Interpretable bioinformatics approaches for pheochromocytoma bioactivity and protein interaction analysis

Area of Science:

• Computational biology
• Network pharmacology
• Oncology

Background:

• Pheochromocytoma (PCC) is a rare neuroendocrine tumor driven by oncogenic c-Myc/Max complexes.
• Molecular interactions and targeted therapies for PCC remain poorly understood.

Purpose of the Study:

• To elucidate key molecular mechanisms and bioactive motifs in PCC using an integrative computational pipeline.
• To identify critical protein targets and actionable molecular signatures for targeted therapy development.

Main Methods:

• Integrative bioinformatics, network biology, and machine learning (Random Forest, SVM, Gradient Boosting).
• Genetic programming for structural motif identification from 5000 bioactive molecules (ChEMBL).
• Protein-protein interaction (PPI) network construction and analysis (STRING, BioGRID), community detection (Girvan-Newman), and explainable AI (XAI) for interpretability.

Main Results:

• Machine learning models achieved high accuracy (mean accuracy: 0.98, AUC >0.97) in predicting bioactivity.
• Key determinants of bioactivity include pIC50, molecular weight, lipophilicity, and hydrogen-bonding properties.
• MYC, MAX, and EP300 identified as central hubs in PCC molecular networks, enriched for cancer-related pathways.

Conclusions:

• The study successfully identified critical molecular targets and bioactive motifs for pheochromocytoma.
• The findings provide a computational foundation for developing novel targeted therapies for PCC and other rare cancers.
• Further experimental validation is needed to confirm the computational predictions.