Curbing false discovery rates in interpretation of genome-wide expression profiles

Bernard Ycart; Frédéric Pont; Jean-Jacques Fournié

doi:10.1016/j.jbi.2013.09.004

Curbing false discovery rates in interpretation of genome-wide expression profiles

J Biomed Inform. 2014 Feb:47:58-61. doi: 10.1016/j.jbi.2013.09.004. Epub 2013 Sep 21.

Authors

Bernard Ycart¹, Frédéric Pont², Jean-Jacques Fournié²

Affiliations

¹ Université Joseph Fourier, Grenoble, France; Laboratoire Jean Kuntzmann, CNRS UMR 5224, Grenoble, France; Laboratoire d'Excellence 'TOUCAN', Toulouse, France. Electronic address: [email protected].
² Laboratoire d'Excellence 'TOUCAN', Toulouse, France; INSERM UMR1037, Cancer Research Center of Toulouse, Toulouse, France; Université Toulouse III Paul-Sabatier, Toulouse, France; ERL 5294 CNRS, Toulouse, France.

PMID: 24060601
DOI: 10.1016/j.jbi.2013.09.004

Abstract

Fisher's exact test is widely used in biomedical research, particularly in genomic profile analysis. Since in most databases, the frequency distribution of genes is right skewed, we show here that its use can lead to excessive false-positive discoveries. We propose to apply Zelen's exact test on a stratification of the gene set; this solves the false discovery problem, and should avoid misleading interpretations of lists of genes produced by various genome-wide analysis technologies.

Keywords: Genomics; Statistics; Transcriptomics.

Publication types

Research Support, Non-U.S. Gov't

MeSH terms

Algorithms
Computational Biology / methods
Databases, Factual
False Positive Reactions
Gene Expression Profiling*
Genome-Wide Association Study*
Genomics
Humans
Neoplasms / genetics
Proteomics / methods
Research Design
Transcription, Genetic
Transcriptome