Identifying Relevant Covariates in RNA-seq Analysis by Pseudo-Variable Augmentation

Area of Science:

• Genomics
• Bioinformatics
• Statistical Genetics

Background:

• RNA-sequencing (RNA-seq) is crucial for identifying differentially expressed genes.
• RNA-seq datasets often contain relevant and irrelevant covariates that can complicate gene expression analysis.
• Ignoring or incorrectly adjusting for covariates can compromise the accurate identification of differentially expressed genes.

Purpose of the Study:

• To develop a robust variable selection method for RNA-sequencing data.
• To accurately identify relevant covariates while controlling the false selection rate.
• To improve the detection of differentially expressed genes in the presence of complex covariate structures.

Main Methods:

• Proposed a novel variable selection method utilizing pseudo-variables.
• Implemented a strategy to control the expected proportion of irrelevant selected covariates.
• Developed the FSRAnalysisBS function within the R package csrnaseq.

Main Results:

• The proposed method accurately selects relevant covariates.
• The method effectively controls the false selection rate below a specified threshold.
• Demonstrated superior performance compared to existing methods for differential gene expression analysis with covariates.

Conclusions:

• The new variable selection approach enhances the reliability of differential gene expression analysis.
• This method offers a significant improvement for RNA-seq studies with complex covariate data.
• The csrnaseq R package provides a practical implementation for researchers.