Search Menu
DOE PAGES title logo U.S. Department of Energy
Office of Scientific and Technical Information
Search Scholarly Publications

Title: Transferable Dynamic Molecular Charge Assignment Using Deep Neural Networks

Journal Article · · Journal of Chemical Theory and Computation
ORCiD logo [1]; ORCiD logo [1]; ORCiD logo [2];  [3]; ORCiD logo [1];  [4];  [5]; ORCiD logo [1]; ORCiD logo [1]
  1. Los Alamos National Lab. (LANL), Los Alamos, NM (United States)
  2. Los Alamos National Lab. (LANL), Los Alamos, NM (United States); Univ. of Florida, Gainesville, FL (United States)
  3. Los Alamos National Lab. (LANL), Los Alamos, NM (United States); Univ. of Southern California, Los Angeles, CA (United States)
  4. Univ. of North Carolina, Chapel Hill, NC (United States)
  5. Univ. of Florida, Gainesville, FL (United States)
+ Show Author Affiliations

We report that the ability to accurately and efficiently compute quantum-mechanical partial atomistic charges has many practical applications, such as calculations of IR spectra, analysis of chemical bonding, and classical force field parametrization. Machine learning (ML) techniques provide a possible avenue for the efficient prediction of atomic partial charges. Modern ML advances in the prediction of molecular energies [i.e., the hierarchical interacting particle neural network (HIP-NN)] has provided the necessary model framework and architecture to predict transferable, extensible, and conformationally dynamic atomic partial charges based on reference density functional theory (DFT) simulations. Utilizing HIP-NN, we show that ML charge prediction can be highly accurate over a wide range of molecules (both small and large) across a variety of charge partitioning schemes such as the Hirshfeld, CM5, MSK, and NBO methods. To demonstrate transferability and size extensibility, we compare ML results with reference DFT calculations on the COMP6 benchmark, achieving errors of 0.004e– (elementary charge). This is remarkable since this benchmark contains two proteins that are multiple times larger than the largest molecules in the training set. An application of our atomic charge predictions on nonequilibrium geometries is the generation of IR spectra for organic molecules from dynamical trajectories on a variety of organic molecules, which show good agreement with calculated IR spectra with reference method. Critically, HIP-NN charge predictions are many orders of magnitude faster than direct DFT calculations. Lastly, these combined results provide further evidence that ML (specifically HIP-NN) provides a pathway to greatly increase the range of feasible simulations while retaining quantum-level accuracy.

Research Organization:
Los Alamos National Laboratory (LANL)
Sponsoring Organization:
USDOE National Nuclear Security Administration (NNSA); Laboratory Directed Research and Development (LDRD)
Grant/Contract Number:
AC52-06NA25396
OSTI ID:
1467336
Report Number(s):
LA-UR-18-22005
Journal Information:
Journal of Chemical Theory and Computation, Journal Name: Journal of Chemical Theory and Computation Journal Issue: 9 Vol. 14; ISSN 1549-9618
Publisher:
American Chemical SocietyCopyright Statement
Country of Publication:
United States
Language:
English

References (74)

Atom-centered symmetry functions for constructing high-dimensional neural network potentials journal February 2011
Symmetry-Adapted Machine Learning for Tensorial Properties of Atomistic Systems dataset January 2018
Machine Learning of Partial Charges Derived from High-Quality Quantum-Mechanical Calculations text January 2018
Active learning of linearly parametrized interatomic potentials text January 2016
Machine Learning Predictions of Molecular Properties: Accurate Many-Body Potentials and Nonlocality in Chemical Space text January 2015
Virtual Exploration of the Small-Molecule Chemical Universe below 160 Daltons journal February 2005
An approach to computing electrostatic charges for molecules journal April 1984
Constructing high-dimensional neural network potentials: A tutorial review journal March 2015
Fourier series of atomic radial distribution functions: A molecular fingerprint for machine learning models of quantum chemical properties
  • von Lilienfeld, O. Anatole; Ramakrishnan, Raghunathan; Rupp, Matthias
  • International Journal of Quantum Chemistry, Vol. 115, Issue 16 https://doi.org/10.1002/qua.24912
journal April 2015
Asparagine and glutamine: using hydrogen atom contacts in the choice of side-chain amide orientation 1 1Edited by J. Thornton journal January 1999
Bonded-atom fragments for describing molecular charge densities journal January 1977
Molecular graph convolutions: moving beyond fingerprints journal August 2016
Multi-fidelity machine learning models for accurate bandgap predictions of solids journal March 2017
Active learning of linearly parametrized interatomic potentials journal December 2017
Spectral neighbor analysis method for automated generation of quantum-accurate interatomic potentials journal March 2015
Automatic atom type and bond type perception in molecular mechanical calculations journal October 2006
Can the electronegativity equalization method predict spectroscopic properties? journal February 2015
10 Residue Folded Peptide Designed by Segment Statistics journal August 2004
Machine Learning for Silver Nanoparticle Electron Transfer Property Prediction journal October 2017
Machine Learning of Partial Charges Derived from High-Quality Quantum-Mechanical Calculations journal February 2018
Prediction Errors of Molecular Machine Learning Models Lower than Hybrid DFT Error journal October 2017
Resolving Transition Metal Chemical Space: Feature Selection for Machine Learning and Structure–Property Relationships journal November 2017
Machine Learning Predictions of Molecular Properties: Accurate Many-Body Potentials and Nonlocality in Chemical Space journal June 2015
Genetic Optimization of Training Sets for Improved Machine Learning Models of Molecular Properties journal March 2017
A Universal Approach to Solvation Modeling journal June 2008
Automation of the CHARMM General Force Field (CGenFF) II: Assignment of Bonded Parameters and Partial Atomic Charges journal November 2012
Virtual Exploration of the Chemical Universe up to 11 Atoms of C, N, O, F:  Assembly of 26.4 Million Structures (110.9 Million Stereoisomers) and Analysis for New Ring Systems, Stereochemistry, Physicochemical Properties, Compound Classes, and Drug Discovery journal January 2007
An Automated Force Field Topology Builder (ATB) and Repository: Version 1.0 journal October 2011
Charge Model 5: An Extension of Hirshfeld Population Analysis for the Accurate Description of Molecular Interactions in Gaseous and Condensed Phases
  • Marenich, Aleksandr V.; Jerome, Steven V.; Cramer, Christopher J.
  • Journal of Chemical Theory and Computation, Vol. 8, Issue 2, p. 527-541 https://doi.org/10.1021/ct200866d
journal February 2012
970 Million Druglike Small Molecules for Virtual Screening in the Chemical Universe Database GDB-13 journal July 2009
Intermolecular Interactions in Complex Liquids: Effective Fragment Potential Investigation of Water– tert -Butanol Mixtures journal February 2012
Quantum-chemical insights from deep tensor neural networks journal January 2017
Designing a 20-residue protein journal April 2002
Quantum chemistry structures and properties of 134 kilo molecules journal August 2014
ANI-1, A data set of 20 million calculated off-equilibrium conformations for organic molecules journal December 2017
Machine learning exciton dynamics journal January 2016
The S66x8 benchmark for noncovalent interactions revisited: explicitly correlated ab initio methods and density functional theory journal January 2016
ANI-1: an extensible neural network potential with DFT accuracy at force field computational cost journal January 2017
Addressing uncertainty in atomistic machine learning journal January 2017
Machine learning molecular dynamics for the simulation of infrared spectra journal January 2017
Machine learning for quantum dynamics: deep learning of excitation energy transfer properties journal January 2017
The TensorMol-0.1 model chemistry: a neural network augmented with long-range physics journal January 2018
Electronic Population Analysis on LCAO–MO Molecular Wave Functions. I journal October 1955
Systematic optimization of long-range corrected hybrid density functionals journal February 2008
The electronegativity equalization method and the split charge equilibration applied to organic systems: Parametrization, validation, and comparison journal July 2009
Atom-centered symmetry functions for constructing high-dimensional neural network potentials journal February 2011
A neural network potential-energy surface for the water dimer based on environment-dependent atomic energies and charges journal February 2012
Self‐consistent molecular orbital methods. XXIII. A polarization‐type basis set for second‐row elements journal October 1982
Natural population analysis journal July 1985
Perspective: Fifty years of density-functional theory in chemical physics journal May 2014
Structure-based sampling and self-correcting machine learning for accurate calculations of potential energy surfaces and vibrational levels journal June 2017
Non-covalent interactions across organic and biological subsets of chemical space: Physics-based potentials parametrized from machine learning journal June 2018
Hierarchical modeling of molecular energies using a deep neural network journal June 2018
Maximally resolved anharmonic OH vibrational spectrum of the water/ZnO(101¯0) interface from a high-dimensional neural network potential journal June 2018
SchNet – A deep learning architecture for molecules and materials journal June 2018
Less is more: Sampling chemical space with active learning journal June 2018
Molecular dynamics simulation of infrared spectra and average structure of benzoic acid crystal journal June 1988
High-throughput and data mining with ab initio methods journal December 2004
Machine learning of molecular electronic properties in chemical compound space journal September 2013
DrugBank 4.0: shedding new light on drug metabolism journal November 2013
High-dimensional neural-network potentials for multicomponent systems: Applications to zinc oxide journal April 2011
How to represent crystal structures for machine learning: Towards fast prediction of electronic properties journal May 2014
Accelerated materials property predictions and design using motif-based fingerprints journal July 2015
Interatomic potentials for ionic systems with density functional accuracy based on charge densities obtained by a neural network journal July 2015
Fast and Accurate Modeling of Molecular Atomization Energies with Machine Learning journal January 2012
Symmetry-Adapted Machine Learning for Tensorial Properties of Atomistic Systems journal January 2018
Generalized Neural-Network Representation of High-Dimensional Potential-Energy Surfaces journal April 2007
Density functional theory: Its origins, rise to prominence, and future journal August 2015
Structure of Glucoamylase from Saccharomycopsis fibuligera at 1.7 Å Resolution journal September 1998
Moment Tensor Potentials: A Class of Systematically Improvable Interatomic Potentials journal January 2016
Query by committee conference January 1992
AtomicChargeCalculator: interactive web-based calculation of atomic charges in large biomolecular complexes and drug-like molecules journal October 2015
AtomicChargeCalculator: interactive web-based calculation of atomic charges in large biomolecular complexes and drug-like molecules collection January 2015
Quantum chemistry structures and properties of 134 kilo molecules collection January 2019

Cited By (12)

Deep learning for UV absorption spectra with SchNarc: First steps toward transferability in chemical compound space journal October 2020
Neural Network Potentials: A Concise Overview of Methods journal April 2022
Synergistic Approach of Ultrafast Spectroscopy and Molecular Simulations in the Characterization of Intramolecular Charge Transfer in Push-Pull Molecules journal January 2020
Design, Parameterization, and Implementation of Atomic Force Fields for Adsorption in Nanoporous Materials journal September 2019
Approaching coupled cluster accuracy with a general-purpose neural network potential through transfer learning journal July 2019
Unexpectedly high cross-plane thermoelectric performance of layered carbon nitrides journal January 2019
Operators in quantum machine learning: Response properties in chemical space journal February 2019
Compressing physics with an autoencoder: Creating an atomic species representation to improve machine learning models in the chemical sciences journal August 2019
Incorporating long-range physics in atomic-scale machine learning journal November 2019
Dynamical matrix propagator scheme for large-scale proton dynamics simulations journal March 2020
Synergistic Approach of Ultrafast Spectroscopy and Molecular Simulations in the Characterization of Intramolecular Charge Transfer in Push-Pull Molecules journal January 2020
Incorporating long-range physics in atomic-scale machine learning text January 2019

Similar Records

Lightweight and effective tensor sensitivity for atomistic neural networks
Journal Article · 2023 · Journal of Chemical Physics · OSTI ID:2441369
+4 more
Bond order predictions using deep neural networks
Journal Article · 2021 · Journal of Applied Physics · OSTI ID:1805727
+2 more
Numerical study on the partitioning of the molecular polarizability into fluctuating charge and induced atomic dipole contributions
Journal Article · 2015 · Journal of Physical Chemistry. A, Molecules, Spectroscopy, Kinetics, Environment, and General Theory · OSTI ID:1344918
+3 more

Related Subjects

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CHEMISTRY
97 MATHEMATICS AND COMPUTING