VAMPnets for deep learning of molecular kinetics

By Andreas Mardt, Luca Pasquali, Hao Wu, Frank No'e. 2017

682 citations recorded by the source corpus.

There is an increasing demand for computing the relevant structures, equilibria, and long-timescale kinetics of biomolecular processes, such as protein-drug binding, from high-throughput molecular dynamics simulations. Current methods employ transformation of simulated coordinates into structural features, dimension reduction, clustering the dimension-reduced data, and estimation of a Markov state model or related model of the interconversion rates between molecular structures. This handcrafted approach demands a substantial amount of modeling expertise, as poor decisions at any step will lead to large modeling errors. Here we employ the variational approach for Markov processes (VAMP) to develop a deep learning framework for molecular kinetics using neural networks, dubbed VAMPnets. A VAMPnet encodes the entire mapping from molecular coordinates to Markov states, thus combining the whole data processing pipeline in a single end-to-end framework. Our method performs equally or better than state-of-the-art Markov modeling methods and provides easily interpretable few-state kinetic models. Extracting kinetic models from high-throughput molecular dynamics (MD) simulations is laborious and prone to human error. Here the authors introduce a deep learning framework that automates construction of Markov state models from MD simulation data.

Subfields: Materials Chemistry, Molecular Biology, Statistical and Nonlinear Physics

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Materials Chemistry

In Molecular Biology

In Statistical and Nonlinear Physics

Citations in the corpus

Deep Generative Markov State Models (2018, methodology): “One 250 ns trajectory with a storage interval of 1 ps is used and split 80%/20% for training and validation – see [15] for details of the simulation setup.”
Learning Parameters and Constitutive Relationships with Physics Informed Deep Neural Networks (2018, background): “Recently, data-driven approximations of Koopman operators, including dynamic mode decomposition [26], diﬀusion maps [26], delay co-ordinates [8, 4, 1], and neural networks [27, 23, 15, 14, 17], have gained signiﬁcant attention.”
Integrating Scientific Knowledge with Machine Learning for Engineering and Environmental Systems (2020, methodology): “Though dynamic mode decomposition [197] is the most common technique for approximating the Koopman operator, many recent approaches have been made to approximate Koopman operator embeddings with deep learning models that outperform existing methods [170, 185, 191, 200, 209, 210, 260, 284, 307].”
Data-driven discovery of coordinates and governing equations (2019, methodology): “Numerous recent approaches leverage machine learning to identify dynamical systems models from data [19–26], with many using neural networks to model time-series data [18, 27–38].”
Exploring chemical compound space with quantum-based machine learning (2019, background): “evidence has been presented that it is equally possible to build ML models of SM ensemble averages (integrating out the atomistic configurational degrees of freedom), such as free energy, entropy, or kinetic pathways [37, 38].”
Variational Approach for Learning Markov Processes from Time Series Data (2017, methodology): “One feasible way of addressing this challenge is to approximate singular components by deep neural networks, which yields the concept of VAMPnet [33].”
Integrating Physics-Based Modeling with Machine Learning: A Survey (2020, methodology): “In molecular kinetics, reduced order modeling frameworks are used for approximating the Koopman operator [171, 262].”
Machine learning for molecular simulation (2019, methodology): “In order to obtain a propagator that can be interpreted as a Markov state model, [21] chose to use a SoftMax layer as an output layer, thus transforming the spectral representation to a soft indicator function similar to spectral clustering methods such as PCCA+ [123, 124].”