Prioritizing Pain-Associated Targets with Machine Learning.

Area of Science:

• Genetics and Bioinformatics
• Pharmacology
• Computational Biology

Background:

• Despite numerous gene associations with pain, molecular mechanisms remain largely unknown, limiting current analgesic targets.
• Existing pain therapeutics target a narrow range of validated pathways.

Purpose of the Study:

• To develop a machine learning (ML) ensemble model for predicting novel molecular targets across 17 pain categories.
• To identify potential drug targets, focusing on G-protein-coupled receptors (GPCRs), ion channels, and protein kinases.

Main Methods:

• Utilized transcriptomics, proteomics, and gene ontology data to train the ML model.
• Employed an ensemble approach to prioritize genes associated with pain modulation.
• Evaluated model performance using Area Under the Receiver Operating Characteristic curve (AUROC).

Main Results:

• The ML model achieved an average AUROC of 0.839 for predicting novel pain targets.
• Highest prediction accuracy was observed for arthritis pain (AUROC = 0.929).
• Identified hundreds of potential novel targets, including specific GPCRs like GPR132 and GPR109B for rheumatoid arthritis.

Conclusions:

• Gene-pain association predictions revealed enrichment in cytokine, calcium, and GABA-related signaling pathways.
• The identified targets provide a foundation for future research into novel analgesics.
• This approach can accelerate the development of safer and more effective pain management strategies.