A METHOD FOR THE DETECTION OF MEANINGFUL AND REPRODUCIBLE GROUP SIGNATURES FROM GENE EXPRESSION PROFILES

摘要

Gene expression microarrays are commonly used to detect the biological signature of andisease or to gain a better understanding of the underlying mechanism of how a groupnof drugs treat a specific disease. The outcome of such experiments, e.g. the signature, isna list of differentially expressed genes. Reproducibility across independent experimentsnremains a challenge. We are interested in creating a method that can detect the sharednsignature of a group of expression profiles, e.g. a group of samples from individuals withnthe same disease or a group of drugs that treat the same therapeutic indication. Wenhave developed a novel Weighted Influence — Rank of Ranks (WIMRR) method, and wendemonstrate its ability to produce both meaningful and reproducible group signatures