MATS: a Bayesian framework for flexible detection of differential alternative splicing from RNA-Seq data

Area of Science:

• Genomics and Molecular Biology
• Bioinformatics and Computational Biology

Background:

• Ultra-deep RNA sequencing is crucial for genome-wide analysis of pre-messenger RNA (pre-mRNA) alternative splicing.
• Accurate statistical methods are needed to identify differential alternative splicing patterns from RNA sequencing (RNA-Seq) data.

Purpose of the Study:

• To develop a flexible Bayesian statistical framework for hypothesis testing of differential alternative splicing patterns.
• To create a method, MATS (multivariate analysis of transcript splicing), for analyzing RNA-Seq data to detect differential alternative splicing.

Main Methods:

• Developed MATS, a Bayesian framework utilizing a multivariate uniform prior to model correlations in exon splicing patterns.
• Employed Markov chain Monte Carlo (MCMC) with simulation-based adaptive sampling for calculating P-values and false discovery rates (FDR).
• Designed MATS for flexibility, allowing hypothesis testing for user-defined differential alternative splicing patterns.

Main Results:

• Evaluated MATS performance using simulated and real RNA-Seq datasets.
• Achieved an 86% RT-PCR validation rate for differential exon skipping events with a MATS FDR <10% in an analysis of ESRP1-regulated splicing.
• MATS FDR estimates showed good agreement with experimental validation rates across tested exons.

Conclusions:

• MATS is an effective and flexible approach for detecting differential alternative splicing from RNA-Seq data.
• The framework provides robust statistical analysis for identifying splicing events of interest.