EBSeq: an empirical Bayes hierarchical model for inference in RNA-seq experiments.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Messenger RNA (mRNA) expression is vital for development, differentiation, and disease.
• RNA sequencing (RNA-seq) enables genome-wide identification of differentially expressed (DE) genes and isoforms.
• Existing statistical methods for DE gene analysis are often misapplied to isoform-level analysis, leading to inaccuracies.

Purpose of the Study:

• To develop a robust statistical method for identifying DE isoforms from RNA-seq data.
• To address the limitations of applying gene-level methods to isoform expression estimation.
• To improve the accuracy and power of DE isoform detection.

Main Methods:

• Development of EBSeq, an empirical Bayesian approach for DE isoform identification.
• Application of EBSeq to RNA-seq experiments comparing multiple biological conditions.
• Utilizing R package for implementation and data analysis.

Main Results:

• EBSeq demonstrates substantially improved power and performance in identifying DE isoforms compared to existing methods.
• The method proves robust for identifying DE genes as well.
• Empirical Bayesian methods enhance the accuracy of isoform expression analysis.

Conclusions:

• EBSeq provides a powerful and accurate solution for DE isoform analysis in RNA-seq.
• The developed method overcomes limitations of traditional count-based approaches for isoform inference.
• EBSeq is a valuable tool for researchers investigating gene and isoform expression.