DifferentialRegulation: a Bayesian hierarchical approach to identify differentially regulated genes

Area of Science:

• Bioinformatics
• Computational Biology
• Molecular Biology

Background:

• Transcriptomics typically analyzes mature spliced mRNA.
• Investigating both spliced and unspliced mRNA reveals gene regulation insights.
• Current methods for spliced/unspliced inference often lack group comparison capabilities and struggle with quantification uncertainty due to multi-mapping reads.

Purpose of the Study:

• To develop a robust method for comparing the relative abundance of unspliced mRNA between experimental conditions.
• To address the challenge of quantification uncertainty in spliced/unspliced mRNA inference.
• To provide a flexible tool applicable to both bulk and single-cell RNA sequencing data.

Main Methods:

• Developed DifferentialRegulation, a Bayesian hierarchical method.
• Employed a latent variable approach to model quantification uncertainty by allocating reads to their gene/transcript and splice version.
• Designed benchmarks to evaluate performance against existing methods.

Main Results:

• DifferentialRegulation demonstrates strong performance in sensitivity and error control in benchmark tests.
• The method effectively models quantification uncertainty inherent in spliced and unspliced mRNA data.
• The tool is versatile, supporting both bulk and single-cell RNA sequencing data.

Conclusions:

• DifferentialRegulation offers a powerful new approach for differential analysis of gene expression dynamics using both spliced and unspliced mRNA.
• The method enhances the ability to compare gene regulation across different experimental conditions, particularly in complex datasets.
• The availability as a Bioconductor R package facilitates its adoption in the research community.