Differentially expressed heterogeneous overdispersion genes testing for count data.

Area of Science:

• Genomics
• Bioinformatics
• Statistical Genetics

Background:

• RNA sequencing (RNA-seq) is crucial for analyzing gene expression.
• Detecting differentially expressed (DE) genes is key, but current methods struggle with overdispersion and small sample sizes.
• Overdispersion, where read count variance exceeds the mean, reduces statistical power.

Purpose of the Study:

• To introduce a novel RNA-seq analysis procedure, heterogeneous overdispersion genes testing (DEHOGT).
• To address limitations in existing DE gene detection methods, particularly overdispersion and limited sample sizes.
• To enhance the power and flexibility of differential gene expression analysis.

Main Methods:

• Developed DEHOGT, a procedure utilizing heterogeneous overdispersion modeling and post-hoc inference.
• Integrated sample information across conditions for adaptive overdispersion modeling.
• Employed a gene-wise estimation scheme for improved detection power with limited replicates and numerous conditions.

Main Results:

• DEHOGT demonstrated superior performance over DESeq2 and EdgeR on synthetic RNA-seq data.
• The method showed enhanced power in detecting differentially expressed genes.
• Application to microglial cell RNA-seq data revealed potentially more DE genes under stress hormone treatments.

Conclusions:

• DEHOGT offers a more robust and powerful approach for differential gene expression analysis in RNA-seq.
• The method effectively handles overdispersion and benefits from larger numbers of conditions.
• DEHOGT has significant implications for biological research, including studies on microglial responses to stimuli.