Computational Prediction of the Therapeutic Mechanisms From the Turmeric-Pueraria-Coix Seed Botanical Formula in Rheumatoid Arthritis.

Rheumatoid arthritis (RA) is characterized by progressive synovial inflammation and immune-metabolic reprogramming. The polyphenol-rich Turmeric-Pueraria-Coix seed (TPC) formula shows therapeutic potential against RA, though its multi-target mechanisms remain uncharacterized. To explore how TPC modulates ferroptosis and oxidative stress in RA, we integrated systems nutrition network analysis with multicohort transcriptomic datasets to identify candidate targets. Ensemble machine learning algorithms and independent cohort validation prioritized key biomarkers. Molecular dynamics (MD) simulations evaluated the dynamic behavior and interaction persistence of polyphenol-target complexes, and single-cell in silico overexpression and knockout analyses characterized the regulatory consequences of perturbing these targets within RA synovial networks. Analysis identified 85 bioactive phytochemicals in the TPC formula, forming a polyphenol-dominant matrix mapping to 40 RA targets. Algorithmic and cross-cohort prioritization isolated six core genes (ALOX5, CD38, CDK1, EZH2, MAPK8, and PTGS2) linked to RA synovial transcriptomic patterns, with varying predictive performance across external cohorts. SHAP analysis highlighted PTGS2, CD38, and MAPK8; however, MD simulations prioritized CD38 and MAPK8 as stable targets of coumestrol, whereas PTGS2 exhibited less dynamic stability. Single-cell perturbation analysis revealed similar regulatory responses following MAPK8 and PTGS2 modulation, whereas CD38 knockout produced a distinct network response. These findings link TPC-derived polyphenols to CD38/MAPK8 signaling and ferroptosis in RA, providing testable molecular hypotheses for experimental validation.