An Adaptive Force Matching Potential for Alanine Developed with Møller–Plesset Perturbation Theory and Smooth Fourier Transform Correction Map

Developing accurate force fields for biomolecules remains a significant challenge due to the subtle energetic differences between various conformational states. We present a novel force field model for polyalanine, ALAMP2_25, developed using adaptive force matching (AFM) with Møller-Plesset perturbation theory at the second order (MP2) as the reference method. By fitting smaller model compounds and transferring parameters to larger peptides, we overcome the limitations of traditional AFM approaches and enable the use of more accurate electronic structure methods. The ALAMP2_25 model incorporates a new correction scheme, Smooth Fourier Transform-based φ, ψ correction map (SFT-CMAP), which efficiently describes φ, ψ coupling with reduced overfitting. Our model demonstrates good agreement with experimental <i>J</i>-coupling data for hydrated polyalanine and shows improved transferability to <i>N</i>-methylated cyclic alanine when compared to previously reported DFT based models. The developed framework provides a pathway for creating accurate force fields for a broader range of amino acids and biomolecules, enabling first-principles-based simulations of complex biological systems with applications in protein folding, ligand binding, and drug design.