Integrating genomics and proteomics data to predict drug effects using binary linear programming.

Area of Science:

• Systems Biology
• Computational Biology
• Pharmacology

Background:

• The Library of Integrated Network-Based Cellular Signatures (LINCS) project seeks to understand cellular responses to perturbations.
• Current methods for predicting compound effects often lack cell-specificity.
• Understanding cell-specific pathways is crucial for effective drug discovery.

Purpose of the Study:

• To develop a novel computational approach for inferring cell-specific pathways.
• To predict compound effects on cellular signaling using gene expression and phosphoproteomics data.
• To elucidate the mechanisms of action for various compounds in cancer cell lines.

Main Methods:

• Inferred potential compound targets and created a generic pathway map using gene expression data.
• Utilized binary linear programming (BLP) to optimize pathway topology based on phosphorylation data.
• Validated the model's predictive accuracy using cross-validation with known compounds and cell lines (MCF7, PC3).

Main Results:

• The developed model successfully inferred cell-specific pathways for MCF7 and PC3 cell lines.
• Cross-validation demonstrated high accuracy in predicting compound effects.
• The model elucidated specific mechanisms for compounds like trichostatin A, MS-275, staurosporine, and digoxigenin.

Conclusions:

• The novel computational model effectively infers cell-specific pathways and predicts compound efficacy.
• This approach advances our understanding of cellular signaling and facilitates drug mechanism elucidation.
• The model shows promise for personalized medicine and drug development by predicting compound responses.