MntJULiP and Jutils: differential splicing analysis of RNA-seq data with covariates

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• RNA sequencing (RNA-seq) data analysis is challenged by complex datasets with multiple confounding variables.
• Existing RNA-seq analysis methods often lack the sophistication to handle these confounders, leading to potential inaccuracies.

Purpose of the Study:

• To introduce MntJULiP and Jutils, enhanced programs for differential splicing detection and visualization in RNA-seq data.
• To develop methods that accurately model and adjust for covariates (e.g., sex, age) in RNA-seq analyses.

Main Methods:

• MntJULiP utilizes a Bayesian linear mixture model to detect intron-level splicing differences, adjusted for covariates.
• Jutils provides visualization tools including heatmaps, sashimi plots, Venn diagrams, and Principal Component Analysis (PCA) maps.
• The methods were applied to GTEx brain RNA-seq samples to analyze the impact of sex and age on splicing patterns.

Main Results:

• Covariate modeling in MntJULiP significantly reduces false positives, achieving >90% precision and outperforming competing methods.
• Analysis of GTEx frontal cortex data revealed increased splicing differences with greater age group disparities.
• Clustering of covariate-adjusted data identified a distinct subgroup with unique splicing programs across the age span.

Conclusions:

• The enhanced MntJULiP and Jutils programs offer a robust solution for analyzing complex RNA-seq data with covariates.
• These tools provide high precision in differential splicing detection and enable detailed visualization of covariate effects.
• The findings highlight age-related splicing variations and identify specific subgroups with distinct splicing behaviors.