Approximate inference of gene regulatory network models from RNA-Seq time series data

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• RNA-Seq data presents challenges for gene regulatory network inference due to its count-based nature.
• Existing methods struggle with the statistical properties of RNA sequencing measurements.

Purpose of the Study:

• To develop a novel model for inferring gene regulatory network structures from RNA-Seq time series data.
• To address the limitations of current methods by explicitly modeling RNA-Seq data distributions.

Main Methods:

• Utilized a negative binomial distribution to model RNA-Seq count observations.
• Employed sparse regression with a horseshoe prior for learning dynamic Bayesian networks.
• Applied variational inference for efficient estimation of model parameters.

Main Results:

• The proposed method demonstrated improved performance in learning directed gene networks compared to other sparse regression techniques on synthetic data.
• Successfully applied to human neuronal stem cell differentiation data to infer gene interactions.

Conclusions:

• The model enhances gene regulatory network inference from RNA-Seq time series by accurately capturing data distributions.
• Approximate inference enables rapid network structure learning with modest computational resources.