Using S+ArrayAnalyzer to Improve Gene Expression Analysis and Deploy Best Practices

Speakers: Michael O'Connell, Ph.D., Director Life Science Solutions, Insightful Corporation & Richard Park, Computational Data Analyzer, Joslin Diabetes Center

Listen to the archived Web cast.

Downloads:

• Webcast presentation materials [6.8MB PDF]
• Scientific Computing Article: State-of-the-art Analytics Help Researchers Understand Type-1 Diabetes [15MB PDF]

Listen to the archived web cast and discover:

• How S+ArrayAnalyzer implements linear models and ANOVA for time course and multifactor experiments
• How Joslin Diabetes Research Center uses S+ArrayAnalyzer to deploy best practices in microarray analysis to its scientific researchers

Also Receive a:

• Demonstration of a microarray time course analysis and the associated workflows available in S+ArrayAnalyzer including:
  • data import,
    
  • probe-level analysis (e.g. RMA, GC-RMA),
    
  • QC and filtering,
    
  • differential expression (linear models),
    
  • clustering,
    
  • annotation and
    
  • gene list management
• Description of workflows for analysis of both low replicate and more complex multi-factor experiments
• Summary of how to increase your ROI on microarray experiments by applying rigorous and appropriate statistical analyses to different classes of microarray experiments

Abstract
Microarrays allow simultaneous studies of gene expression for thousands of genes under many different experimental conditions. As such, there is a need for 1-way and 2-way ANOVA models and models for analysis of saturated designs, to be fit rapidly to the thousands of genes in any given experiment. We describe the fitting of such models in the analysis of time course and multifactor experiments using S+ArrayAnalyzer.

The ANOVA models allow simple specification of contrasts between the time points and experimental factor levels. Summaries of differential expression on the time increments and between factor levels provide a lens through which one can view the underlying cellular events and changes in the cell phenotype. We present such summaries using volcano plots, parallel coordinate plots and annotation analysis. The plots are displayed as interactive S+Graphlets which allow convenient linking to metadata for annotation of the genes on the plots.

Genes are identified for detailed study in the increments analyses using p-values from the overall F-tests and contrast F-tests, in addition to fold change and other intuitive criteria. Such genes may then be summarized through partitioning cluster analysis methods and annotated as groups. Such grouping and group annotation provide convenient summaries of gene function across the time series.

We illustrate the analysis approach with a time course dataset from the Affymetrix murine chip mgu74av2. This analysis also describes the workflows available through S+ArrayAnalyzer including data import, probe-level analysis (e.g. RMA, GC-RMA), QC and filtering, differential expression, clustering, annotation and gene list management.

We also describe how S+ArrayAnalyzer can be deployed to single users and user communities in a real world setting at Joslin Diabetes Research Center. This includes a description of S+ArrayAnalyzer deployed through S-PLUS on the Windows desktop and through S-PLUS Server on Windows and UNIX server environments.

Presenter Information
Michael O'Connell, Ph.D., is director of Life Science Solutions at Insightful Corporation. He has more than 15 years experience in the medical device, informatics and health-care statistics arena, having published more than 30 papers on statistics, data mining and health-care applications. This has included statistical methods work in the areas of non-parametric regression, experimental design, calibration and mixed models; and applications such as DNA amplification, diagnostics, drug delivery and microarray data analysis.

Richard Park, computational data analyzer at Joslin Diabetes Center, builds and deploys innovative IT solutions that support data management and analysis at Joslin's Immunology and Immunogenetics Section. Specifically, their work supports researchers extracting intelligence from data to understand mechanisms and develop therapies for Type-1 Diabetes.