Covariate-dependent negative binomial factor analysis of RNA sequencing data.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• High-throughput sequencing, particularly RNA sequencing (RNA-seq), is fundamental in biomedical research.
• Analyzing coordinated gene expression patterns can reveal cellular mechanisms for improved disease understanding and treatment.
• Existing co-expression network methods for microarray data may introduce bias when applied to RNA-seq data.

Purpose of the Study:

• To develop a method specifically adapted for RNA-seq data to identify functional gene modules.
• To capture coordinated gene expression changes while accounting for covariate effects.
• To overcome limitations of existing methods, such as ad-hoc choices in data processing and normalization.

Main Methods:

• Developed a fully Bayesian covariate-dependent negative binomial factor analysis (dNBFA) method.
• The dNBFA model directly analyzes RNA-seq count data, incorporating covariate information.
• Employs novel data augmentation techniques for efficient Bayesian inference of model parameters.

Main Results:

• The dNBFA method effectively captures coordinated gene expression changes in RNA-seq data.
• It successfully identifies gene modules with significant differential expression relevant to complex diseases.
• Experimental results demonstrate its power in uncovering meaningful biological insights.

Conclusions:

• dNBFA is a robust tool for analyzing RNA-seq data to discover biologically relevant gene modules.
• The method avoids common biases and preprocessing steps associated with other network-based approaches.
• dNBFA offers a powerful approach for understanding complex diseases through gene expression analysis.