Improvement of MPBioPath for Reactome Scale Perturbation Analysis

MP-BioPath uses rigorous mathematical programming to predict exactly how genetic mutations or drug perturbations cascade through Reactome's biological pathways. The current model needs better integration of tissue-specific realities and address the issue of artificially diluting predictive signals when multiple proteins share similar roles (entity-set dilution). AND While the Reactome team generated an expansion of these logic networks across their entire database, this massive architecture remains unvalidated at scale. This project aims to transform MP-BioPath into a scalable, context-aware predictive engine. I will architect an automated validation pipeline to stress-test the newly generated Reactome-wide logic networks, and then use this infrastructure to systematically benchmark MP-BioPath against traditional, topology-agnostic methods like Gene Set Enrichment Analysis (GSEA). I will build a transcriptomic-weighting module as the context later: by feeding real-world RNA-seq data into the optimization model, this extension will act as "live traffic updates" for the biological pathways, dynamically adjusting node weights based on actual tissue expression to resolve entity-set dilution and enable highly specific predictions.