TRANSCRIPTOME 2002: From Functional Genomics to Systems
Encoding Biological Prior Information Into Gene Expression Analysis
Michael F. Ochs, Frank J. Manion, Ghislain Bidaut, Thomas Moloshok, Jeffrey D. Grant Fox Chase Cancer Center, Philadelphia, PA
The fundamental goal of gene expression data analysis is to recover biological information in the form of coexpression groups and signaling or metabolic pathway information. However, present methods assign each gene to only one coexpression group, which limits their ability to recover fundamental biological behavior, since genes often lie in multiple coexpression groups, which are activated in response to different external stimuli. In addition, published algorithms are generally based on standard statistical tools and data mining methods and do not encode any biological information in their structure. We present a new system designed to overcome these limitations. The algorithm is based on our previous work in identifying spectral signatures from mixtures through the incorporation of prior knowledge. We show preliminary results from the analysis of public domain data, which demonstrate the ability of the system to identify genes that are multiply expressed in response to different stimuli. In addition we discuss the importance of inclusion of biological information in the form of the underlying model encoded within the analysis software. Finally, future additions to the algorithm, which should increase its power to recover biologically significant features in the data will be discussed.
Return to Table of Contents * Speaker Abstracts * Poster Abstracts * View the Photos
Return to Meetings Home Page
This site produced by the Human Genome Management Information System of Oak Ridge National Laboratory.