RNA-seq analysis for detecting quantitative trait-associated genes.

Area of Science:

• Genomics
• Bioinformatics
• Statistical Genetics

Background:

• RNA-sequencing (RNA-seq) is widely used for differential gene expression analysis.
• Detecting genes associated with quantitative traits using RNA-seq with large biological replicates remains underexplored.

Purpose of the Study:

• To apply and evaluate linear models for detecting trait-associated genes (TAGs) in RNA-seq data.
• To compare the performance of regression-based methods against existing differential gene expression analysis tools.

Main Methods:

• Application of ordinary and robust regression models to analyze two real RNA-seq datasets.
• Comparison of proposed methods with DESeq2 and Voom using simulation studies and qRT-PCR validation.
• Analysis of human obesity and Holstein milk production datasets with large numbers of biological replicates.

Main Results:

• Linear regression methods, particularly robust regression, showed significantly lower false discovery rates compared to two-group comparison approaches.
• Robust regression exhibited higher precision in identifying trait-associated genes (TAGs).
• The proposed methods outperformed existing tools like DESeq2 and Voom in simulation and experimental validation.

Conclusions:

• Linear models, especially robust regression, are effective for detecting trait-associated genes (TAGs) in RNA-seq studies with numerous replicates.
• These methods offer improved precision and reduced false discoveries for quantitative trait analysis.
• The approach is expected to be valuable for various RNA-seq studies involving continuous response traits given decreasing sequencing costs.