MTHFR Variant Hypothesis Prioritization: Computational Prioritization of Selected MTHFR Variant States for Experimental Follow-Up

Computational hypothesis-prioritization study comparing wild-type, single-variant, and compound heterozygous MTHFR states using AlphaFold 3 and Boltz-2. Across 64 structural predictions (10 independent seeds per configuration), the compound heterozygous dimer yielded the lowest reported interaction-confidence values across assessed dimer-interface and ligand-associated metrics. Three of four key metrics survived Bonferroni correction (ipTM p=0.035, pTM p=0.031, pLDDT@429 p=0.000005). RMSD validation against PDB 6FCX confirmed high prediction accuracy (1.42 A). Molecular dynamics simulations (OpenMM) are in progress. All sequences, analysis scripts, and results are available at https://github.com/DSMPromo/mthfr-target-validation