TailEnrich enables discovery of dysregulated genes through disease sample enrichment testing

Identifying differentially expressed genes is fundamental to elucidating the molecular mechanisms underlying complex diseases. However, in complex disorders such as Alzheimer's disease (AD), the genes exhibiting dysregulated expression may vary among patient subsets, reflecting the heterogeneity of the underlying pathological processes. Consequently, pathogenic genes are profoundly dysregulated in only a subset of patients. Conventional differential expression tools, such as DESeq2 and edgeR, are primarily designed to detect population-level expression shifts and may have limited power to identify subgroup-specific dysregulation. To overcome this limitation, we developed TailEnrich, a novel computational method that discovers dysregulated genes within patient subsets. TailEnrich independently tests the extreme left (downregulated) or right (upregulated) tails of a gene's expression distribution for disease sample enrichment. We quantify this enrichment using a new metric, i.e. the PE (Percentage of Enriched disease samples) score. On simulated datasets, TailEnrich demonstrates markedly improved accuracy over DESeq2 and edgeR in identifying dysregulated genes specific to patient subsets. Applied to AD gene expression data, TailEnrich uncovers a distinct, biologically relevant set of genes within key pathological pathways that are missed by mean shift-based approaches. TailEnrich provides a novel method for identifying dysregulated genes, and readily extends to other omics data including genomic variants and proteomics to uncover pathogenic genes in complex diseases.