Expected value of experimentation in causal discovery from gene expression studies

• Authors: Gregory F. Cooper;Changwon Yoo
• Venue: Expected value of experimentation in causal discovery from gene expression studies
• Year: 2003

Microarray technology has opened a new era for gene expression studies. It allows a relatively quick and easy way to measure the cellular mRNA levels of many different genes at once. Large datasets generated by the microarray experiments provide good opportunities to learn causal relationships among genes. This dissertation investigates how to model and apply the expected value of experimentation (EVE) to discover causal pathways from gene expression data. I introduce a system called GEEVE (causal discovery in Gene Expression data using EVE), which implements EVE in discovering causal pathways using gene expression data. GEEVE provides (1) recommendations of which experiments to perform; (2) recommendations of the number of measurements to include in the experimental design; and (3) a Bayesian analysis that combines prior knowledge with the results of recent microarray experimental results to derive posterior probabilities of gene regulation relationships. My dissertation hypothesis is that the use of the GEEVE system will assist biologists in designing and interpreting microarray experiments in the process of discovering gene-regulation pathways.To evaluate the GEEVE system, I first used a gene expression simulator to generate data from specified models of gene regulation. The results show that the system gives better results than two recently published approaches (1) in learning the generating models of gene regulation and (2) in recommending experiments to perform. I also evaluated the GEEVE system using a randomized control study that involved 10 biologists, some of whom used GEEVE and some of whom did not. The results show that biologists who used GEEVE reached correct causal assessments about gene regulation more often than did those biologists who did not use GEEVE. The GEEVE users also reached their assessments in a more cost-effective manner. The GEEVE methodology is applicable to the discovery of networks of causal relationships in general, not just gene regulation relationships. For example, in the biology domain, when high throughout protein-level measurements are available, the approach taken by GEEVE can be applied to help learn the causal relationships among genes and proteins.