dearseq

dearseq performs differential expression analysis of RNA-seq data using a variance component score test that controls the false discovery rate without assuming a specific distribution for the data.

Key Features:

• Variance component score test: Implements a robust variance component score test for differential expression analysis that does not assume a true distribution of RNA-seq data.
• Precision weights for heteroscedasticity: Accounts for data heteroscedasticity through precision weights.
• Gene-wise analysis: Performs gene-level differential expression analysis.
• Gene set analysis: Performs gene set-level differential expression analysis.
• Repeated and longitudinal data: Handles repeated or longitudinal RNA-seq data structures.
• False discovery rate control: Effectively controls the false discovery rate while maintaining statistical power.

Scientific Applications:

• RNA-seq differential expression analysis: Identifies differentially expressed genes from RNA-seq data with reduced false positive rates compared to commonly used methods.
• Time-course gene set analysis: Analyzes longitudinal RNA-seq data to identify temporally changing gene sets.
• Tuberculosis and infectious disease studies: Applied to tuberculosis studies to produce biologically meaningful results with fewer apparent false positives.

Methodology:

The method uses a variance component score test with precision weights to account for heteroscedasticity. Mathematical proofs, simulations, and real data validation demonstrate FDR control and statistical power. Methods are detailed in Agniel & Hejblum (2017) Biostatistics and Gauthier et al. (2020) NAR Genomics and Bioinformatics.