DMFF: An Open-Source Automatic Differentiable Platform for Molecular Force Field Development and Molecular Dynamics Simulation

Reversible molecular simulation for training classical and machine-learning force fields

The next generation of force fields for molecular dynamics will be developed using a wealth of data. Training systematically with experimental data remains a challenge, however, especially for machine-learning potentials. Differentiable molecular simulation calculates gradients of observables with respect to parameters through molecular dynamics trajectories. Here, we improve this approach by explicitly calculating gradients using a reverse-time simulation with effectively constant memory cost and a computation count similar to the forward simulation. The method is applied to learn all-atom water and gas diffusion models with different functional forms and to train a machine-learning potential for diamond from scratch. Comparison to ensemble reweighting indicates that reversible simulation can provide more accurate gradients and train to match time-dependent observables.

MDRefine: A Python package for refining molecular dynamics trajectories with experimental data

Molecular dynamics (MD) simulations play a crucial role in resolving the underlying conformational dynamics of molecular systems. However, their capability to correctly reproduce and predict dynamics in agreement with experiments is limited by the accuracy of the force-field model. This capability can be improved by refining the structural ensembles or the force-field parameters. Furthermore, discrepancies with experimental data can be due to imprecise forward models, namely, functions mapping simulated structures to experimental observables. Here, we introduce MDRefine, a Python package aimed at implementing the refinement of the ensemble, the force field, and/or the forward model by comparing MD-generated trajectories with the experimental data. The software consists of several tools that can be employed separately from each other or combined together in different ways, providing a seamless interpolation between these three different types of refinement. We use some benchmark cases to show that the combined approach is superior to separately applied refinements. MDRefine has been released as an open-source package under the LGPLv2+ license. Source code, documentation, and examples are available at https://pypi.org/project/MDRefine and https://github.com/bussilab/MDRefine.

DeePMD-kit v3: A Multiple-Backend Framework for Machine Learning Potentials

In recent years, machine learning potentials (MLPs) have become indispensable tools in physics, chemistry, and materials science, driving the development of software packages for molecular dynamics (MD) simulations and related applications. These packages, typically built on specific machine learning frameworks, such as TensorFlow, PyTorch, or JAX, face integration challenges when advanced applications demand communication across different frameworks. The previous TensorFlow-based implementation of the DeePMD-kit exemplified these limitations. In this work, we introduce DeePMD-kit version 3, a significant update featuring a multibackend framework that supports TensorFlow, PyTorch, JAX, and PaddlePaddle backends, and demonstrate the versatility of this architecture through the integration of other MLP packages and of differentiable molecular force fields. This architecture allows seamless back-end switching with minimal modifications, enabling users and developers to integrate DeePMD-kit with other packages using different machine learning frameworks. This innovation facilitates the development of more complex and interoperable workflows, paving the way for broader applications of MLPs in scientific research.

Screening and Design of Aqueous Zinc Battery Electrolytes Based on the Multimodal Optimization of Molecular Simulation

Aqueous batteries, such as aqueous zinc-ion batteries (AZIB), have garnered significant attention because of their advantages in intrinsic safety, low cost, and eco-friendliness. However, aqueous electrolytes tend to freeze at low temperatures, which limits their potential industrial applications. Thus, one of the core challenges in aqueous electrolyte design is optimizing the formula to prevent freezing while maintaining good ion conductivity. However, the experimental trial-and-error approach is inefficient for this purpose, and existing simulation tools are either inaccurate or too expensive for high-throughput phase transition predictions. In this work, we employ a small amount of experimental data and differentiable simulation techniques to develop a multimodal optimization workflow. With minimal human intervention, this workflow significantly enhances the prediction power of classical force fields for electrical conductivity. Most importantly, the simulated electrical conductivity can serve as an effective predictor of electrolyte freezing at low temperatures. Generally, the workflow developed in this work introduces a new paradigm for electrolyte design. This paradigm leverages both easily measurable experimental data and fast simulation techniques to predict properties that are challenging to access by using either approach alone.

Active-Learning Assisted General Framework for Efficient Parameterization of Force-Fields

This work presents an efficient approach to optimizing force field parameters for sulfone molecules using a combination of genetic algorithms (GA) and Gaussian process regression (GPR). Sulfone-based electrolytes are of significant interest in energy storage applications, where accurate modeling of their structural and transport properties is essential. Traditional force field parametrization methods are often computationally expensive and require extensive manual intervention. By integrating GA and GPR, our active learning framework addresses these challenges by achieving optimized parameters in 12 iterations using only 300 data points, significantly outperforming previous attempts requiring thousands of iterations and parameters. We demonstrate the efficiency of our method through a comparison with state-of-the-art techniques, including Bayesian Optimization. The optimized GA-GPR force field was validated against experimental and reference data, including density, viscosity, diffusion coefficients, and surface tension. The results demonstrated excellent agreement between GA-GPR predictions and experimental values, outperforming the widely used OPLS force field. The GA-GPR model accurately captured both bulk and interfacial properties, effectively describing molecular mobility, caging effects, and interfacial arrangements. Furthermore, the transferability of the GA-GPR force field across different temperatures and sulfone structures underscores its robustness and versatility. Our study provides a reliable and transferable force field for sulfone molecules, significantly enhancing the accuracy and efficiency of molecular simulations. This work establishes a strong foundation for future machine learning-driven force field development, applicable to complex molecular systems.

Application of modern artificial intelligence techniques in the development of organic molecular force fields

The molecular force field (FF) determines the accuracy of molecular dynamics (MD) and is one of the major bottlenecks that limits the application of MD in molecular design. Recently, artificial intelligence (AI) techniques, such as machine-learning potentials (MLPs), have been rapidly reshaping the landscape of MD. Meanwhile, organic molecular systems feature unique characteristics, and require more careful treatment in both model construction, optimization, and validation. While an accurate and generic organic molecular force field is still missing, significant progress has been made with the facilitation of AI, warranting a promising future. In this review, we provide an overview of the various types of AI techniques used in molecular FF development and discuss both the advantages and weaknesses of these methodologies. We show how AI methods provide unprecedented capabilities in many tasks such as potential fitting, atom typification, and automatic optimization. Meanwhile, it is also worth noting that more efforts are needed to improve the transferability of the model, develop a more comprehensive database, and establish more standardized validation procedures. With these discussions, we hope to inspire more efforts to solve the existing problems, eventually leading to the birth of next-generation generic organic FFs.

A Neural-Network-Based Mapping and Optimization Framework for High-Precision Coarse-Grained Simulation

The accuracy and efficiency of a coarse-grained (CG) force field are pivotal for high-precision molecular simulations of large systems with complex molecules. We present an automated mapping and optimization framework for molecular simulation (AMOFMS), which is designed to streamline and improve the force field optimization process. It features a neural-network-based mapping function, DSGPM-TP (deep supervised graph partitioning model with type prediction). This model can accurately and efficiently convert atomistic structures to CG mappings, reducing the need for manual intervention. By integrating bottom-up and top-down methodologies, AMOFMS allows users to freely combine these approaches or use them independently as optimization targets. Moreover, users can select and combine different optimizers to meet their specific mission. With its parallel optimizer, AMOFMS significantly accelerates the optimization process, reducing the time required to achieve optimal results. Successful applications of AMOFMS include parameter optimizations for systems such as POPC and PEO, demonstrating its robustness and effectiveness. Overall, AMOFMS provides a general and flexible framework for the automated development of high-precision CG force fields.

Refining potential energy surface through dynamical properties via differentiable molecular simulation

Recently, machine learning potential (MLP) largely enhances the reliability of molecular dynamics, but its accuracy is limited by the underlying ab initio methods. A viable approach to overcome this limitation is to refine the potential by learning from experimental data, which now can be done efficiently using modern automatic differentiation technique. However, potential refinement is mostly performed using thermodynamic properties, leaving the most accessible and informative dynamical data (like spectroscopy) unexploited. In this work, through a comprehensive application of adjoint and gradient truncation methods, we show that both memory and gradient explosion issues can be circumvented in many situations, so the dynamical property differentiation is well-behaved. Consequently, both transport coefficients and spectroscopic data can be used to improve the density functional theory based MLP towards higher accuracy. Essentially, this work contributes to the solution of the inverse problem of spectroscopy by extracting microscopic interactions from vibrational spectroscopic data.

A Refined Set of Universal Force Field Parameters for Some Metal Nodes in Metal-Organic Frameworks

Metal-organic frameworks (MOFs) exhibit promise as porous materials for carbon capture due to their design versatility and large pore sizes. The generic force fields (e.g., UFF and Dreiding) use one universal set of Lennard-Jones parameters for each element, while MOFs have a much richer local chemical environment than those chemical environments used to fit the UFF. When MOFs contain hard-Lewis acid metals, UFF systematically overestimates CO2 uptakes. To address this, we developed a workflow to affordably and efficiently generate reliable force fields to predict CO2 adsorption isotherms of MOFs containing metals from groups IIA (Mg, Ca, Sr, and Ba) and IIIA (Al, Ga, and In), connected to various carboxylate ligands. This method uses experimental isotherms as input. The optimal parameters are obtained by minimizing the loss function of the experimental and simulated isotherms, in which we use the Multistate Bennett Acceptance Ratio (MBAR) theory to derive the functionality relationship of loss functions in terms of force field parameters.

Metal-Organic Frameworks through the Lens of Artificial Intelligence: A Comprehensive Review

Metal-organic frameworks (MOFs) are a class of hybrid porous materials that have gained prominence as a noteworthy material with varied applications. Currently, MOFs are in extensive use, particularly in the realms of energy and catalysis. The synthesis of these materials poses considerable challenges, and their computational analysis is notably intricate due to their complex structure and versatile applications in the field of material science. Density functional theory (DFT) has helped researchers in understanding reactions and mechanisms, but it is costly and time-consuming and requires bigger systems to perform these calculations. Machine learning (ML) techniques were adopted in order to overcome these problems by implementing ML in material data sets for synthesis, structure, and property predictions of MOFs. These predictions are fast, efficient, and accurate and do not require heavy computing. In this review, we discuss ML models used in MOF and their incorporation with artificial intelligence (AI) in structure and property predictions. The advantage of AI in this field would accelerate research, particularly in synthesizing novel MOFs with multiple properties and applications oriented with minimum information.

Synergistic Integration of Physical Embedding and Machine Learning Enabling Precise and Reliable Force Field

Machine-learning force fields have achieved significant strides in accurately reproducing the potential energy surface with quantum chemical accuracy. However, this approach still faces several challenges, e.g., extrapolating to uncharted chemical spaces, interpreting long-range electrostatics, and mapping complex macroscopic properties. To address these issues, we advocate for a synergistic integration of physical principles and machine learning techniques within the framework of a physically informed neural network (PINN). This approach involves incorporating physical knowledge into the parameters of the neural network, coupled with an efficient global optimizer, the Tabu-Adam algorithm, proposed in this work to augment optimization under strict physical constraint. We choose the AMOEBA+ force field as the physics-based model for embedding and then train and test it using the diethylene glycol dimethyl ether (DEGDME) data set as a case study. The results reveal a breakthrough in constructing a precise and noise-robust machine learning force field. Utilizing two training sets with hundreds of samples, our model exhibits remarkable generalization and density functional theory (DFT) accuracy in describing molecular interactions and enables a precise prediction of the macroscopic properties such as the diffusion coefficient with minimal cost. This work provides valuable insight into establishing a fundamental framework of the PINN force field.

Machine-learned molecular mechanics force fields from large-scale quantum chemical data

The development of reliable and extensible molecular mechanics (MM) force fields-fast, empirical models characterizing the potential energy surface of molecular systems-is indispensable for biomolecular simulation and computer-aided drug design. Here, we introduce a generalized and extensible machine-learned MM force field, espaloma-0.3, and an end-to-end differentiable framework using graph neural networks to overcome the limitations of traditional rule-based methods. Trained in a single GPU-day to fit a large and diverse quantum chemical dataset of over 1.1 M energy and force calculations, espaloma-0.3 reproduces quantum chemical energetic properties of chemical domains highly relevant to drug discovery, including small molecules, peptides, and nucleic acids. Moreover, this force field maintains the quantum chemical energy-minimized geometries of small molecules and preserves the condensed phase properties of peptides and folded proteins, self-consistently parametrizing proteins and ligands to produce stable simulations leading to highly accurate predictions of binding free energies. This methodology demonstrates significant promise as a path forward for systematically building more accurate force fields that are easily extensible to new chemical domains of interest.

The Open Force Field Initiative: Open Software and Open Science for Molecular Modeling

Force fields are a key component of physics-based molecular modeling, describing the energies and forces in a molecular system as a function of the positions of the atoms and molecules involved. Here, we provide a review and scientific status report on the work of the Open Force Field (OpenFF) Initiative, which focuses on the science, infrastructure and data required to build the next generation of biomolecular force fields. We introduce the OpenFF Initiative and the related OpenFF Consortium, describe its approach to force field development and software, and discuss accomplishments to date as well as future plans. OpenFF releases both software and data under open and permissive licensing agreements to enable rapid application, validation, extension, and modification of its force fields and software tools. We discuss lessons learned to date in this new approach to force field development. We also highlight ways that other force field researchers can get involved, as well as some recent successes of outside researchers taking advantage of OpenFF tools and data.

Developing a Differentiable Long-Range Force Field for Proteins with E(3) Neural Network-Predicted Asymptotic Parameters

Accurately describing long-range interactions is a significant challenge in molecular dynamics (MD) simulations of proteins. High-quality long-range potential is also an important component of the range-separated machine learning force field. This study introduces a comprehensive asymptotic parameter database encompassing atomic multipole moments, polarizabilities, and dispersion coefficients. Leveraging active learning, our database comprehensively represents protein fragments with up to 8 heavy atoms, capturing their conformational diversity with merely 78,000 data points. Additionally, the E(3) neural network (E3NN) is employed to predict the asymptotic parameters directly from the local geometry. The E3NN models demonstrate exceptional accuracy and transferability across all asymptotic parameters, achieving an R2 of 0.999 for both protein fragments and 20 amino acid dipeptide test sets. The long-range electrostatic and dispersion energies can be obtained using the E3NN-predicted parameters, with an error of 0.07 and 0.02 kcal/mol, respectively, when compared to symmetry-adapted perturbation theory (SAPT). Therefore, our force fields demonstrate the capability to accurately describe long-range interactions in proteins, paving the way for next-generation protein force fields.

FFParam-v2.0: A Comprehensive Tool for CHARMM Additive and Drude Polarizable Force-Field Parameter Optimization and Validation

Developing production quality CHARMM force-field (FF) parameters is a very detailed process involving a variety of calculations, many of which are specific for the molecule of interest. The first version of FFParam was developed as a standalone Python package designed for the optimization of electrostatic and bonded parameters of the CHARMM additive and polarizable Drude FFs by using quantum mechanical (QM) target data. The new version of FFParam has multiple new capabilities for FF parameter optimization and validation, with an emphasis on the ability to use condensed-phase target data in optimization. FFParam-v2 allows optimization of Lennard-Jones (LJ) parameters using potential energy scans of interactions between selected atoms in a molecule and noble gases, viz., He and Ne, and through condensed-phase calculations, from which experimental observables such as heats of vaporization and free energies of solvation may be obtained. This functionality serves as a gold standard for both optimizing parameters and validating the performance of the final parameters. A new bonded parameter optimization algorithm has been introduced to account for simultaneously optimizing multiple molecules sharing parameters. FFParam-v2 also supports the comparison of normal modes and the potential energy distribution of internal coordinates towards each normal mode obtained from QM and molecular mechanics calculations. Such comparison capability is vital to validate the balance among various bonded parameters that contribute to the complex normal modes of molecules. User interaction has been extended beyond the original graphical user interface to include command-line interface capabilities that allow for integration of FFParam in workflows, thereby facilitating the automation of parameter optimization. With these new functionalities, FFParam is a more comprehensive parameter optimization tool for both beginners and advanced users.

Integrating physics in deep learning algorithms: a force field as a PyTorch module

MOTIVATION

Deep learning algorithms applied to structural biology often struggle to converge to meaningful solutions when limited data is available, since they are required to learn complex physical rules from examples. State-of-the-art force-fields, however, cannot interface with deep learning algorithms due to their implementation.

RESULTS

We present MadraX, a forcefield implemented as a differentiable PyTorch module, able to interact with deep learning algorithms in an end-to-end fashion.

AVAILABILITY AND IMPLEMENTATION

MadraX documentation, together with tutorials and installation guide, is available at madrax.readthedocs.io.

PhyNEO: A Neural-Network-Enhanced Physics-Driven Force Field Development Workflow for Bulk Organic Molecule and Polymer Simulations

An accurate, generalizable, and transferable force field plays a crucial role in the molecular dynamics simulations of organic polymers and biomolecules. Conventional empirical force fields often fail to capture precise intermolecular interactions due to their negligence of important physics, such as polarization, charge penetration, many-body dispersion, etc. Moreover, the parameterization of these force fields relies heavily on top-down fittings, limiting their transferabilities to new systems where the experimental data are often unavailable. To address these challenges, we introduce a general and fully ab initio force field construction strategy, named PhyNEO. It features a hybrid approach that combines both the physics-driven and the data-driven methods and is able to generate a bulk potential with chemical accuracy using only quantum chemistry data of very small clusters. Careful separations of long-/short-range interactions and nonbonding/bonding interactions are the key to the success of PhyNEO. By such a strategy, we mitigate the limitations of pure data-driven methods in long-range interactions, thus largely increasing the data efficiency and the scalability of machine learning models. The new approach is thoroughly tested on poly(ethylene oxide) and polyethylene glycol systems, giving superior accuracies in both microscopic and bulk properties compared to conventional force fields. This work thus offers a promising framework for the development of advanced force fields in a wide range of organic molecular systems.