Modeling signal transduction from protein phosphorylation to gene expression

Area of Science:

• Computational biology
• Systems biology
• Genomics and proteomics

Background:

• Cellular signaling networks are crucial for understanding regulatory mechanisms.
• Large-scale genomic and proteomic data enable data-driven discovery of signaling pathways.
• Identifying stimulus-responsive pathways is key for biological insight and cancer research.

Purpose of the Study:

• To develop and apply a computational method for integrating transcriptomic and proteomic data to identify signaling pathways.
• To infer stimulus-specific signaling networks in rat and human cells.
• To explore the application of data-driven pathway inference in cancer research.

Main Methods:

• A novel Bayesian Network (BN) framework was developed to integrate transcriptomic data with proteomic data (protein phosphorylation states).
• Genes and proteins were represented as nodes in the BN, with edges indicating regulatory relationships.
• An efficient inference algorithm incorporated prior pathway knowledge and performed data-driven network structure searches.

Main Results:

• The BN framework successfully inferred sparse, cell-specific signaling networks for rat and human cells using gene expression and proteomic datasets.
• Distinct signaling pathways were identified for rat and human cells, reflecting their differential responses to stimuli.
• The model demonstrated strong performance in the SBV IMPROVER challenge.

Conclusions:

• The data-driven approach effectively reconstructs signaling pathways by integrating multi-omics data.
• This methodology can identify aberrant signaling pathways in cancer subtypes, contributing to personalized cancer research.
• The framework advances the understanding of cellular signaling and its dysregulation in disease.