Bayesian variable selection for gene expression modeling with regulatory motif binding sites in neuroinflammatory...

Area of Science:

• Computational Biology
• Genomics
• Bioinformatics

Background:

• Gene transcription is regulated by multiple transcription factors (TFs) acting coordinately.
• Existing computational methods often identify individual TF-binding sites (TFBSs) but overlook their interdependence.
• Understanding complex regulatory networks is crucial for deciphering gene expression patterns.

Purpose of the Study:

• To develop and apply a computational framework for modeling the coordinated regulation of gene expression by TFBSs.
• To investigate the relationship between TFBSs and gene expression levels, considering TF family and individual member motifs.
• To account for the complexity of transcription regulation through advanced statistical modeling.

Main Methods:

• A three-step approach involving data preprocessing, TFBS identification and scoring, and predictive modeling.
• Utilized regression models with Bayesian variable selection (Gibbs sampler) for identifying significant TFBSs.
• Employed linear and probit regression to model gene expression based on TFBS predictors.

Main Results:

• Successfully modeled the relationship between TFBSs and gene expression using family-wise and member-specific motifs.
• Identified intricate regulatory roles of TFs in neuroinflammatory events using spinal cord injury gene expression data.
• The Bayesian variable selection approach effectively identified key TFBSs and reduced false positives.

Conclusions:

• The proposed systematic approach provides a comprehensive perspective on dissecting transcription regulation.
• This method facilitates the generation of plausible hypotheses for combinatorial TF regulation.
• Accurate modeling of TFBS interdependence is essential for understanding molecular events driving cellular and tissue-level phenotypes.