TorchMD: A Deep Learning Framework for Molecular Simulations (2020-12-22T00:00:00.000000Z)

TL;DR

TorchMD is presented, a framework for molecular simulations with mixed classical and machine learning potentials that enables learning and simulating neural network potentials and provides a useful tool set to support molecular simulations of machinelearning potentials.

Abstract

Molecular dynamics simulations provide a mechanistic description of molecules by relying on empirical potentials. The quality and transferability of such potentials can be improved leveraging data-driven models derived with machine learning approaches. Here, we present TorchMD, a framework for molecular simulations with mixed classical and machine learning potentials. All force computations including bond, angle, dihedral, Lennard-Jones, and Coulomb interactions are expressed as PyTorch arrays and operations. Moreover, TorchMD enables learning and simulating neural network potentials. We validate it using standard Amber all-atom simulations, learning an ab initio potential, performing an end-to-end training, and finally learning and simulating a coarse-grained model for protein folding. We believe that TorchMD provides a useful tool set to support molecular simulations of machine learning potentials. Code and data are freely available at github.com/torchmd.

Authors

F. Noé

6 papers

Stefan Doerr

1 papers

Maciej Majewski

1 papers

TL;DR

Abstract

Authors

References57 items

End-to-End Differentiable Molecular Mechanics Force Field Construction

Temperature-transferable coarse-graining of ionic liquids with dual graph convolutional neural networks.

Coarse graining molecular dynamics with graph neural networks.

Active learning and neural network potentials accelerate molecular screening of ether-based solvate ionic liquids.

Ensemble learning of coarse-grained molecular dynamics force fields with a kernel approach.

TorchANI: A Free and Open Source PyTorch-Based Deep Learning Implementation of the ANI Neural Network Potentials

JAX, M.D.: End-to-End Differentiable, Hardware Accelerated, Molecular Dynamics in Pure Python

FCHL revisited: Faster and more accurate quantum machine learning.

SciPy 1.0: fundamental algorithms for scientific computing in Python

Coarse-graining molecular systems by spectral matching.

GROUNDS OF MODERN MODELS AND SYSTEMS OF ORGANIZATIONAL CREATIVITY SUPPORT

The SIRAH 2.0 Force Field: Altius, Fortius, Citius.

Coarse-graining auto-encoders for molecular dynamics

Machine Learning of Coarse-Grained Molecular Dynamics Force Fields

Variational selection of features for molecular kinetics.

SchNetPack: A Deep Learning Toolbox For Atomistic Systems.

Faculty Opinions recommendation of OpenMM 7: Rapid development of high performance algorithms for molecular dynamics.

Open Force Field Consortium: Escaping atom types using direct chemical perception with SMIRNOFF v0.1

Automated design of collective variables using supervised machine learning.

DeePCG: Constructing coarse-grained models via deep neural networks.

SchNet - A deep learning architecture for molecules and materials.

The TensorMol-0.1 model chemistry: a neural network augmented with long-range physics

Drug Discovery and Molecular Dynamics: Methods, Applications and Perspective Beyond the Second Timescale.

SchNet: A continuous-filter convolutional neural network for modeling quantum interactions

Modeling the Mechanism of CLN025 Beta-Hairpin Formation

OpenMM 7: Rapid development of high performance algorithms for molecular dynamics

Optimized parameter selection reveals trends in Markov state models for protein folding.

Lessons learned from comparing molecular dynamics engines on the SAMPL5 dataset

Convolutional Networks on Graphs for Learning Molecular Fingerprints

ff14SB: Improving the Accuracy of Protein Side Chain and Backbone Parameters from ff99SB.

Adam: A Method for Stochastic Optimization

Quantum chemistry structures and properties of 134 kilo molecules

On-the-Fly Learning and Sampling of Ligand Binding by High-Throughput Molecular Simulations.

Perspective: Coarse-grained models for biomolecular systems.

Perspective on the Martini model.

Coarse-graining methods for computational biology.

Identification of slow molecular order parameters for Markov model construction.

Simple few-state models reveal hidden complexity in protein folding

How Fast-Folding Proteins Fold

How robust are protein folding simulations with respect to force field parameterization?

High-Throughput All-Atom Molecular Dynamics Simulations Using Distributed Computing

Discovery through the computational microscope.

ACEMD: Accelerating Biomolecular Dynamics in the Microsecond Time Scale.

Crystal structure of a ten-amino acid protein.

Coarse-grained models of protein folding: toy models or predictive tools?

A multiscale coarse-graining method for biomolecular systems.

Improving efficiency of large time‐scale molecular dynamics simulations of hydrogen‐rich systems

A generalized reaction field method for molecular dynamics simulations

The time machine.

Comparison of simple potential functions for simulating liquid water

Advances in Protein Chemistry

Di ff erentiable molecular simulation can learn all the parameters in a coarse - grained force fi eld for proteins

End-to-End Differentiable , Hardware Accelerated , Molecular Dynamics in Pure Python

PyTorch Light-ning

JAX: composable transformations of Python+NumPy programs

[Advances in Protein Chemistry] Protein Simulations Volume 66 || Free Energy Calculations and Ligand Binding

The Nature of Statistical Learning Theory

Field of Study

Journal Information

Name

Page

Volume

Venue Information

Name

Type

URL

Alternate Names