EMPIRICAL BAYES ANALYSIS OF RNA SEQUENCING EXPERIMENTS WITH AUXILIARY INFORMATION

• 成果类型: Article
• 署名作者: Liang, Kun
• 署名单位: University of Waterloo
• 刊物名称: ANNALS OF APPLIED STATISTICS
• ISSN/ISSBN: 1932-6157
• DOI: 10.1214/19-AOAS1270
• 发表日期: 2019
• 页码: 2452-2482
• 关键词: false discovery rate; null hypotheses; PROPORTION; psoriasis; expression; ORACLE; VALUES; genes; rates; POWER

摘要

Finding differentially expressed genes is a common task in high-throughput transcriptome studies. While traditional statistical methods rank the genes by their test statistics alone, we analyze an RNA sequencing dataset using the auxiliary information of gene length and the test statistics from a related microarray study. Given the auxiliary information, we propose a novel nonparametric empirical Bayes procedure to estimate the posterior probability of differential expression for each gene. We demonstrate the advantage of our procedure in extensive simulation studies and a psoriasis RNA sequencing study. The companion R package calm is available at Bioconductor.