Scalable Hybrid Deep Neural Networks/Polarizable Potentials Biomolecular Simulations including Long-range Effects

Théo Jaffrelot-Inizan (1) , Thomas Plé (1) , Olivier Adjoua (1) , Pengyu Ren (2) , Hatice Gökcan (3) , Olexandr Isayev (3) , Louis Lagardère (1, 4) , Jean-Philip Piquemal (1, 2)

1 LCT - Laboratoire de chimie théorique
2 Biomedical Engineering [Austin]
3 CMU - Carnegie Mellon University [Pittsburgh]
4 IP2CT - Institut Parisien de Chimie Physique et Théorique

Deep-HP is a scalable extension of the Tinker-HP multi-GPUs molecular dynamics (MD) package enabling the use of Pytorch/TensorFlow Deep Neural Networks (DNNs) models. Deep-HP increases DNNs MD capabilities by orders of magnitude offering access to ns simulations for 100k-atom biosystems while offering the possibility of coupling DNNs to any classical (FFs) and many-body polarizable (PFFs) force fields. It allows therefore to introduce the ANI-2X/AMOEBA hybrid polarizable potential designed for ligand binding studies where solvent-solvent and solvent-solute interactions are computed with the AMOEBA PFF while solute-solute ones are computed by the ANI-2x DNN. ANI-2X/AMOEBA explicitly includes AMOEBA's physical long-range interactions via an efficient Particle Mesh Ewald implementation while preserving ANI-2X's solute short-range quantum mechanical accuracy. The DNNs/PFFs partition can be user-defined allowing for hybrid simulations to include biosimulation key ingredients such as polarizable solvents, polarizable counter ions, etc... ANI-2X/AMOEBA is accelerated using a multiple-timestep strategy focusing on the models contributions to low-frequency modes of nuclear forces. It primarily evaluates AMOEBA forces while including ANI-2x ones only via correction-steps resulting in an order of magnitude acceleration over standard Velocity Verlet integration. Simulating more than 10 μs, we compute charged/uncharged ligands solvation free energies in 4 solvents, and absolute binding free energies of host-guest complexes from SAMPL challenges. ANI-2X/AMOEBA average errors are within chemical accuracy opening the path towards large-scale hybrid DNNs simulations, at force-field cost, in biophysics and drug discovery.

Mots clés: Molecular dynamics; Molecular dynamics simulations; AMOEBA polarizable force field; ANI neural network; Neural network; Neural network models; Binding free energy; Binding free energy prediction; Solvation free energy; Multiple time stepping; Quantum chemistry; Force field molecular dynamics; Machine learning applications; Machine learning and deep learning; Long-range Interactions; Tinker-HP; Deep-HP; Machine learning; Force field

Domaines: Chimie théorique et/ou physique; Intelligence artificielle [cs.AI]

Théo Jaffrelot-Inizan, Thomas Plé, Olivier Adjoua, Pengyu Ren, Hatice Gökcan, et al.. Scalable Hybrid Deep Neural Networks/Polarizable Potentials Biomolecular Simulations including Long-range Effects. Chemical Science, 2023, 14 (5438-5452), ⟨10.1039/D2SC04815A⟩. ⟨hal-03738403⟩