State predictive information bottleneck
Abstract
We report the ability to make sense of the massive amounts of high-dimensional data generated from molecular dynamics simulations is heavily dependent on the knowledge of a low-dimensional manifold (parameterized by a reaction coordinate or RC) that typically distinguishes between relevant metastable states, and which captures the relevant slow dynamics of interest. Methods based on machine learning and artificial intelligence have been proposed over the years to deal with learning such low-dimensional manifolds, but they are often criticized for a disconnect from more traditional and physically interpretable approaches. To deal with such concerns, in this work we propose a deep learning based state predictive information bottleneck approach to learn the RC from high-dimensional molecular simulation trajectories. We demonstrate analytically and numerically how the RC learnt in this approach is connected to the committor in chemical physics and can be used to accurately identify transition states. A crucial hyperparameter in this approach is the time delay or how far into the future the algorithm should make predictions about. Through careful comparisons for benchmark systems, we demonstrate that this hyperparameter choice gives useful control over how coarse-grained we want the metastable state classification of the system to be. We thus believe thatmore »
- Authors:
-
- University of Maryland, College Park, MD (United States)
- Publication Date:
- Research Org.:
- Univ. of Maryland, College Park, MD (United States)
- Sponsoring Org.:
- USDOE Office of Science (SC), Basic Energy Sciences (BES); National Science Foundation (NSF)
- OSTI Identifier:
- 1853699
- Alternate Identifier(s):
- OSTI ID: 1773945; OSTI ID: 1991186
- Grant/Contract Number:
- SC0021009; CHE180027P; ACI-1548562
- Resource Type:
- Accepted Manuscript
- Journal Name:
- Journal of Chemical Physics
- Additional Journal Information:
- Journal Volume: 154; Journal Issue: 13; Journal ID: ISSN 0021-9606
- Publisher:
- American Institute of Physics (AIP)
- Country of Publication:
- United States
- Language:
- English
- Subject:
- 71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSICS; 37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CHEMISTRY; molecular dynamics; probability theory; transition state; chemical physics; Markov processes; artificial neural networks; artificial intelligence; machine learning; peptides; molecular simulations
Citation Formats
Wang, Dedi, and Tiwary, Pratyush. State predictive information bottleneck. United States: N. p., 2021.
Web. doi:10.1063/5.0038198.
Wang, Dedi, & Tiwary, Pratyush. State predictive information bottleneck. United States. https://doi.org/10.1063/5.0038198
Wang, Dedi, and Tiwary, Pratyush. Mon .
"State predictive information bottleneck". United States. https://doi.org/10.1063/5.0038198. https://www.osti.gov/servlets/purl/1853699.
@article{osti_1853699,
title = {State predictive information bottleneck},
author = {Wang, Dedi and Tiwary, Pratyush},
abstractNote = {We report the ability to make sense of the massive amounts of high-dimensional data generated from molecular dynamics simulations is heavily dependent on the knowledge of a low-dimensional manifold (parameterized by a reaction coordinate or RC) that typically distinguishes between relevant metastable states, and which captures the relevant slow dynamics of interest. Methods based on machine learning and artificial intelligence have been proposed over the years to deal with learning such low-dimensional manifolds, but they are often criticized for a disconnect from more traditional and physically interpretable approaches. To deal with such concerns, in this work we propose a deep learning based state predictive information bottleneck approach to learn the RC from high-dimensional molecular simulation trajectories. We demonstrate analytically and numerically how the RC learnt in this approach is connected to the committor in chemical physics and can be used to accurately identify transition states. A crucial hyperparameter in this approach is the time delay or how far into the future the algorithm should make predictions about. Through careful comparisons for benchmark systems, we demonstrate that this hyperparameter choice gives useful control over how coarse-grained we want the metastable state classification of the system to be. We thus believe that this work represents a step forward in systematic application of deep learning based ideas to molecular simulations.},
doi = {10.1063/5.0038198},
journal = {Journal of Chemical Physics},
number = 13,
volume = 154,
place = {United States},
year = {Mon Apr 05 00:00:00 EDT 2021},
month = {Mon Apr 05 00:00:00 EDT 2021}
}
Works referenced in this record:
Accurate sampling using Langevin dynamics
journal, May 2007
- Bussi, Giovanni; Parrinello, Michele
- Physical Review E, Vol. 75, Issue 5
A Variational Approach to Modeling Slow Processes in Stochastic Dynamical Systems
journal, January 2013
- Noé, Frank; Nüske, Feliks
- Multiscale Modeling & Simulation, Vol. 11, Issue 2
Determination of reaction coordinates via locally scaled diffusion map
journal, March 2011
- Rohrdanz, Mary A.; Zheng, Wenwei; Maggioni, Mauro
- The Journal of Chemical Physics, Vol. 134, Issue 12
Metadynamics Enhanced Markov Modeling of Protein Dynamics
journal, January 2018
- Biswas, Mithun; Lickert, Benjamin; Stock, Gerhard
- The Journal of Physical Chemistry B, Vol. 122, Issue 21
Variational encoding of complex dynamics
journal, June 2018
- Hernández, Carlos X.; Wayment-Steele, Hannah K.; Sultan, Mohammad M.
- Physical Review E, Vol. 97, Issue 6
Canonical sampling through velocity rescaling
journal, January 2007
- Bussi, Giovanni; Donadio, Davide; Parrinello, Michele
- The Journal of Chemical Physics, Vol. 126, Issue 1
Describing Protein Folding Kinetics by Molecular Dynamics Simulations. 1. Theory †
journal, May 2004
- Swope, William C.; Pitera, Jed W.; Suits, Frank
- The Journal of Physical Chemistry B, Vol. 108, Issue 21
Reweighted autoencoded variational Bayes for enhanced sampling (RAVE)
journal, August 2018
- Ribeiro, João Marcelo Lamim; Bravo, Pablo; Wang, Yihang
- The Journal of Chemical Physics, Vol. 149, Issue 7
Kinetic Pathways of Ion Pair Dissociation in Water
journal, May 1999
- Geissler, Phillip L.; Dellago, Christoph; Chandler, David
- The Journal of Physical Chemistry B, Vol. 103, Issue 18
Reaction Rate Theory in Coordination Number Space: An Application to Ion Solvation
journal, March 2016
- Roy, Santanu; Baer, Marcel D.; Mundy, Christopher J.
- The Journal of Physical Chemistry C, Vol. 120, Issue 14
Obtaining reaction coordinates by likelihood maximization
journal, August 2006
- Peters, Baron; Trout, Bernhardt L.
- The Journal of Chemical Physics, Vol. 125, Issue 5
Using metadynamics to explore complex free-energy landscapes
journal, March 2020
- Bussi, Giovanni; Laio, Alessandro
- Nature Reviews Physics, Vol. 2, Issue 4
T RANSITION P ATH S AMPLING : Throwing Ropes Over Rough Mountain Passes, in the Dark
journal, October 2002
- Bolhuis, Peter G.; Chandler, David; Dellago, Christoph
- Annual Review of Physical Chemistry, Vol. 53, Issue 1
GROMACS: High performance molecular simulations through multi-level parallelism from laptops to supercomputers
journal, September 2015
- Abraham, Mark James; Murtola, Teemu; Schulz, Roland
- SoftwareX, Vol. 1-2
Potential energy surface for a seven-atom reaction. Thermal rate constants and kinetic isotope effects for CH4+OH
journal, April 2000
- Espinosa-Garcı́a, J.; Corchado, J. C.
- The Journal of Chemical Physics, Vol. 112, Issue 13
Author Correction: VAMPnets for deep learning of molecular kinetics
journal, October 2018
- Mardt, Andreas; Pasquali, Luca; Wu, Hao
- Nature Communications, Vol. 9, Issue 1
Markov State Model Reveals Folding and Functional Dynamics in Ultra-Long MD Trajectories
journal, November 2011
- Lane, Thomas J.; Bowman, Gregory R.; Beauchamp, Kyle
- Journal of the American Chemical Society, Vol. 133, Issue 45
Reaction coordinates of biomolecular isomerization
journal, May 2000
- Bolhuis, P. G.; Dellago, C.; Chandler, D.
- Proceedings of the National Academy of Sciences, Vol. 97, Issue 11
Calculating rate constants and committor probabilities for transition networks by graph transformation
journal, May 2009
- Wales, David J.
- The Journal of Chemical Physics, Vol. 130, Issue 20
Galerkin approximation of dynamical quantities using trajectory data
journal, June 2019
- Thiede, Erik H.; Giannakis, Dimitrios; Dinner, Aaron R.
- The Journal of Chemical Physics, Vol. 150, Issue 24
GROMACS: A message-passing parallel molecular dynamics implementation
journal, September 1995
- Berendsen, H. J. C.; van der Spoel, D.; van Drunen, R.
- Computer Physics Communications, Vol. 91, Issue 1-3
Past–future information bottleneck for sampling molecular reaction coordinate simultaneously with thermodynamics and kinetics
journal, August 2019
- Wang, Yihang; Ribeiro, João Marcelo Lamim; Tiwary, Pratyush
- Nature Communications, Vol. 10, Issue 1
Automatic Method for Identifying Reaction Coordinates in Complex Systems †
journal, April 2005
- Ma, Ao; Dinner, Aaron R.
- The Journal of Physical Chemistry B, Vol. 109, Issue 14
PLUMED 2: New feathers for an old bird
journal, February 2014
- Tribello, Gareth A.; Bonomi, Massimiliano; Branduardi, Davide
- Computer Physics Communications, Vol. 185, Issue 2
Reaction Coordinates and Mechanistic Hypothesis Tests
journal, May 2016
- Peters, Baron
- Annual Review of Physical Chemistry, Vol. 67, Issue 1
Spectral gap optimization of order parameters for sampling complex molecular systems
journal, February 2016
- Tiwary, Pratyush; Berne, B. J.
- Proceedings of the National Academy of Sciences, Vol. 113, Issue 11
Representation Learning: A Review and New Perspectives
journal, August 2013
- Bengio, Y.; Courville, A.; Vincent, P.
- IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 35, Issue 8
Reaction coordinates and rates from transition paths
journal, April 2005
- Best, R. B.; Hummer, G.
- Proceedings of the National Academy of Sciences, Vol. 102, Issue 19
Diffusion maps, spectral clustering and reaction coordinates of dynamical systems
journal, July 2006
- Nadler, Boaz; Lafon, Stéphane; Coifman, Ronald R.
- Applied and Computational Harmonic Analysis, Vol. 21, Issue 1, p. 113-127
Diffusion Maps, Reduction Coordinates, and Low Dimensional Representation of Stochastic Systems
journal, January 2008
- Coifman, R. R.; Kevrekidis, I. G.; Lafon, S.
- Multiscale Modeling & Simulation, Vol. 7, Issue 2
Machine learning approaches for analyzing and enhancing molecular dynamics simulations
journal, April 2020
- Wang, Yihang; Lamim Ribeiro, João Marcelo; Tiwary, Pratyush
- Current Opinion in Structural Biology, Vol. 61
Learning reaction coordinates via cross-entropy minimization: Application to alanine dipeptide
journal, August 2020
- Mori, Yusuke; Okazaki, Kei-ichi; Mori, Toshifumi
- The Journal of Chemical Physics, Vol. 153, Issue 5
Dependence of the Rate of LiF Ion-Pairing on the Description of Molecular Interaction
journal, November 2015
- Pluhařová, Eva; Baer, Marcel D.; Schenter, Gregory K.
- The Journal of Physical Chemistry B, Vol. 120, Issue 8
Transition path sampling and the calculation of rate constants
journal, February 1998
- Dellago, Christoph; Bolhuis, Peter G.; Csajka, Félix S.
- The Journal of Chemical Physics, Vol. 108, Issue 5
Transition networks for modeling the kinetics of conformational change in macromolecules
journal, April 2008
- Noé, Frank; Fischer, Stefan
- Current Opinion in Structural Biology, Vol. 18, Issue 2