Inference of sparse combinatorial-control networks from gene-expression data: a message passing approach

Area of Science:

• Systems Biology
• Genomics
• Bioinformatics

Background:

• Transcriptional gene regulation controls essential cellular processes like development and response to environmental changes.
• Gene regulation involves complex networks of transcription factors and other molecules, often analyzed using gene-expression data.
• Existing models often assume transcription factor independence, leading to inaccuracies with correlated gene expression.

Purpose of the Study:

• To develop a new algorithm for inferring combinatorial control networks from gene-expression data.
• To improve the prediction of gene regulatory interactions by accounting for combinatorial control.
• To overcome limitations of models assuming transcription factor independence.

Main Methods:

• A novel algorithm employing a message-passing approach to infer combinatorial gene regulatory networks.
• The algorithm avoids explicit sampling over putative gene-regulatory networks.
• Application to artificial cell-cycle network models and large-scale yeast gene expression datasets.

Main Results:

• Successfully recovered the structure of an artificial yeast cell-cycle network.
• Identified combinatorial regulations in a large-scale yeast gene expression dataset.
• Applied to the Pleiotropic Drug Resistance (PDR) network for medical relevance.

Conclusions:

• The developed algorithm recovers biologically meaningful gene interactions, validated by experimental results.
• Predicts new instances of combinatorial gene control, enhancing regulatory network inference.
• Demonstrates the value of incorporating combinatorial control into models for extracting more interactions from microarray data.