical Support:

AFFYMETRIX, INC.3420 Central ExpresswaySanta Clara, CA 95051 USA

Tel: 1-888-362-2447 (1-888-DNA-CHIP)Fax: 1-408-731-5441

[email protected]

[email protected]

AFFYMETRIX UK Ltd.,Voyager, Mercury Park,Wycombe Lane, Wooburn Green,High Wycombe HP10 0HHUnited Kingdom

UK and Others Tel: +44 (0) 1628 552550France Tel: 0800919505Germany Tel: 01803001334Fax: +44 (0) 1628 552585

[email protected]

[email protected]

Affymetrix Japan K.K.Mita NN Bldg. 16F4-1-23 Shiba Minato-ku, Tokyo 108-0014 Japan

Tel. 03-5730-8200Fax: 03-5730-8201

[email protected]

[email protected]

www.affymetrix.com


Chapter 2
Installation and Setup


Chapter 2

13


Installation and Setup

The Affymetrix® Expression Console™ software is a stand-alone application. It can be installed on computers that have GeneChip® Operating System (GCOS) software, Affymetrix GeneChip® Command Console™ (AGCC) software, or neither.

Software Requirements

The Expression Console software can be installed with the following operating systems:

• Microsoft Windows 2000 Professional with service pack 4.0 or higher

• Microsoft Windows XP with service pack 2.0 or higher

Minimum Hardware Recommendations

The minimum hardware recommendations are:

• Memory (RAM): 1 GB*

• Hard drive: 20 GB** (sufficient space should be available to meet user data requirements)

• Processor: 2.0 GHz Intel Pentium or higher

* Due to the large number of datapoints produced by the Exon Arrays, an additional 1GB of RAM is highly recommended.

**The larger file sizes associated with Exon Array data should be taken into account when calculating the necessary free space requirement.

Affymetrix recommends that if you are using GCOS CEL files, you

should use the Data Transfer Tool (DTT) provided by Affymetrix to

move the CEL files out of the GCOS directory.



Installation Instructions

To install the Expression Console software:

1. Go to www.affymetrix.com and download the software from the following location:Support / By Product / Software / Affymetrix® Expression Console™ Software

2. Unzip the downloaded software package.

3. Double-click on setup.exe to install the software.

4. Follow the directions provided by the installer.

5. The setup process installs the required Microsoft components, which includes the .NET 2.0 framework.

Setup Profile

The Affymetrix® Expression Console™ software is ready for configuration. Follow the steps below to configure the software for sample analysis.

Create a Profile

A profile is used as a method to group options and parameters so that those parameters can be used again. Once the software is started, the profile can only be changed when there is no study is open, by clicking

, selecting Edit → Change User Profile from the drop-down menu, or going to Toolbox → Configuration → Specify User Profile.

To open the software and setup a profile, perform the following steps:

If there is a previous version of Expression Console software

installed, the installation software prompts the user to remove it

before installing the new version.


Chapter 2 | Installation and Setup 15


1. Open the Expression Console™ application by selecting Start → Programs → Affymetrix → Expression Console.The Expression Console software window opens with the Profile Information dialog box displayed.

2. Type in a name for your profile and click OK.By entering a profile name, you retain the specific analysis settings and quality control thresholds for each profile entered.

A list of previously created profiles, if any, are found in the drop-down menu on the Profile Information dialog box.

Affymetrix® Expression Console™ software - Profile Information Window

You can select a different profile without terminating the program,

but the current study must be closed to open another profile.



Delete A Profile

The list of previously created profiles are found in the drop-down menu on the Profile Information dialog box.

To remove profiles no longer needed:

1. Close the study window, if open.

2. Select File → Utilities.

3. Select the User Profile Management tab (Figure 2.1)

4. Highlight the profile to be removed.

5. Click Delete.

Figure 2.1

User Profile Management Tab



Library Files

Library Files – Download Option

The Expression Console software requires information stored in library files (array types) to analyze the CEL files generated by GCOS or Affymetrix GeneChip® Command Console™ (AGCC) software. These files are available from Affymetrix and can be downloaded within the Expression Console application.When you click OK in the Profile Information window the first time (Figure 2.1), a dialog box opens asking you to direct the software to your GeneChip library files folder (Figure 2.2). You can select any location for the library files folder

You can select any location for the library files folder. However,

once you direct the software to the folder location, do not place any

library files in a subfolder. The Expression Console application

cannot find library files in a subfolder!

If the Affymetrix GeneChip® Command Console™ (AGCC) software

is installed on your system, the Expression Console™ application

defaults to the files in C:\Command_Console\Library.

If the Affymetrix GeneChip® Operating System software (GCOS) is

installed on your system, Affymetrix recommends that you do notselect the GCOS library file directory as the library file directory for

Expression Console, to avoid confusion.



Library files can be downloaded from NetAffx™.

1. Select File → Download Library Files. A dialog box opens requesting your account information for NetAffx (Figure 2.3).To obtain a NetAffx account, go to www.affymetrix.com, click Register at the top of the Affymetrix main page and follow the instructions.

Expression Console downloads library files from NetAffx for

analysis, but these are not registered with GCOS and are not

sufficient to scan arrays.

Figure 2.2

Browse For Folder window to locate GeneChip library files folder




2. Enter your registered email address and password.The NetAffx Library Files window opens (Figure 2.4). This window contains a complete list of Affymetrix library files that can be downloaded. Library files previously downloaded are marked currently installed.

Figure 2.3

NetAffx Account Information dialog box



3. Check the library files needed and click the Download button.The appropriate files are downloaded to the folder you previously indicated. The green status bar at the bottom of the window highlights during the download process.If the Abort button is selected, the download process stops with a message indicating which library files failed to download.

Figure 2.4

NetAffx Library Files window - downloading a checked library file



Library Files – Copying Files Manually

For computers that are not connected to the internet and therefore cannot take advantage of the library file download option, it is possible to manually copy the necessary files to the computer with Expression Console:

1. Create a folder on the computer to hold the library files for the Expression Console application.

2. Copy the necessary files from the CD or other removable media to the library file folder. In order for the default report controls to be identified for the array type, this should be done with the application closed.

A. For 3' Expression Arrays, only the .psi and .cdf files need to be copied to the directory.

B. For Exon Arrays, minimally the .clf, .pgf, .bgp, and .qcc files need to be copied to the directory.

When you create a new study and add CEL/CHP files that do not

have corresponding library files loaded in the correct folder, you will

be prompted to download the appropriate library files.

Figure 2.5

Status Window - indicating missing library file

Do not create subdirectories within the library file folder. The

Expression Console software does not look at subdirectories.



To continue, proceed to Chapter 3, Creating a Study.


Chapter 3
Creating a Study


Chapter 3

25


Creating a Study

This section describes how to create a study with probe level intensity files (CEL) and probe level summarization files (CHP) in the Affymetrix® Expression Console™ software.

Getting Started

To get started using the Affymetrix® Expression Console™ software, the user creates a study consisting of a collection of probe cell intensity files (CEL) and/or probe level summarization files (CHP) and their associated sample information for analysis and examination.

The study table contains the following:

• File names (CEL or CHP)

• Sample attribute information from either the .xml files for GCOS (requires the use of DTT to transfer the data from the GCOS database) or .arr files (sample/array) from AGCC

• An indication of whether or not CHP files pass the user-defined tolerances for the array metrics. See Chapter 8, Controls and Thresholds.

• Whether or not the results are on a linear or log scale

By default, the algorithms PLIER and MAS 5.0 are linear; the RMA algorithm is in the log scale. Under the edit menu, the scale can be changed, but this does not change the scale of the data in the CHP file. See Appendix A, Algorithms.

Follow the steps below to create a study in the Expression Console software.

GCOS users must use DTT v1.1, using the Flat File option, to

transfer files to be analyzed by the Expression Console software

from the GCOS database to an independent folder. More detailed

instructions can be found at www.affymetrix.com; then go to

Support/Technical/Tutorial/GCOS.




Study Window

Follow the steps below to create a study to analyze CEL files or view CHP file data.

1. Create a study by first selecting one of the following:

- The icon

- File → New Study

- Toolbox → Study → Create New Study.

The Affymetrix Study window opens.

2. To add CEL files from your data set of interest to the study window, do one of the following:

For further analysis in an existing study, select , or File → OpenStudy, or Open → Existing Study from the Toolbox.

Figure 3.1

Affymetrix Study window

Chapter 3 | Creating a Study 27


- Click the Add Intensity Files button

- Select File → Add to Study → Probe Cell Intensity Files

- In the workflow toolbox, select Study → Add Intensity Files.

The Select Probe Cell Intensity Files window opens. Sample attributes from .arr or .xml files are displayed for the associated CEL files, if available.

Figure 3.2

Affymetrix Study window with CEL files



3. To add previously created CHP files to the study window , select the Add Summarization Files button, click File → Add to Study → Probe Level Summarization Files, or go to the workflow toolbox and select Study → Add Summarization Files.The Select Probe Level Summarization Files window opens and sample attributes from .chp files are displayed.

Figure 3.3

Select Probe Cell Intensity Files window

Use the Files of type menu selection at the bottom of the window

to filter the CHP files based on the selected algorithm type.



After CHP files are added to the study, either at the end of an analysis or directly by clicking the Add Summarization Files button, the metrics and controls associated with that algorithm run are compared against the user defined thresholds. Any files that fail to meet any one of the criteria are highlighted.

4. In the Select Probe Cell Intensity Files or Select Probe Level Summarization Files window, click the column heading of choice to sort the files based on that attribute.

Figure 3.4

Select Probe Level Summarization Files window

On a system with AGCC, the Expression Console software uses the

AGCC index to locate the matching ARR file; otherwise, it pairs the

CEL/CHP and ARR files by matching the root names.

On a system with GCOS, the Expression Console application will

use the sample attribute information from the .xml files, if the .xml

files are in the same folder as the CEL or CHP files and have the

same root name (e.g., 12345.xml and 12345.cel).



5. Select the files to be added and click Open to add the selected CEL/CHP files to the study window.

Library Files

To analyze the selected study files, the Expression Console application automatically locates the associated library files in the indicated folder, when the software is first installed. If the selected library path is not specified, Expression Console software defaults to the AGCC library files folder, if it is present.

If the correct library files have not been downloaded to the selected library file directory, the following error message displays when you attempt to analyze your files:

The status window at the bottom of the Expression Console application window lists the missing library files. If you need to establish a new path:

1. Select Edit → Set library path and locate the new directory.

2. Select the CEL files to analyze and click Open.The CEL files you selected are populated in the Affymetrix Study window with their accompanying attributes.

Figure 3.5

Error message - opening CEL files for analysis

All studies must be closed to activate the Set library path menu

selection.



Study Window Controls

The study window in the Expression Console application controls the files that are analyzed by the software. Within the study window the default sort order for the tables and graphs is determined. In addition, a sample attribute can be selected that will be pre-pended to the array names for display in the tables and graphs.

File Consolidation

A study can contain files that are located in multiple directories on the computer. To aid in file management, the Expression Console application can move all of the files associated with a study to a single directory. Studies cannot be consolidated when a study is open.


2. Select the File Consolidation Tab (Figure 3.7).

3. Browse to find the study to be consolidated.

4. Select an output directory to hold all of the associated files.

5. Click Consolidate and then Close.

Figure 3.6

Affymetrix Study window with CEL files



Analyzing Files

To analyze a 3' Expression Array data set, proceed to Chapter 4, 3’ Expression Array Analysis.

To analyze an Exon Array data set, proceed to Chapter 5, Exon Array Analysis.

Figure 3.7

File Consolidation Tab

File consolidation moves the files to the selected directory.


Chapter 4
3’ Expression Array Analysis


Chapter 4

35


3’ Expression Array Analysis

Probe cell intensity data from 3' Expression Array analyses is further analyzed in the Affymetrix® Expression Console™ software application using the MAS5, RMA, and PLIER algorithms to create CHP files. Follow the instructions below to analyze 3' Expression Array data.

Analysis Controls

Affymetrix microarrays contain the hybridization, labeling and housekeeping controls that help determine the success of the hybridizations. For more information on the interpretation of these controls, see the whitepaper, Data Analysis Fundamentals, at for the time www.affymetrix.com.

To aid in the examination of these controls, the Expression Console application displays summarized probe information in tabular format. In order for this feature to function properly, the controls must be identified prior to analysis. Once the controls have been identified for a particular GeneChip (probe array type), they are saved and only need to be updated if the user wishes to modify the controls.

For most standard Affymetrix GeneChip® Arrays, a standard set of defaults has been provided; however, it is recommended that the user should verify all controls before initiating their analysis. For details on defining the controls and their thresholds, refer to Chapter 8, Controls and Thresholds.

Analysis Algorithms

By default, the 3' Expression Array workflows are set as follows:

• PLIER workflow is Quantile normalization with PM-MM

• RMA workflow is Quantile normalization and has a general background correction

• MAS 5.0 workflow is set so that all probe sets are scaled to TGT = 500

For more details about the algorithms, see Appendix A, Algorithms and Appendix B, Algorithm Parameters and Outputs.




Analyze Data

To analyze 3' Expressson Array data:

1. Create a new study or open an existing study. See Chapter 3, Creating a Study.

2. In the study window, check the CEL files for Analysis.

3. To analyze the selected CEL files in the study, select Analysis → 3' Expression Arrays and select the appropriate workflow (MAS5, RMA, or PLIER) (Figure 4.1).

The suffix dialog box for the selected workflow opens (Figure 4.2).

You can also look at the controls alone to obtain a cursory look at

the success of the hybridization. This allows you to remove the

obvious outliers prior to the time investment of analyzing the full

array.

Figure 4.1

3’ IVT Expression Array - MAS5 Algorithm analysis

Chapter 4 | 3’ Expression Array Analysis 37


For advanced analysis, select Analysis → Advanced Expression Configuration and choose the appropriate Configuration or Mask file menu (Figure 4.3).

For details on how to utilize the advanced analysis features, see Chapter 9, Advanced Analysis.If you want to run a script (an automated analysis with preset parameters) to generate CHP files, select Analysis → Expression Analysis Scripts and create a new script or open an existing script (Figure 4.4). For details on how to set up an analysis script, see Chapter 9, Advanced Analysis.

Figure 4.2

Expression Console - MAS5 - File name suffix dialog box

Figure 4.3

Expression Console application - Advanced Analysis



4. Enter a suffix to identify the specific samples after analysis.

When you click OK, analysis begins. The progress of analysis is displayed in the Status window (Figure 4.5).

Results

Several minutes may be required for processing results, depending on the number of arrays selected. Once processing is complete, the study window shows a new group containing the newly summarized data (CHP files). The new group is named with the type of algorithm selected.

Figure 4.4

Expression Analysis scripts menu

Figure 4.5

MAS5 Analysis files in the Status window

Chapter 4 | 3’ Expression Array Analysis 39


The files are listed in the status window at the bottom of your screen as they are generated, and the Affymetrix Study window (Figure 4.6) is populated with the corresponding CHP files.

Highlighted CHP files (Figure 4.7) contain metrics outside of defined thresholds.

Figure 4.6

Affymetrix Study window - MAS5 analysis results

Figure 4.7

Affymetrix Study window - MAS5 analysis results.


Chapter 5
Exon Array Analysis


Chapter 5

43


Exon Array Analysis

Probe cell intensity data (CEL) from Affymetrix GeneChip® Exon Arrays are analyzed in the Affymetrix® Expression Console™ software. The application uses the RMA-sketch workflow for both Exon and Gene level analyses to create CHP files. Follow the instructions below to analyze Exon Array data.

Affymetrix® Exon Arrays include the following:

• Human – HuEx-1_0-st-v2

• Mouse – MoEx-1_0-st-v1

• Rat – RaEx-1_0-st-v1

Specify Controls

To analyze data from the Affymetrix Exon Arrays:

1. Download the appropriate library files, File → Download Library Files, (for example, human Exon Arrays require HuEx-1_0-st-v2 library files).

2. Create a new study or open an existing study. See Chapter 3, Creating a Study.

3. Add the CEL files you want to use to create CHP files by clicking the Add Intensity Files button.

4. Select the level of analysis by clicking the Run Analysis button in the Affymetrix Study window to see the Available Analyses menu box or select Analysis to display a drop-down menu (Figure 5.1). Then perform one of the following:

To use custom library files for an analysis, see Chapter 9, AdvancedAnalysis.



- For the Gene level, select Analysis → Gene Level (Figure 5.2), or select the Gene Level from the drop-down menu (Figure 5.1). Then select the annotation level, Core or Extended. These describe transcript structure confidence levels (for the annotation used to define the transcript) that are annotations broadly defined as follows:

– Core: limits analysis to exon-level probe sets that map to BLAT alignments of mRNA with annotated full-length CDS regions.– Extended: limits analysis to transcripts that are defined by exon-level probe sets that map to cDNA alignments and their annotations based on cDNA alignments, in addition to transcripts defined by the core category.– Full: analysis includes transcripts that consist of exon-level probe sets that map to sets of ab-initio gene predictions in addition to transcripts defined by the core and extended categories.For further explanation of confidence levels, see the Affymetrix whitepaper, Exon Probeset Annotations and Transcript Cluster Groupings, at www.affymetrix.com; then go to Support/Technical Documentation/White Papers.

Figure 5.1

Select the Run Analysis button in the Affymetrix Study window


Chapter 5 | Exon Array Analysis 45


- For the Exon level, select Analysis → Exon Level, or select one of the exon level analysis options from the drop-down menu (Figure 5.2). Then select the confidence level, Core, Extended, or All. These are gene confidence levels, which are input transcript annotations broadly defined as follows:

– Core: limits analysis to exons that consist of BLAT alignments of mRNA with annotated full-length CDS regions.– Extended: limits analysis to cDNA alignments and their annotations based on cDNA alignments, in addition to core defined exons.– Full: analysis uses exons derived from sets of ab-initio gene predictions in addition to the core and extended exons.– All: consists of all three confidence levels plus probe sets that map to more than one gene or to genes that do not align to the genome.For further explanation of confidence levels, see the Affymetrix whitepaper, Exon Probeset Annotations and Transcript Cluster Groupings, at www.affymetrix.com; then go to Support/Technical Documentation/White Papers.

- For Controls, select Analysis → Controls Only → Controls (Figure 5.3).

A subset of probes, on all arrays, are identified as controls. This option summarizes expression values for controls only.

Figure 5.2

Gene Level probe set with confidence modes: Core, Extended, Full




The RMA-sketch workflow is automatically selected by the software.

5. Click OK.The suffix dialog box for the selected array and confidence level opens.

You can look at the controls alone to obtain a cursory look at the

success of the hybridization. This allows you to remove the obvious

outliers prior to the time investment of analyzing the full array.

Figure 5.3

Exon analysis - Controls Only

Figure 5.4

Available Analyses - Controls Only - Controls



6. Enter a suffix to identify the specific samples after analysis.

7. Click OK; analysis begins.

The progress of the analysis can be tracked in the status window (Figure 5.6) (Figure 5.7). Since Exon Arrays are processed as a batch, the CHP files are not added to the study until the completion of the analysis. The files are listed in the status window as they are generated; then the Affymetrix Study window (Figure 5.8) is populated with the corresponding CHP files.

Figure 5.5

Expression Console - Exon Core - File name suffix dialog box

You can add a suffix to the file name in order to further identify your

samples, or you can leave it blank. The summarization method is

automatically included in the file name.

The processing time for the CEL files depends on a number of

factors, which include the number of CEL files, level of analysis,

number of probes under consideration, amount of available RAM,

and computer processor speed. See Chapter 2, Minimum HardwareRecommendations.



Figure 5.6

Exon Core level analysis files in the Status window

Figure 5.7

Gene Core level analysis files in the Status window

Figure 5.8

Affymetrix Study window - Gene core level analysis results



The CHP files are ready for QC analysis within the Expression Console software or other compatible software. CHP files are located in the same directory as the first CEL files in your input list. Outliers, samples outside the threshold boundaries, are highlighted in orange (Figure 5.9).

Figure 5.9

Affymetrix Study Window - Gene core level analysis results with outliers


Chapter 6
QC Tables and Graphs


Chapter 6

53


QC Tables and Graphs

To help researchers establish quality control processes for gene expression analyses, Affymetrix has developed several controls. Researchers are encouraged to monitor these controls on a regular basis to assess assay data quality. Many of these control metrics are generated as a product of the primary analysis by the respective analysis algorithms. To help with monitoring these values, a collection of tools for viewing and graphing metrics associated with the array and the analysis are provided in the Affymetrix® Expression Console™ software. These metrics include but are not limited to:

• Hybridization controls

• Labeling controls

• Internal control genes (Housekeeping controls)

• Global array metrics

• Algorithm parameters

This section describes how to generate graphs and tables allowing the quality of individual hybridizations included in a single study to be easily assessed and gives some introductory guidance on their interpretation. For more detailed information about graph and table interpretation, refer to the whitepapers, Data Analysis Fundamentals and Exon Arrays Quality Analysis, at www.affymetrix.com.

In general, Affymetrix highly encourages users to create a running log of the parameters to monitor quality and potentially flag outlier samples. Evaluation of particular samples should be based on the examination of all sample and array performance metrics in light of this history. A good, general rule for the examination of these types of quality control data is to look for outliers when compared to other highly related samples. For example, tissue A may have an overall low level of gene expression so that the percent of probes detected may normally be between 10–15%. Therefore, a %Present = 11 would not be an indication of a problem. However, if tissue B normally has a %Present between 45–50, and one sample has %Present = 11, that is an indication of a problem. Examination of which metrics or controls are outliers will provide insight into the source of the problem and possible solutions.




The Expression Console software has the ability to store user defined thresholds for the QC metrics and highlight, in the reports and study tables, those metrics outside the thresholds and arrays that contain them. Expression Console software comes with an initial set of defaults for both metrics and thresholds, which the user can modify according to individual preferences. For more information, see Chapter 9, Advanced Analysis.

Reports

Reports available in the Expression Console software provide a tabular view of all QC metrics, including:

• Algorithm parameters (See Appendix B, Algorithm Parameters and Outputs for a list of parameters, outputs, and their definitions.)

• Global analysis metrics

• Corner +, Corner -, Center +, and Center -, if available

• Count and percentages of detection calls

• Average signal value for each detection call type

• Signals, detections, and 3'/5' ratio for spike and housekeeping controls

Note that these metrics are algorithm specific. For details about the MAS 5.0 algorithm parameters, see the GeneChip® Operating System Software (GCOS) manual (PN 701439) and the Statistical Algorithms Reference Guide. For details about the PLIER algorithm parameters, see the Guide to Probe Logarithmic Intensity Error (PLIER) Estimation technical note and the Quality Assessment of Exon Arrays whitepaper at www.affymetrix.com.

To view the full set of report metrics, select Report → View Full Report.

http://www.adobe.com

Chapter 6 | QC Tables and Graphs 55


Full Report

A Full Report (Report → View Full Report) displays a table of the chosen algorithm parameters and the resulting array QC metrics contained within the CHP files. The Threshold Test column displays whether or not all the metrics are within or one or more metrics are outside of the user-specified thresholds.

Figure 6.1

Full Report - 3' Expression Array with MAS5



Custom Report

Customization of the report to define a subset of columns (report metrics) for the tabular display is possible by selecting Report → New Report and checking the categories you want to display in the table.

The list of available metrics to plot is based on the CHP files in the

study. For example, if you have MAS 5.0 CHP files open, then

parameters associated with PLIER are not visible.

Figure 6.2

New Report - 3' Expression Array with MAS5



Delete A Custom Report

To delete a Custom Report:

1. Click File → Utilities.

2. Select the Parameter Management tab (Figure 6.3).

3. In the Custom Reports section, select the report to be deleted.

4. Select Delete and then Close.

The categories available for selection in the Report window depend

on the expression array selected and the algorithms used to

generate the CHP files in the study.

Figure 6.3

Parameter Management Tab



Graphs

In addition to the tabular display of the metrics, the Expression Console software supports several different graphical views, including signal histograms, box plots, line graphs, and heat maps depicting the correlation between arrays. The graphical displays, especially the box plots and heat maps, are useful in identifying outlier samples.

Probe Cell Intensity View

The Probe Cell Intensity View (Graph → Probe Cell Intensity View) is a view of the CEL intensities arranged by physical position on the array. This view is used to quickly determine if any features on the array do not have a signal. For in-depth image inspection and gridding evaluation, it is recommended that customers use GCOS or the manual grid application available within the Affymetrix GeneChip Command Console (AGCC) software package (Figure 6.5).

Figure 6.4

Graph menu items



The display area can be adjusted using the arrow keys, while the zoom level is controlled by the + and - buttons. Moving the mouse over a cell in the image displays the features XY coordinate and intensity values. Selecting a region (click and drag the mouse) zooms the display to show the selected region. Right clicking on the image enables the following options:

• Select a gray scale palette (Gray)

• Select a color scale palette that goes from black to red to yellow to white (Heat)

• Adjust the range for the color scale (Scale)

• Save the image in PNG format

• Copy the image to the clipboard (Copy to clipboard)

Figure 6.5

Probe Cell Intensity View

A particular file may appear solid black when you first open the

Probe Cell Intensity View. Use the View → Zoom feature to zoom

in on a particular area of investigation.



Signal Histogram

The signal histogram (Graph → Signal Histogram) is enabled when the CHP files are selected in the study window (Figure 6.6). This graph displays a histogram plot of the signal values for the selected CHP files in a box or line graph. A legend of the input CHP files is displayed in the upper right corner of the histogram.

A right click popup menu provides the following options for the user:

• Save the graphic as a PNG file (Save as PNG)

• Save the underlying data used from the histogram to a TXT file. (Save data as TXT)


• Use bars in the histogram to display the values (Bars only)

• Use lines in the histogram to display the values (Lines only)

• Use bars and lines to display the values (Bars + Lines)

• Edit the scale of the Y axis (Scale)

Figure 6.6

Signal Histogram - 3' IVT with MAS5 algorithm



A slider just below the histogram allows you to adjust the zoom factor of the X axis and a scroll bar allows you to adjust the visible region of the graph.

Box Plot

The box plot graph shows a standard box plot of data from either CHP or CEL files. A general rule to use when examining box plots is to look for individual arrays that are dramatically different from the others and most importantly from other replicates in the same group.

A right-click popup menu is provided for all of the box plots containing:

• Save the graphic as a PNG file (Save as PNG)

• Save the data to a TXT file (Save data as TXT)


A slider is provided on the X axis to adjust the zoom factor. This is the number of files to display on the X axis.

Box Plot – Probe Cell Intensity

The probe cell intensity box plots are generated from CEL file probe cell intensity values. This graph is generated by selecting Graph → Box Plot - Probe Cell Intensity. The probe cell intensity creates a box plot of the probe intensity values for each array. Probe cell intensities are prior to analysis/summarization and have not been normalized; therefore, some differences in the distributions are to be expected. At the feature intensity level, the Exon Arrays have about 6 million probes; some delays in the generation of these graphs are expected.



Relative Probe Cell Intensity

The distribution of the ratio of the intensity of each probe to the median probe intensity across all of the selected arrays is summarized in the Box Plot - Relative Probe Cell Intensity. Therefore, the plot compares the distribution of intensities on each array to the median probe intensity value for the group. As such, it is a good way to identify arrays with divergent probe intensity distributions relative to the other arrays in the study.

Figure 6.7

Box Plot - Probe Cell Intensity - 3' IVT Expression Array



Figure 6.8

Box Plot - Relative Probe Cell Intensity - 3' IVT Expression Array

Figure 6.9

Box Plot - Relative Probe Cell Intensity - Exon Array



Box Plot – Signal

The Signal Box Plot is generated from summarized probe set signal values in CHP files. This graph is generated by selecting the Graph → Box Plot - Signal menu item.

Figure 6.10

Box Plot - Signal - 3' IVT Expression Array

Figure 6.11

Box Plot - Signal - Exon Array



Box Plot – Relative Signal

The Relative Signal Box Plot summarizes the distribution of the ratio of signal for each probe set to the median probe set signal across all the selected arrays. Therefore, the plot compares the distribution of probe set Signal values on each array to the median array for the group. As a result, it is a good way to identify arrays with divergent signal distributions relative to the other arrays in the study.

MvA Plot

The MvA plot (Figure 6.13) is a comparison plot comparing M (magnitude of change) on the Y-axis versus A on the X-axis, where M = log(Signal array1) - log(Signal array2) and A (average log(Signal)). The Y-axis is displayed on a log2 scale with green threshold lines for +/- two-fold changes. The color coding of the plot indicates the density of probes represented by that data point. The following are displayed:

Figure 6.12

Box Plot - Relative Signal - 3' IVT Expression Array



• The title of the graph, which indicates the two CHP files used to create the MvA plot

• The Pearson’s correlation (r2) value

Correlations

The relationship between two variables is described by their correlation. Two standard statistical measures of linear correlation (Pearson’s and Spearman) are provided in the Expression Console software. These are used to compare the signal estimates or detection p-values between two arrays. To simplify the interpretation of this data, the correlation values are presented as a heat map (Figure 6.14).

Figure 6.13

MvA Plot - 3' Expression Array with MAS5 algorithm

The Spearman test is used with fewer samples since it is a rank-

based test. Pearson’s test is used when there are more data, which

allows normalcy to be confirmed.



For all of the heat maps, a right-click popup menu enables the following:

• The graphic to be saved in PNG format (Save as PNG)

• The underlying r2 values to be saved as a TXT file (Save data as TXT)

• The image to be copied to the clipboard (Copy to clipboard)

• A dialog box to adjust the min/max values to use to map values to a color scale (Edit Scale)

In addition, a slider is provided on either axis to adjust the zoom factor controlling the number of files to show on a given axis. A scroll bar is provided to adjust the visible region of the graph. Clicking on a cell in the graph displays the associated files in the lower left part of the window. The display includes the file names and the value.

Pearson’s Correlation (Signal)

Signal concordance is evaluated using the Pearson’s correlation coefficient (r2) value to compare the signal values in two CHP files. The heat map contains a pairwise comparison of the signal values from all the selected CHP files, where the r2 values have been converted into a pseudocolor scale.



Pearson’s Correlation (Detection P-Value)

P-value concordance is evaluated using the Pearson’s correlation coefficient (r2) to compare the p-values from two CHP files. The resulting heat map contains a pairwise comparison of the detection p-values from all of the selected CHP files, where the r2 values have been converted into a pseudocolor scale.

Figure 6.14

Pearson’s Correlation (signal) - 3' IVT Expression Array

Only MAS5 and Exon level analyses calculate detection p-values to

generate the graph associated with them.



Spearman Rank Correlation (Signal)

This is the Spearman rank correlation of the signal values between two CHP files. The heat map contains a pairwise comparison of the signal values from all the selected CHP files, where the Spearman r2 values have been converted into a pseudocolor scale.

Figure 6.15

Pearson’s Correlation (detection p-value) - 3' IVT Expression Array



Spearman Rank Correlation (Detection P-Value)

This is the Spearman rank correlation of the p-values between two CHP files. P-values are only available in MAS5 CHP files and Exon Arrays analyzed at the exon level. The heat map contains a pairwise comparison of the detection p-values from all the selected CHP files, where the Spearman r2 values have been converted into a pseudocolor scale.

Figure 6.16

Spearman Rank Correlation (signal) - 3' IVT Expression Array

Only MAS5 and Exon level analyses calculate detection p-values to

generate the graph associated with them.



Line Graphs

Line graphs allow the user to graph metrics or signal values for specified groups of probe sets for selected CHP files.

Report Metrics

The report metrics is a line graph of the selected metrics (Graph → Line Graph - Report Metrics) calculated and stored in the header of the CHP. These include the algorithm parameters, global array metrics, and spike-in and housekeeping control values that are stored in the CHP file. The available metrics for graphing depend on the metrics available for the CHP files in the current study. Metrics not present in the CHP file cannot be graphed.

Figure 6.17

Spearman Rank Correlation (detection p-value) - 3' Expression Array



Probe List

The probe list is a line graph of probe sets defined in a probe list file). Select Graph → Line Graph - Probe List, the Select the probe list window opens (Figure 6.19). When you select the probe list of interest, Expression Console software creates a line graph of the probe sets (Figure 6.20).

Figure 6.18

Report Metrics - 3' Expression Array analyzed with MAS5



Figure 6.19

Select the probe list window

Figure 6.20

Probe List Line Graph - 3' Expression Array



New Probe List

The New Probe List window (Graph → New Probe List) enables users to create their own probe list by selecting the probe sets they want to investigate (Figure 6.23).

Figure 6.21

Probe List Line Graph - Exon Array



Figure 6.22

New Probe List - 3' Expression Array

Figure 6.23

New Probe List - Exon Array



Open Probe List

To open a probe list:

1. Select Graph → Open Probe List (Figure 6.24), and select the probe list.

2. Click Open.The Probe List of interest opens for viewing (Figure 6.25).

Figure 6.24

Select the probe list

Figure 6.25

View the Probe List



Delete a Probe List

To delete a probe list:



3. In the Probe List section, select the probe list to be deleted.

4. Click Delete and then Close.

Figure 6.26

Parameter Management Tab - Delete a Probe List


Chapter 7
Exporting Data


Chapter 7

81


Exporting Data

The Affymetrix® Expression Console™ software enables the user to export the data in tables and graphs to various text or pdf documents.

Exporting

Tables and Graphs to PDF

All of the currently opened/displayed tables and graphs can be exported to a PDF format. Select Export → Export All Tables/Graphs To PDF and browse to the chosen folder.

Figure 7.1

Tables and graphs in PDF format



Probe Set Results (pivot table) to TXT

The Export → Export Probe Set Results (pivot table) to TXT menu item displays a table of probe set results from the CHP files in the study.

1. Select Edit → Probe Level Summarizations Report Options. The Probe Level Summarizations Report Options dialog box opens (Figure 7.2).

2. Select the options to be displayed in the probe set results (pivot table):- Signal

- Detection p-value (MAS5 and PLIER workflows)

- Detection (MAS5)

- #Pairs (MAS5)

- #Used Pairs (MAS5)

3. Select Export → Export Probe Set Results.

Figure 7.2

Probe Level Summarization Report Options

P-values are only available when using MAS5 summarization for 3'

Expression Arrays and Exon Arrays analyzed at the exon level. No

p-values are available for the gene level analysis irrespective of the

summarization method chosen.

Chapter 7 | Exporting Data 83


Figure 7.3

Export Probe Set Results (pivot table) to TXT



A Report to GCOS RPT File

The Expression Console software is capable of creating GCOS formatted RPT files for MAS 5.0 CHP files. However, they cannot be generated by the other applications. The summarization workflows offered (i.e., RMA and PLIER-based workflows) are created with the Expression Console software, but not other applications. Use the Export → Export Report to GCOS RPT File menu item to save the CHP file information for all checked files in the current study to RPT files. Each CHP file creates a RPT file in the same directory using the same base file name.

A Study to TXT File

To export the Expression Console study table to TXT, right-click on the study window and select Export Study To TXT.

Figure 7.4

Export Report to GCOS RPT File

Figure 7.5

Export Study to TXT

Chapter 7 | Exporting Data 85


A Table as TXT File

After running an analysis, select Report → View Full Report (or alternatively, a user customized report selection). When the report displays (Figure 7.6), right-click on the report and select Export Table As TXT.

Figure 7.6

Export table as text



Saving Images

Images of the graphs and plots in Expression Console can be exported for use in other applications two ways:

• The image can be copied to the clipboard by right-clicking the image and selecting Copy to clipboard.

• Save the image as PNG by right-clicking on the graph and selecting Save As PNG (Figure 7.7) or select File → Save As PNG. The application prompts the user for a name and location to save the file.

The Expression Console software allows you to save a graph as a PNG (Portable Network Graphic), which is an image format that enables the graphic to be easily moved into another application or document without loss of quality.

Figure 7.7

Save the histogram as a PNG


Chapter 8
Controls and Thresholds


Chapter 8

89


Controls and Thresholds

This section describes how to identify and modify the controls for 3' Expression Arrays and how to set user-defined thresholds for both Exon and 3' Expression Arrays. (Currently, modification of the controls for Exon Arrays is not supported with the Expression Console™ software.) For guidance on how to set thresholds for individual metrics for 3' Expression Arrays, refer to the whitepaper Data Analysis Fundamentals, and for Exon Arrays, refer to the whitepaper Exon Arrays Quality Analysis.

Report Controls

To help researchers establish quality control processes for gene expression analyses, Affymetrix has developed several controls which allow researchers to monitor assay data quality. These include but are not limited to:

• hybridization controls

• labeling controls

• internal control genes

• algorithm parameters

• algorithm outputs

The Expression Console software provides functionality to display and highlight individual metrics and the arrays containing metrics outside of user-defined tolerances.

In general, Affymetrix highly encourages users to create a running log of these parameters to monitor quality and potentially flag outlier samples. The QC functionality built into the Expression Analysis Software can help with this. Evaluation of particular samples should be based on the examination of all sample and array performance metrics in light of the history of the metrics performance in an individual tissue and array type.

When analysis results are viewed within the software, the metrics for the report controls are compared to a set of user definable thresholds. Any results identified as being outside of the selected thresholds are tagged as Outside of Bounds.



For 3' Expression Arrays, the Report Controls window (Figure 8.1) enables the user to modify the group of probe sets that constitute the spike and housekeeping controls. The probe array type is selected from the drop-down menu at the top of the window to modify the control probes for that probe array type only. The list of control probe sets on the array are displayed on the left side of the window. By default, the probe sets are filtered to show only the AFFX controls using the Filter button at the bottom. The filter is case sensitive.

In order for the report to contain control information, the controls

must be defined before the analysis is run. If additional or different

controls are required, the analysis must be rerun following

redefining the controls.

Figure 8.1

3' IVT Expression Array Report Controls window

Chapter 8 | Controls and Thresholds 91


Spike Controls

The 20x Eukaryotic Hybridization Controls are spiked into the hybridization cocktail, independent of RNA sample preparation, and are therefore used to evaluate sample hybridization efficiency on gene expression arrays. The default spike controls are listed as:

• AFFX-r2-Ec-BioB

• AFFX-r2-Ec-BioC

• AFFX-r2-Ec-BioD

• AFFX-r2-P1-Cre

Internal Control Genes (Housekeeping Genes)

Internal control genes, or housekeeping genes, are gene transcripts that are constitutively expressed in most samples. These transcripts serve as internal controls, are useful for monitoring the quality of the starting sample, and are subject to any variability in the labeling of the sample and hybridization to the array, for 3' Expression Arrays. For Human, Mouse, and Rat 3' Expression Array types, β-actin and GAPDH are used to assess RNA sample and assay quality. Specifically, the signal values of the 3' probe sets for actin and GAPDH are most informative and, therefore, as a general recommendation, these should be compared to the signal values of the corresponding 5' probe sets. The ratio of the 3' probe set to the 5' probe set should generally be less than 3 for the One Cycle Labeling protocol. For more details on interpreting the housekeeping genes, see the whitepaper, Data Analysis Fundamentals at www.affymetrix.com. The Housekeeping controls are:

• GAPDH

• β-Actin

BioB is at the level of assay sensitivity (1:100,000 complexity ratio)

and should be called Present at least 70% of the time. BioC, BioD,

and cre should always be called Present with increasing signal

values.




Labeling Controls

Poly-A RNA controls can be used to monitor the entire target labeling process. Each eukaryotic GeneChip probe array contains probe sets from several B. subtilis genes that are absent in eukaryotic samples (lys, phe, thr, and dap). These Poly-A RNA controls are in vitro synthesized, and the polyadenylated transcripts for the B. subtilis genes are premixed at staggered concentrations. The Poly-A controls can be spiked into a complex RNA sample, carried through the sample preparation process, and evaluated like the internal control genes. The GeneChip® Poly-A RNA Control Kit (P/N 900433) contains the following four exogenous, premixed control spikes:

• Lys: AFFX-r2-Bs-lys (1:100,000)

• Phe: AFFX-r2-Bs-phe (1:50,000)

• Thr: AFFX-r2-Bs-thr (1:25,000)

• Dap: AFFX-r2-Bs-dap (1:6,667)

Identify / Remove Controls

Most 3' Expression Array types have a set of controls identified by default. Follow the steps below to change or set the report controls for the specified array type.

1. Select , or Edit → 3’ Expression Report Controls.The Report Controls dialog box opens (Figure 8.2).

2. Choose the array type from the Select Array Type drop-down menu at the top of the dialog box.

Control probe set names are unique to each array design.

All of the Poly-A controls should be called Present with increasing

signal values in the order of lys, phe, thr, dap.



3. In the probeset box , located on the left side of the window, highlight a probe set ID and select the 5'> to make that probe set the 5' probe, M> the middle probe, or 3'> the 3' probe.The upper box is used to define the spike-in control probe set IDs and the lower box the housekeeping control probe set IDs. For each probe set ID, a minimum of two probesets must be identified (for example, a 3' and a 5').

4. After you identify the 3', M, or 5' probe sets, select the >> button to load it into the control list.

5. Add or remove probe set IDs by clicking on the appropriate button or .Control metrics for the probe set IDs contained in the two boxes of the right hand side of the window will be included in the report for any subsequent analyses performed.

Figure 8.2

3’ Expression Report Controls dialog box



Delete Report Controls

To delete all of the controls for an array type:



3. In the Report Controls section, select the array type.

4. Click Delete to remove the controls for that array type.

To view all available probesets, clear the text box next to the Filter

button, then click the Filter button. A list of all the probesets on the

array are displayed.

Or, to view specific probeset controls, type the identifying letters

and/or numbers in the text field and click the Filter button.

Figure 8.3

Parameter Management Tab - Delete Report Thresholds



Report Thresholds

The report thresholds (Figure 8.4) define boundary conditions against a set of metrics that are computed during the analysis step. These metrics include report controls and statistics based on probe set Signal and Detection calls.

Fixed thresholds are applied to selected metrics in one of three ways:

• Compare metric 1 to threshold value – Example, compare the Percent Present calls to a fixed value to confirm that the metric passes a minimum or maximum value.

• Compare metric 1 to metric 2 – Example, compare two items, such as the average signal value for the Cre control against the average signal value for the bioD control to confirm that the detected signals are consistent with the known relative abundance in the sample.

• Compare metric 1 to average of metric 1 across arrays by standard deviation – Example, compare the spread of values such as scaling factors across all selected arrays of the same type. The calculation uses the mean and standard deviation to describe the spread of the data being compared. The final in-bounds range is then determined as the mean plus and minus a user defined factor times the standard deviation times the user selected multiplier.



Define / Modify Report Thresholds

The Report Thresholds listed are default settings and can be used without further modification on the part of the user. Report thresholds can be changed at anytime and the tables will be dynamically updated to reflect the changes once they have been saved.

To set report thresholds:

1. Select Edit → Report Thresholds, , or Toolbox → Configurations → Specify report thresholds to open the window.

Figure 8.4

Report Thresholds window

If a desired metric is not present in the report for a 3' Expression

Array type, the 3' Expression Controls must first be modified and

then the analysis must be run.



2. Select the comparison type:- Compare metric 1 to threshold value

- Compare metric 1 to metric 2

- Compare metric 1 to average of metric 1 across arrays by standard deviation

3. Select the comparison operator – less than, greater than, or equal to.

4. Select the item/value/range multiplier to compare to. The availability of this option is dependent on the comparison type you selected.

5. To delete a threshold item, click in the box so that it is highlighted in yellow, then click the Remove button.

Delete Report Thresholds

To delete all of the thresholds for an array type:



3. In the Report Thresholds section, select the array type.

4. Click Delete to remove the report thresholds for that array type.

To set thresholds for algorithm metrics, a CHP file analyzed with

that algorithm must be loaded into the study window.

If a CHP is not selected, the software is unable to determine the

metrics that are available for thresholding; therefore, one cannot

add thresholds for new metrics without first selecting a CHP file.



Figure 8.5

Parameter Management Tab - Delete Report Thresholds


Chapter 9
Advanced Analysis


Chapter 9

101


Advanced Analysis

Each analysis workflow provided in the Expression Console™ application has several parameters, which can be changed from the defaults provided by Affymetrix. This section describes how to change the parameters and save them to a configuration file. The file is then saved to the user profile for use in later analyses. Since Exon and 3' Expression Arrays have different algorithm options available, each array type has its own custom analysis configuration setup dialog.

3' Expression Array Configurations

This section describes how to set the algorithm parameters for MAS 5.0, PLIER, and RMA for use with 3' Expression Arrays. For additional information about setting MAS 5.0 parameters, see the white paper, Data Analysis Fundamentals and the Statistical Algorithms Reference Guide, on the Affymetrix web site. For explanations of the different normalization and background correction strategies for PLIER and RMA, see Appendix A, Algorithms.

Affymetrix Power Tools (www.affymetrix.com → Support/Developers’ Network/Analysis/Affymetrix Power Tools) is a

command line application that provides the ability to further

customize algorithm parameters and is intended for use by more

experienced users.

Affymetrix Power Tools is also available as source code and binaries

for several different computing platforms. These can be found at

the same location on the Affymetrix web site.




New Configuration – MAS5

1. Select Analysis → Advanced Expression Configurations.The Custom Analysis Algorithms dialog box opens (Figure 9.1).

2. Select the MAS5 algorithm and click OK.The New MAS5 Parameter settings dialog box opens (Figure 9.2).

3. Select the settings and click Save.

Figure 9.1

Advanced Analysis - MAS5 Custom Analysis Algorithms dialog box

Figure 9.2

New MAS5 Parameter Settings dialog box

Chapter 9 | Advanced Analysis 103


4. To change the detection parameters or mask out a subset of probes, click the Advanced button (Figure 9.2).The window expands to include Detection Parameters (Figure 9.3).

5. Enter new parameters and click Save.The Save window opens.If you select the Scale to select sets radio button, the Probe Set MSK file field is highlighted (Figure 9.4). Use the browse button

to find the appropriate MSK file.

Figure 9.3

New MAS5 Parameter Settings - Advanced window

Figure 9.4 Scale to select sets



6. Enter a name for this collection of parameters (Figure 9.5), and click Save.

7. To analyze array data with the new parameters, select Analysis → Execute Advanced Configuration, and select the appropriate file name (Figure 9.6).

To create a new MSK file, see Create a New Mask (MSK) File, on page 109.

Figure 9.5

Name the new configuration



A dialog box opens to announce analysis initiation (Figure 9.7).

8. Enter a suffix for run identification, if needed, and click OK to start the analysis.

Figure 9.6

Execute Advanced Configurations

Figure 9.7

Advanced Analysis Suffix Identifier



New Configuration – PLIER Workflow

1. Select Analysis → Advanced Expression Configurations.

The Custom Analysis Algorithms dialog box opens (Figure 9.8).

2. Select the PLIER workflow and click OK.The New custom workflow for PLIER dialog box opens (Figure 9.9).

3. Select the settings and click Save.A dialog box opens to announce analysis initiation (Figure 9.10).

Figure 9.8

Advanced Analysis - PLIER Custom Analysis Algorithms dialog box

Figure 9.9

Advanced Configuration for PLIER




New Configuration – RMA

1. Select Analysis → Advanced Expression Configurations.The Custom Analysis Algorithms dialog box opens

2. Select the RMA algorithm and click OK.The New custom configuration for RMA dialog box opens (Figure 9.11).

Figure 9.10


Figure 9.11

Advanced Configuration for RMA



The RMA Background Correction field is automatically applied to the PM-only probes when selecting RMA.

3. Select the settings and click Save.A dialog box opens to announce analysis initiation (Figure 9.12).


Delete an Advanced Configuration

To delete an advanced configuration:



3. In the appropriate Configuration section, select the configuration to be deleted.

4. Click Delete.

Since RMA has its own background correction algorithm,

Expression Console software does not allow this to be modified.

Figure 9.12




MASK Files

Create a New Mask (MSK) File

The Expression Console application uses MSK files for either setting the scaling factor or excluding or masking certain user-selected probe pairs from an expression analysis. The Selected Probe Sets scaling option, in the advanced configuration for the MAS5 algorithm, adjusts the trimmed mean signal of the selected probe sets on a probe array to the user-specified target signal value. Expression Console software utilizes user-selected probe sets (specified by a Scale Factor mask file) to calculate the trimmed mean signal and derive the scale factor for the probe array so that Target Signal = Scale Factor x Trimmed Mean Signal of selected probe sets. For a selected set of arrays, predefined msk files are available from Affymetrix at www.affymetrix.com. The other use for mask files is to allow user-selected probe pairs to be excluded or masked from an expression analysis.

Figure 9.13

Parameter Management Tab - Delete Configuration




To create a MASK file:

1. Open Analysis → Advanced Expression Configurations and select New Mask (MSK) File (Figure 9.14).

The Select the Array Type dialog box opens.

For human, mouse, and rat, a list of 100 normalization genes is

provided to the user.

Figure 9.14

Advanced Expression Configurations - New Mask File



2. Select the array type and click Open.

A window opens displaying all probe sets for that array type (Figure 9.17).

Figure 9.15

Select the Array Type

Figure 9.16

New MASK (MSK) File dialog box



3. Double-click on the probe sets of interest to highlight each set. When you click Save, an MSK file of the selected probe sets is created.

4. Make changes if necessary and click Save, or click the box to exit.

Open Mask (MSK) File

Mask Files are specific for the probe array type. The Expression Console software will not open a mask file that is incompatible with the currently selected probe array type.

To view or edit an existing MSK file:

1. Select Analysis → Advanced Expression Configurations → Open Mask (MSK) File.The Select the MSK file dialog box opens.

Figure 9.17

Probe Sets for Expression Array types



2. Select the MSK file and click Open.The selected MSK file opens displaying the selected probe sets, which are highlighted.

3. Make changes if necessary and click Save, or click the box to exit.

MAS 5.0 CHP Files in the Command Console Format

By default, the Expression Console software stores the MAS5 analysis results (Signal, Detection, Detection P-Value, # of Probe Pairs, and # of Probe Pairs Used) in a CHP file using the new Affymetrix GeneChip® Command Console™ (AGCC) format. This new format provides additional features such as :

• An embedded unique file identifier

• A copy of the CEL and DAT headers (if the input CEL/DAT files are also in the AGCC file format)

Figure 9.18

Select the MSK File dialog box



The new file features provide the ability to trace the file’s lineage independent of the file name.To configure the software to write the MAS5 CHP files in the newer AGCC format:

• Select Analysis → Advanced Expression Configurations → Save MAS5 CHP files in Command Console format (Figure 9.19).

Figure 9.19

Save MAS5 CHP files in Command Console format



MAS 5.0 CHP Files in the GCOS Format

To support the transition of software applications to use the newer Affymetrix file parsers (those capable of reading both the older CHP file format as well as the newer AGCC CHP file format), the Expression Console software has the ability to write the MAS5 analysis results to the older GCOS format CHP file.

To configure the software to write MAS5 CHP files in the older GCOS format:

• Select Analysis → Advanced Expression Configurations → Save MAS5 CHP files in GCOS format (Figure 9.20).

For more information about the Command Console (AGCC) file format and C++/Java parsers that read both GCOS and Command Console files, see www.affymetrix.com; then go to Support/Developer/Fusion.

Figure 9.20

Save MAS5 CHP files in GCOS format

Analysis results from PLIER and RMA cannot be stored in GCOS

formatted CHP files.




Advanced Configuration Exon Analysis

1. Select Analysis → Advanced Exon Configurations. The Advanced Exon Setup dialog box opens (Figure 9.21).

A. Select the Analysis Type:

The setup box in the upper left-hand corner enables users to select their analysis level for the configuration. For details on the different analysis levels, see the whitepaper, Exon Probeset Annotations and Transcript Cluster Groupings, at www.affymetrix.com. In the Expression Console software, each level is paired with a default set of library files for that analysis and array type provided by Affymetrix. To use other versions of the probe group file, the intensity layout file, background probes, or the

Figure 9.21

Advanced Exon Setup dialog box




QC probeset file, browse to the desired file on the appropriate line of the library file section at the bottom of the screen. It is always possible to reload the default files for an analysis by clicking the Default Files button. To use a different meta-probeset file for Gene level analysis or probeset file for Exon analysis, select Other and browse to the file you wish to use in the library file section.

B. Select the Summarization Method. For more information about the summarization methods, see the algorithm details in Appendix A, Algorithms and the whitepapers, Gene Signal Estimates from Exon Arrays and Guide to Probe Logarithmic Intensity Error (PLIER) Estimation, at www.affymetrix.com.

C. Select the Background Correction. For more information about background correction, see the whitepaper, Exon Background Correction, at www.affymetrix.com.

D. Select the Normalization Method. For more information about normalization, see the whitepaper, Gene Signal Estimates from Exon Arrays, at www.affymetrix.com.

E. Confirm the library file selections in the bottom portion of the window. Note that selecting a confidence level in the Analysis Type automatically selects the appropriate file from Affymetrix.If you want to use your own custom meta-probeset file or probelist file, select Other in the analysis level to activate the browse feature for that line.

2. Click Save.






3. Enter a name for this configuration (Figure 9.22), and click Save.

4. To analyze array data with the new parameters,

A. Select Analysis → Execute Advanced Configuration.

B. Select the appropriate file name.

Figure 9.22

Name the new configuration



5. To delete an advanced configuration:

A. Select File → Utilities.

1) Select the Parameter Management tab (Figure 9.24).

2) In the appropriate Configuration section, select the configuration to be deleted.

Figure 9.23

Execute Advanced Configurations

Figure 9.24

Parameter Management Tab - Delete Configuration



3) Click Delete.

A dialog box opens to announce analysis initiation.

B. Enter a suffix for run identification, if needed, and click OK to start the analysis.

Figure 9.25



Chapter 1
0 Analysis Scripts


Chapter 1
0

123


Analysis Scripts

To streamline the analysis workflow within the Expression Console™ software, the user can create an analysis script. In the analysis script, the user selects and saves a specific analysis configuration and a set of graphs and tables that are automatically generated at the end of each analysis. Expression Console analysis scripts include the following features:

• Data table can be exported as a tab-delimited TXT file

• Selected graphs and tables can be exported as PDF files

• User-defined scripts are associated with a user profile

Analysis Script Creation

To create an analysis script:

1. Select Analysis → Exon Analysis Scripts → New Script.The New Script dialog box (Figure 10.1) displays.

Scripts are associated with the user profile open at the time the

script is created.

Analyses that have array-specific details contained within them,

such as custom MAS 5.0 and custom Exon analysis scripts, are

array-type specific.



2. Select the workflow from the drop-down menu. If a study is open, then the analysis methods in the list are restricted by the array type in the study.

3. Select the table and graphs to be generated after the completion of the analysis from the relevant sections.

4. If desired, select the probe set results table for a tab-delimited export of the expression data.

5. If desired, select the PDF option to save a PDF of the selected graphs and tables.

Figure 10.1

New Script dialog box

The application prompts the user to name the exported files at the

time of the analysis script execution.

Chapter 10 | Analysis Scripts 125


Analysis Script Deletion

To delete an analysis script:



3. In the Scripts section, select the script to be deleted.

4. Click Delete.

Figure 10.2

Parameter Management Tab - Delete Analysis Script



Analysis Script Execution

To execute an analysis script:

1. Open a study.

2. Select the CEL files for analysis.

3. Select Analysis → Execute Analysis Script.The Select script menu box display (Figure 10.3).

Figure 10.3

Select Script menu

If the script calls for any exports, the user is prompted for names

and locations for saving files.


Appendix A
Algorithms


Appendix A

129


Algorithms

This appendix briefly describes the MAS 5.0, RMA, and PLIER algorithms as they are used to analyze data in the Affymetrix® Expression Console™ software. References and links to various publications, which describe the MAS 5.0, RMA, and PLIER algorithms in detail, are given. See a Comparison of Algorithms for a side-by-side comparison of the assumptions, advantages and disadvantages of using each individual algorithm.

MAS 5.0 Algorithm

The MAS 5.0 algorithm uses the Tukey’s biweight estimator to provide a robust mean Signal value and the Wilcoxon’s rank test to calculate a significance or p-value and Detection call for each probe set. Background estimation is provided by a weighted average of the lowest 2% of the feature intensities. Mismatch probes are utilized to adjust the perfect match (PM) intensity. Linear scaling of the feature level intensity values, using the trimmed mean, is the default to make the means equal for all arrays being analyzed.

The MAS 5.0 algorithm (also known as the Statistical Algorithm) analyzes each array independently. As a result, individual probe-specific affinities can not be considered and the ability to detect small changes between experiment and control samples is reduced in comparison to either RMA or PLIER.

The primary use of the MAS 5.0 algorithm is to obtain a quick report regarding the performance of the arrays and to identify any obvious problems before submitting the final set of arrays to one of the multichip analysis methods (RMA, PLIER).

For a more detailed description of the MAS 5.0 algorithm, see the Statistical Algorithms Reference Guide at www.affymetrix.com; then go to Support/By Support Type/Technical Documentation/Technical Notes/Software.




RMA Algorithm

The Robust Multichip Analysis (RMA) algorithm fits a robust linear model at the probe level to minimize the effect of probe-specific affinity differences. This approach:

• Increases sensitivity to small changes between experiment and control samples.

• Minimizes variance across the dynamic range, but does compress calculated fold change values.

RMA consists of three steps:

1. Background adjustment

2. Quantile normalization

3. Summarization

This is a multi-chip analysis approach. Therefore, all arrays intended for comparison should be included together in the summarization step.

For a more detailed description of the RMA algorithm, see the publication, Exploration, Normalization, and Summaries of High Density Oligonucleotide Array Probe Level Data, Biostatistics, April 2003; Vol. 4; Number 2: 249–264.

Appendix A | Algorithms 131


PLIER Algorithm

The Probe Logarithmic Intensity Error Estimation (PLIER) algorithm method produces a signal by accounting for experimentally observed patterns in probe behavior and handling error at the appropriately low and high signal values.

Similar to RMA, the PLIER algorithm also utilizes data across all arrays submitted for analysis to minimize the effect of probe-specific affinity differences. Unlike RMA, which uses a simple global background correction approach, the standard PLIER configuration uses a probe-specific background providing a higher degree of accuracy at the cost of increased signal variance. Exon Arrays use a more advanced and efficient background correction method utilizing surrogate mismatched probes. (See the technical note, GeneChip® Exon Array Design at www.affymetrix.com; then go to Support/By Support Type/Technical Documentation/Technical Notes/Software). Signal variance can be addressed following the PLIER analysis by applying a variance-stabilizing data transformation. A simplistic but effective approach is to add a value of 16 to each and every signal value (PLIER + 16 algorithm).

This is a multi-chip analysis approach. Therefore, all arrays intended for comparison should be included together in the summarization step.

For a more detailed description of the PLIER algorithm, see Guide to Probe Logarithmic Intensity Error (PLIER) Estimation at www.affymetrix.com; then go to Support/By Support Type/Technical Documentation/Technical Notes/Software).





Comparison of Algorithms

Table A.1 A comparison of the MAS5, RMA, and PLIER algorithms used with the Expression Console™ application

Algorithm Advantages Disadvantages Assumptions

MAS5 Single-array algorithm is independent of other data in the dataset:

• Less computationally intensive than RMA or PLIER

• Conservative

• Smooth down�weighting of outliers

• Positive output values

• Minimal bias

Not as sensitive as either RMA or PLIER to small changes in target abundance:

• Limited ability to adjust for probe�specific affinity differences

• Unstable variance at low end

• Lower precision in signal calculation

• Single�array analysis

• Multiplicative error

• Signal is adjusted by the mis�match probes (PM–MM)

• Background imputed to handle negative differences

RMA Minimizes the variance seen across the arrays

• Higher reproducibility of signal over singlearray analyses

• Good differential change detection

• Variance stable on log scale

Not as sensitive as PLIER in the ability to detect small fold changes

• In cases where feature intensities disagree, may have more than one solution (mitigated by median pol�ish)

• Positive bias contributed to signal values

• Compresses fold changes for low intensity probe sets

• Multiple�array analysis

• Multiplicative error

• PM (perfect match) only

• Single background is used to adjust each intensity

PLIER Ability to detect small fold changes:

• Higher reproducibility of signal (lower coefficient of variation) with�out loss of accuracy

• Higher sensitivity to changes in abundance for targets near back�ground

• Dynamic weighting of the most informative probes in an experi�ment to determine signal

• High degree of accuracy for signal and fold change calculations

• Lack of bias

More variance in indivdual signals than seen with RMA

• Computationally intensive

• In cases where feature intensities disagree, may have more than one solution (mitigated by median polish)

• Performance relative to amount of model data provided

• Variance not stable on log scale

• Multiple�array analysis

• Mixed error model PM–MM, PM only, etc.

• Multiple background options

• Smoothly handles intensities below background


Appendix B
Algorithm Parameters and Outputs


Appendix B

135


Algorithm Parameters and Outputs

The Expression Console™ software displays column headings for individual parameters and outputs in the Report window (Report → View Full Report), for each of the 3' Expression Array algorithms (MAS 5.0, RMA, PLIER). This appendix gives column heading definitions.

MAS 5.0 Column Headings

Threshold Test The minimum number of probe pairs a probe set must have in order for the probe set data to be included in the calculation of the report statistics

HZ Number of horizontal zones used in background subtraction

VZ Number of vertical zones used in background subtraction

BG Minimum, maximum, average, and standard deviation of the background intensity calculated for the probe array

Alpha1 (α1) Significance level for the detection p-value in an analysis – Alpha1 is a user-modifiable parameter that is set in the New MAS5 Parameter Settings dialog box (Analysis → Advanced Expression Configurations → New Configuration).If the probe set detection p-value is < alpha1, the call is present.

Alpha2 (α2) Second significance level for the detection p-value in an analysis and a user-modifiable parameter set in the same dialog box as Alpha1 – If the probe set detection p-value is greater than or equal to alpha2, the call is absent. If alpha1 is less than or equal to the Detection p-value < alpha2, the call is marginal.

Tau (τ) A user-modifiable parameter, ideally set to a value that is a little larger than the median of the discrimination scores of the probe sets whose targets are absent, to avoid false detected calls

Noise (Raw Q) The degree of pixel-to-pixel variation among the probe cells used to calculate the background



Scale Factor (SF) The scale factor specified in the Scaling tab of the Expression Analysis Settings dialog box or computed by the algorithm

Norm Factor (NF) The normalization factor computed by the algorithm

Scale Mask Contains the name and location of the mask file if one was used for the analysis

RawQ Baseline Noise – The degree of pixel-to-pixel variation among the probe cells used to calculate the background in the baseline probe array

BG Avg The average background intensity calculated for the probe array

BG Std The standard deviation background intensity calculated for the probe array

BG Max The maximum background intensity calculated for the probe array

BG Min The minimum background intensity calculated for the probe array

Noise Avg The average noise calculated for the probe array

Noise Std The standard deviation noise calculated for the probe array

Noise Max The maximum noise calculated for the probe array

Noise Min The minimum noise calculated for the probe array

Corner+ Avg The average cell intensity for the sense probe cells used in the grid alignment process

Corner+ Count The corner count for the sense probe cells used in the grid alignment process

Corner- Avg The average cell intensity of the antisense probe cells used in the grid alignment process

Corner- Count The corner count of the antisense probe cells used in the grid alignment process

Central- Avg The average cell intensity for the nine probe cells that comprise the cross at the center of an antisense probe array

Central- Count The central count for the nine probe cells that comprise the cross at the center of an antisense probe array

Appendix B | Algorithm Parameters and Outputs 137


#Probe Sets Exceeding Probe Pair Threshold

The number of probe sets that exceed the probe pair threshold

Probe Pair Threshold The minimum number of probe pairs a probe set must have in order for the probe set data to be included in the calculation of the report statistics

Control Direction The direction (sense or antisense) of the target (sample)

#P The number of probe sets present

%P The percent of probe sets present

Signal(P) The average signal for the probe sets defined as present

#M The number of probe sets whose detection is marginal

%M The percent of probe sets whose detection call is marginal

Signal(M) The average signal for probe sets whose detection call is marginal

#A The number of probe sets absent

%A The percent of probe sets called present that are absent

Signal(A) The average signal for the probe sets defined as absent

Signal(All) The average signal for all probe sets on the array



Column Headings for RMA and PLIER

Metrics are dependent on the analysis algorithm, array type, and the level; therefore, not all of the metrics are always present in the Full Report.

Raw Corner+ Avg The average cell intensity prior to any background for the sense probe cells adjustment used in the grid alignment process

Raw Corner+ Count The count of the number of sense probe cells used in grid alignment

Raw Corner- Avg The average cell intensity prior to any background for the antisense probe cells adjustment used in the grid alignment process

Raw Corner- Count The count of the number of antisense probe cells used in grid alignment

Raw Central- Avg The average cell intensity before background adjustment for the sense probe cells that comprise the cross at the center of the array

Raw Central- Count The average cell intensity before background adjustment for the anti sense probe cells that comprise the cross at the center of the array

Spike-probeID-signal The signal for the probe sets that correspond to the labeling and hybridization spike controls

pm_mean The mean signal for all of the PM probes on the array

mm_mean The mean signal for all of the MM probes on the array

bgrd_mean The average signal for the probes used to calculate the background

Appendix B | Algorithm Parameters and Outputs 139


pos_vs_neg_auc The area under the curve (AUC) for areceiver operator curve (ROC) comparing the intron controls to the exon controls by applying a threshold to the probe set summary – The ROC curve is generated by evaluating how well the probe set summary separates the positive controls from the negative controls (e.g., exon from intron). The assumption (which is only valid in part) is that the negative controls are a measure of false positives and the positive controls are a measure of true positives. An AUC of 1 reflects perfect separation whereas as an AUC value of 0.5 would reflect no separation. Note that the AUC of the ROC curve is equivalent to a rank sum statistic used to test for differences in the center of two distributions.

Signal(A) The average signal for the probe sets defined as absent (Exon-level DABG > 0.01)


apt-opt-cdf-file The cdf file used for the analysis

apt-opt-probe-count The total number of probes (features) on the array whether or not the probes are used

apt-opt-qc-groups-file The file used to identify the control probe and probe sets for the analysis run

#P The number of probe sets present – For exon level analysis, the DABG p-value is less than or equal to 0.01.

%P The percent of probe sets present – For the exon level analysis, this is defined as the DABG probe level p-value is less than or equal to 0.01.

Signal(P) The average signal for the probe sets defined as present (Exon-level DABG <= 0.01)

#A The number of probe sets absent – For exon level analysis, the DABG p-value is greater than 0.01.

%A The percent of probe sets called present absent – For exon level analysis, the DABG p-value is greater than 0.01.

Signal(A) The average signal for the probe sets defined as absent (Exon-level DABG > 0.01)




Probe Set Suffixes

A number of standard metrics are created for particular groups of probe sets on the arrays. The key below describes the suffix and descriptions for the following groups of probe sets:

- all_probeset (all probe sets within a group)

- bac_spike (bacterial spikes and hybridization controls)

- polya_spike (labeling controls)

- neg_control (negative control probes)

- pos_control (positive control probes)

KEY: Suffix for Output Definition

_probesets From the QC Report, the number of probe sets actually analyzed

_atoms From the QC Report, the number of PM–MM or PM–GCBG probe pairs actually analyzed

_mean The mean signal value for all the probe sets

_stdev The standard deviation for all of the probe sets analyzed

_mad_residual_mean The mean absolute deviation of the residual for a chip versus all chips in the data set

_mad_residual_stdev The standard deviation of the residual for a chip versus all chips in the data set

_rle_mean The mean absolute relative log expression (RLE) – This metric is generated by taking the probe set summary for a given chip and calculating the difference in log base 2 from the median value of that probeset over all the chips. The mean is then computed from the absolute RLE for all the probe sets for a given CEL file.

_rle_stdev The standard deviation of the relative log expression (RLE) – This metric is generated by taking the probe set summary for a given chip and calculating the difference in log base 2 from the median value of that probeset over all the chips. The standard deviation is then computed from the absolute RLE for all the probe sets for a given CEL file.

_percent_called The percent of probe sets from the exon analysis called present (DABG <= 0.01)


Index

143


Index

Symbols% Present call 53

Numerics3’ Expression Array 3

configurations 101

Aadvanced configuration

delete 108

advanced expression configu-

ration 37

AFFX controls 90

Affymetrix Power Tools 101

Affymetrix technical support 10

AGCC software 13, 29

algorithm

comparisons 132parameters and outputs 135

algorithm parameters 53, 89

algorithms 35

analysis results 38

array metrics 25

Bbacterial spikes 140

b-Actin 91

box plot 61relative signal 64, 65signal 64

box plots 58

CCEL files 3

CHP files 3

confidence levels 44, 46

control

metrics 53, 93

probe set names 92

control probes

negative 140positive 140

controls 35, 89

correlation matrices 4

custom report

delete 57

DDABG 140

Data Transfer Tool 13

detection

calls 95p-values 66, 68

documentation

conventions used 7

DTT 25

EExon Array 3, 43

advanced configuration 116background correction 117human 43mouse 43normalization method 117rat 43summarization method 117

Exon level analysis 43all 45core 45extended 45full 45

exporting

data 81probe set results to TXT 82report to GCOS RPT file 84study to TXT file 84table as TXT file 85tables and graphs to PDF 81

expression analysis scripts 37

expression arrays 3

Expression Console™

advanced analysis 37, 101analysis controls 35analysis results 38configuration file 101controls only analysis 45create profile 14custom report 56data analysis 36delete profile 16detection 54Exon level analysis 43exporting data 81full report 55Gene level analysis 43graphs 58hardware requirements 13hybridization 46installation instructions 13library files 17report metrics 56, 71reports 54signal 54software requirements 13specify controls 43status window 38study window 26workflow 5

FFAQS 7

GGAPDH 91

GCOS

library file directory 17RPT files 84software 13, 29

gene expression 53

Gene level analysis 43core 43, 44extended 44full 44

GeneChip® 3



global array metrics 53

graphs 53

grid evaluation 58

Hheat maps 4, 58, 67

housekeeping controls 53, 71, 90

hybridization 3, 36controls 53

Iidentify controls 92

interanl control genes 91

internal control genes 89

Llabeling controls 53, 89

library files 30copying files manually 21custom 43default location 17download 17error message 30folder 17NetAffx 17set library path 30

line graphs 58, 71

linear correlations 66

linear scale 25

log scale 25

MMAS 5.0 25, 35, 101, 129

background estimation 129CHP files in AGCC format

113CHP files in GCOS format

115column headings 135configuration 102

parameter settings 102

MASK files

create 109

metrics 53

MSK file

new 110open 112scale factor 109

multi-chip analysis 131

MvA plot 4, 65

NNetAffx

account 18account information dialog

box 19library files 20

Ooutliers 3, 49, 58

outside of bounds 89

Pparameter management 94,

119

Pearson’s Correlation (detec-

tion p-value) 68

Pearson’s correlation (Signal) 67

Pearson’s test 66

PLIER 25, 35, 54, 101, 131column headings 138configurations 106relative log expression 140

PM–GCBG 140

PM–MM 140

PNG

format 59images 86

Poly-A RNA controls 92

probe cell intensity view 58

probe level

intensity files 25summarization files 25summarization report op-

tions 82

probe list 72

probe set suffixes 140

profile information dialog box 16

p-values 82concordance 68

QQC 3

analysis 49functionality 89metrics 54

Rrelative probe cell intensity 62

report

controls 89, 90define and modify thresh-

olds 96delete controls 94delete thresholds 97metrics 71thresholds 95

RLE 140

RMA 25, 35, 101, 130background correction algo-

rithm 108column headings 138configurations 107sketch workflow 46

Sscreen captures 8

signal

distributions 4estimates 66histogram 60

Index 145


Spearman Rank Correlation

detection p-value 70signal 69

Spearman test 66

spike controls 90, 91

study window

file consolidation 31sort order for tables and

graphs 31

suffix 47, 105

Ttechnical support 10

threshold 39, 89boundaries 49test 55

Uuser profile management tab

16

Affymetrix Expression Console Software v1.0 - User Guide

Documents

Transcript of Affymetrix Expression Console Software v1.0 - User Guide