Abstract:Our study focused on the implementation and testing of machine learning interatomic potentials (MLIPs) into the AMBER software suite. This implementation enables us to perform a novel type of molecular dynamics simulation utilizing the hybrid machine learning/molecular mechanics (ML/MM) potentials. To underpin the capabilities of ML/MM simulations, we first validated our implementation at a fundamental physical level by confirming energy and momentum conservation laws. The successful validation indicates that our implementation is able to produce adequate and physically interpretable samplings. Building upon this, for the first time to the best of our knowledge, we proposed an ML/MM-compatible thermodynamic integration (TI) protocol to tackle real-world challenges, such as solvation free energy calculation. Our results demonstrate that this computational protocol can predict hydration free energies with an accuracy of less than 1.00 kcal/mol compared to experimental data, paving the way for the use of ML/MM in multiscale simulations to addressing future drug design problems. Moreover, by applying ML/MM in molecular dynamics simulations of protein-ligand complexes, we demonstrated that the adequate samplings enable us to accurately reproduce experimental binding free energies. Thus, our implementation can offer new insights into biomolecular systems using the ML/MM "microscope". Last, we demonstrated that our implementation can achieve nanosecond timescale simulations daily after significant effort being put to improve the code performance. In a conclusion, we have successfully implemented ML/MM potential to AMBER software package after overcoming limitations in current multi-scale simulations including low computational efficiency. We have advanced TI theory allowing us to accurately predict free energies with ML/MM potentials.

What problem does this paper attempt to address?

### Problems the Paper Aims to Solve This paper focuses on integrating machine learning interatomic potentials (MLIPs) into the AMBER software suite to achieve a novel molecular dynamics simulation method, namely hybrid machine learning/molecular mechanics (ML/MM) potential simulations. Through this approach, the authors aim to address the following key issues: 1. **Improving Computational Efficiency of Multiscale Simulations**: - One of the main limitations of current multiscale simulation techniques is low computational efficiency. Traditional quantum mechanics/molecular mechanics (QM/MM) methods, while capable of providing high-precision results, are difficult to significantly accelerate due to the complexity of quantum mechanical calculations. By introducing MLIPs, computational efficiency can be significantly improved while maintaining high accuracy. 2. **Validating the Physical Consistency of the ML/MM Method**: - To ensure the effectiveness and reliability of the ML/MM method, the authors first validated the laws of energy and momentum conservation at the fundamental physical level. Successful validation indicates that the ML/MM method can produce reasonable and physically interpretable sampling results. 3. **Developing Thermodynamic Integration (TI) Protocols for Practical Problems**: - The authors proposed a thermodynamic integration (TI) protocol compatible with ML/MM to address practical challenges, such as the calculation of solvation free energy. Experimental results show that this calculation protocol can predict hydration free energy with an error of less than 1.00 kcal/mol, providing a new avenue for multiscale simulations in future drug design. 4. **Accurately Predicting Binding Free Energy of Protein-Ligand Complexes**: - By applying the ML/MM method in molecular dynamics simulations of protein-ligand complexes, the authors demonstrated that this method can accurately reproduce experimental binding free energies. This indicates the potential application value of the ML/MM method in the simulation of biomolecular systems. 5. **Enhancing Simulation Performance**: - To further improve the performance of ML/MM MD, the authors optimized the code and tested its performance under different hardware configurations. Results show that through parallel computing and optimized memory management, ML/MM MD can complete nanosecond-level simulations in a relatively short time. In summary, this paper aims to develop an efficient and accurate multiscale simulation method by integrating MLIPs into the AMBER software suite, addressing the issues of computational efficiency and accuracy in current molecular dynamics simulations.

Accurate Free Energy Calculation via Multiscale Simulations Driven by Hybrid Machine Learning and Molecular Mechanics Potentials

Amber Free Energy Tools: Interoperable Software for Free Energy Simulations Using Generalized Quantum Mechanical/molecular Mechanical and Machine Learning Potentials.

Towards chemical accuracy for alchemical free energy calculations with hybrid physics-based machine learning / molecular mechanics potentials

Accelerating Metadynamics-Based Free-Energy Calculations with Adaptive Machine Learning Potentials.

Accurate interatomic force field for molecular dynamics simulation by hybridizing classical and machine learning potentials

Machine-Learning-Assisted Free Energy Simulation of Solution-Phase and Enzyme Reactions

An evaluation of machine learning/molecular mechanics end-state corrections with mechanical embedding to calculate relative protein-ligand binding free energies

Advancing Multiscale Molecular Modeling with Machine Learning-Derived Electrostatics

Assessment of Embedding Schemes in a Hybrid Machine Learning/Classical Potentials (ML/MM) Approach

NNP/MM: Accelerating Molecular Dynamics Simulations with Machine Learning Potentials and Molecular Mechanics

How Well Does the Extended Linear Interaction Energy Method Perform in Accurate Binding Free Energy Calculations?

Neural Network Potential with Multi-Resolution Approach Enables Accurate Prediction of Reaction Free Energies in Solution

Machine Learning of Accurate Energy-Conserving Molecular Force Fields

NNP/MM: Accelerating molecular dynamics simulations with machine learning potentials and molecular mechanic

Free Energy Calculations Using the Movable Type Method with Molecular Dynamics Driven Protein-Ligand Sampling

Free energy simulations with the AMOEBA polarizable force field and metadynamics on GPU platform

Improving Molecular Dynamics Simulations for Solid‐Liquid Interface with Machine Learning Interatomic Potentials

Accelerated Computation of Free Energy Profile at Ab Initio Quantum Mechanical/Molecular Mechanics Accuracy Via a Semiempirical Reference-Potential. 4. Adaptive QM/MM

Accelerating ab initio QM/MM Molecular Dynamics Simulations with Multiple Time Step Integration and a Recalibrated Semi-empirical QM/MM Hamiltonian

Considerations in the use of ML interaction potentials for free energy calculations

Assessment of embedding schemes in a hybrid machine learning/classical potentials (ML/MM) scheme