Discovering gene regulatory networks of multiple phenotypic groups using dynamic Bayesian networks.

Area of Science:

• Bioinformatics
• Computational Biology
• Systems Biology

Background:

• Gene regulatory networks (GRNs) are crucial for understanding cellular processes.
• Time series gene expression data offers insights into dynamic biological systems.
• Discovering GRNs from complex data remains a significant challenge.

Purpose of the Study:

• To develop a scalable Bayesian approach for learning dynamic Bayesian networks (DBNs) from gene expression data.
• To enhance the predictive accuracy of GRN models.
• To identify differences and similarities in GRNs across multiple sample groups.

Main Methods:

• Employed a Bayesian approach for learning DBNs, focusing on scalability and predictive accuracy.
• Utilized cross-validated predictive accuracy for optimal model selection.
• Applied the framework to time series transcriptomic datasets from the Gene Expression Omnibus database.

Main Results:

• The proposed DBN approach demonstrated superior performance in preventing overfitting compared to existing techniques.
• Achieved higher classification accuracy than previously reported for anti-cancer therapy response and normal vs. tumor colorectal tissue.
• Identified significant differences in GRNs between normal and cancer colorectal tissues, particularly around oncogenes.

Conclusions:

• The developed DBN framework provides a robust and accurate method for GRN discovery from gene expression data.
• The identified GRN differences in colorectal cancer offer potential targets for novel therapeutic strategies.
• This approach facilitates comparative GRN analysis across different biological conditions or sample groups.