Abstract:We present a novel integration of the ANI neural networks into the Amber software suite, offering a sophisticated machine learning/molecular mechanics (ML/MM) framework. The implementation is designed as a general-purpose tool for the simulation of neutral organic molecules, requiring no additional training for its use beyond the initial setup. The framework leverages a new ANI potential that accurately predicts geometry-dependent atomic partial charges at the Minimal Basis Iterative Stockholder (MBIS) level, enhancing the modeling of electrostatic interactions within ML/MM systems. Additionally, we incorporate a polarization correction to address the distortion effects on the ML subsystem from MM point charges. Our approach is validated through simulations of solvation profiles, vibrational spectra, and torsion free energy profiles of small molecules in aqueous environments, as well as protein-ligand interactions. Our findings demonstrate that this ML/MM framework can approximate QM/MM electrostatic embedding with significantly reduced computational demands, paving the way for more efficient and accurate simulations in computational chemistry.

What problem does this paper attempt to address?

### Problems the Paper Aims to Solve This paper aims to address the integration of ANI neural networks into the Amber software suite to provide an efficient and accurate machine learning/molecular mechanics (ML/MM) framework. Specifically, the paper focuses on the following aspects: 1. **Improving Simulation Accuracy**: By introducing new ANI potential functions, it can accurately predict geometry-dependent atomic partial charges based on the Minimal Basis Iterative Stockholder (MBIS) level, thereby enhancing the modeling of electrostatic interactions in ML/MM systems. 2. **Reducing Computational Cost**: The proposed method can approximate the electrostatic embedding effects of quantum mechanics/molecular mechanics (QM/MM) while significantly reducing computational demands, making the simulation of complex chemical and biological molecular systems more efficient. 3. **Generality**: The designed framework is a general tool applicable to the simulation of neutral organic molecules, which users can use without additional training, requiring only initial setup. 4. **Addressing Polarization Effects**: Polarization corrections are introduced to address the distortion effects from molecular mechanics point charges on the machine learning subsystem. ### Validation Methods To validate the effectiveness and accuracy of the framework, the authors conducted the following types of simulation experiments: 1. **Solvation Profiles**: Simulating the solvation profiles of small molecules in a water environment. 2. **Vibrational Spectra**: Calculating the vibrational spectra of small molecules in a water environment. 3. **Torsional Free Energy Profiles**: Analyzing the torsional free energy profiles of small molecules in a water environment. 4. **Protein-Ligand Interactions**: Evaluating the physical accuracy of protein-ligand interactions. ### Main Contributions - **New ANI-MBIS-q Model**: Predicts MBIS charges directly from the local chemical environment without the need for electron density. - **Efficient ML/MM Interface**: Implements the ANI neural network in Amber's SANDER engine, avoiding the cost of calling Python code each time. - **Wide Applicability**: The framework can be used for various systems without the need to train models from scratch. ### Conclusion Through these simulation experiments, the research results indicate that the proposed ML/MM framework can provide simulation accuracy comparable to QM/MM while significantly reducing computational demands, paving the way for efficient and accurate simulations in computational chemistry.

ANI Neural Networks Meet Electrostatics: A ML/MM Implementation in Amber

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

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

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

Incorporating Neural Networks into the AMOEBA Polarizable Force Field

emle-engine: a flexible electrostatic machine learning embedding package for multiscale molecular dynamics simulations

AENET-LAMMPS and AENET-TINKER: Interfaces for Accurate and Efficient Molecular Dynamics Simulations with Machine Learning Potentials

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

Machine Learning Interatomic Potentials and Long-Range Physics

LAMMPS Implementation of Rapid Artificial Neural Network Derived Interatomic Potentials

AIMNet2: A Neural Network Potential to Meet your Neutral, Charged, Organic, and Elemental-Organic Needs

Scalable Hybrid Deep Neural Networks/Polarizable Potentials Biomolecular Simulations including long-range effects

Software Infrastructure for Next-Generation QM/MM-ΔMLP Force Fields

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

Machine Learning in QM/MM Molecular Dynamics Simulations of Condensed-Phase Systems

Electrostatic Embedding Machine Learning for Ground and Excited State Molecular Dynamics of Solvated Molecules

Accelerated Quantum Mechanics/Molecular Mechanics Simulations Via Neural Networks Incorporated with Mechanical Embedding Scheme.

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

Including physics-informed atomization constraints in neural networks for reactive chemistry

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