Multiplicity Adjustment for Intersection-union Test: DetectingOverlapping Genes from Multiple Microarray Gene Lists

摘要

Detecting overlapping genes from multiple gene lists is a frequently used procedure in microarray- based studies. However in previous studies, the intersection operations of gene lists, formally known as the Intersection-Union Tests (IUTs), were performed without being aware of the incurred changes in Type 1 error rate. We presented an IUT adjustment procedure, called Relaxed IUT (RIUT), which is proved to be less conservative and more powerful for combining independent tests. We demonstrated the superiority of RIUT for detecting overlapping genes by simulation and two real toxicogenomic gene expression data sets. Notably, the better power of RIUT enables it to identify overlapping gene sets leading to known related pathways where traditional IUT fails to find.