Topconfects: a package for confident effect sizes in differential expression analysis provides a more biologically useful ranked gene list

Paul F Harrison; Andrew D Pattison; David R Powell; Traude H Beilharz

doi:10.1186/s13059-019-1674-7

Topconfects: a package for confident effect sizes in differential expression analysis provides a more biologically useful ranked gene list

Genome Biol. 2019 Mar 28;20(1):67. doi: 10.1186/s13059-019-1674-7.

Authors

Paul F Harrison¹, Andrew D Pattison², David R Powell^{3

2}, Traude H Beilharz⁴

Affiliations

¹ Monash Bioinformatics Platform, Monash University, Melbourne, Australia. Paul.harrison@monash.edu.
² Stem Cell and Development Program, Monash Biomedicine Discovery Institute and Department of Biochemistry and Molecular Biology, Monash University, Melbourne, Australia.
³ Monash Bioinformatics Platform, Monash University, Melbourne, Australia.
⁴ Stem Cell and Development Program, Monash Biomedicine Discovery Institute and Department of Biochemistry and Molecular Biology, Monash University, Melbourne, Australia. traude.beilharz@monash.edu.

Abstract

Differential gene expression analysis may discover a set of genes too large to easily investigate, so a means of ranking genes by biological interest level is desired. p values are frequently abused for this purpose. As an alternative, we propose a method of ranking by confidence bounds on the log fold change, based on the previously developed TREAT test. These confidence bounds provide guaranteed false discovery rate and false coverage-statement rate control. When applied to a breast cancer dataset, the top-ranked genes by Topconfects emphasize markedly different biological processes compared to the top-ranked genes by p value.

Keywords: Confidence interval; Differential expression analysis; False coverage-statement rate; False discovery rate; RNA-Seq; TREAT.

Publication types

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

MeSH terms

Breast Neoplasms / genetics
Genetic Techniques*
Humans
Software*