Analyzing gene expression data in terms of gene sets: methodological issues

• DOI: 10.1093/bioinformatics/btm051
• Publication Date: 2007-02-16T01:18:27Z
• Authors: Jelle J. Goeman; Peter Bühlmann
• Keywords: Independence; Expression (computer science)

Abstract Motivation: Many statistical tests have been proposed in recent years for analyzing gene expression data terms of sets, usually from Gene Ontology. These methods are based on widely different methodological assumptions. Some approaches test differential each set against the rest genes, whereas others its own. Also, some a model which genes sampling units, treat subjects as units. This article aims to clarify assumptions behind and indicate preferential methodology testing. Results: We identify crucial needed by majority methods. P-values derived that use takes unit easily misinterpreted, they does not resemble biological experiment actually performed. Furthermore, because these models unrealistic independence assumption between such can be wildly anti-conservative, simulation shows. also argue competitively create an unnecessary rift single testing Contact: [e-mail]