Exaggerated false positives by popular differential expression methods when analyzing human population samples

• DOI: 10.1186/s13059-022-02648-4
• Publication Date: 2022-03-15T07:02:59Z
• Authors: Yumei Li; Xinzhou Ge; Fanglue Peng; Wei Li; Jingyi Jessica Li
• Keywords: False Discovery Rate; Bonferroni correction; Multiple comparisons problem; Rank (graph theory)

Abstract When identifying differentially expressed genes between two conditions using human population RNA-seq samples, we found a phenomenon by permutation analysis: popular bioinformatics methods, DESeq2 and edgeR, have unexpectedly high false discovery rates. Expanding the analysis to limma-voom, NOISeq, dearseq, Wilcoxon rank-sum test, that FDR control is often failed except for test. Particularly, actual FDRs of edgeR sometimes exceed 20% when target 5%. Based on these results, population-level studies with large sample sizes, recommend