Bivariate microarray analysis: statistical interpretation of two-channel functional genomics data.

摘要: Conventional statistical methods for interpreting microarray data require large numbers of replicates in order to provide sufficient levels sensitivity. We recently described a method identifying differentially-expressed genes one-channel 1. Based on the idea that variance structure can itself be reliable measure noise, this allows statistically sound interpretation as few two per treatment condition. Unlike array, two-channel platform simultaneously compares gene expression RNA samples. This leads covariation measured signals. Hence, by accounting model, we significantly increase power test. believe approach has potential overcome limitations existing methods. present here novel analysis involves modeling paired context Bayesian framework. also describe test used identify genes. method, bivariate (BMA), demonstrates dramatically improved sensitivity over approaches. show with only array replicates, it is possible detect changes are at best detected six other Further, combining results from BMA Gene Ontology annotation yields biologically significant ligand-treated macrophage cell system.