Analysis and optimization of weighted ensemble sampling

Aristoff, David

doi:10.1051/m2an/2017046

Cited by 19 publications

(37 citation statements)

References 28 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Future studies will clearly benefit from variance-reduction strategies, which have been proposed. [95][96] The weighted ensemble method was chosen over other rigorous path sampling approaches 10,[27][28][29][30][31][47][48][49][50][51][52] and standard (history-independent) Markov state models (MSMs). [97][98] Compared to other path sampling methods, WE offers fully scalable parallelization and does not require hard-coding within the dynamics engine in order to "catch" trajectories as they cross interfaces.…”

Section: Discussionmentioning

confidence: 99%

Computational Estimation of Microsecond to Second Atomistic Folding Times

et al. 2019

View full text Add to dashboard Cite

Despite the development of massively parallel computing hardware including inexpensive graphics processing units (GPUs), it has remained infeasible to simulate the folding of atomistic proteins at room temperature using conventional molecular dynamics (MD) beyond the μs scale. Here we report the folding of atomistic, implicitly solvated protein systems with folding times τ f ranging from ~10 μs to ~100 ms using the weighted ensemble (WE) strategy in combination with GPU computing. Starting from an initial structure or set of structures, WE organizes an ensemble of GPU-accelerated MD trajectory segments via intermittent pruning and replication events to generate statistically unbiased estimates of rate constants for rare events such as folding; no biasing forces are used. Although the variance among atomistic WE folding runs is significant, multiple independent runs are used to reduce and quantify statistical uncertainty. Folding times are estimated directly from WE probability flux and from history-augmented Markov analysis of the WE data. Three systems were examined: NTL9 at low solvent viscosity (yielding τ f = 0.8 − μs), NTL9 at water-like viscosity (τ f = 0.2 − 2 ms), and Protein G at low viscosity (τ f = 3 − 200 ms). In all cases the folding time, uncertainty, and ensemble properties could be estimated from WE simulation; for Protein G, this characterization required significantly less overall computing than would be required to observe a single folding event with conventional MD simulations. Our results suggest that the use and calibration of force fields and solvent models for precise estimation of kinetic quantities is becoming feasible.

show abstract

Section: Discussionmentioning

confidence: 99%

Computational Estimation of Microsecond to Second Atomistic Folding Times

et al. 2019

View full text Add to dashboard Cite

show abstract

“…Future studies will clearly benefit from variance-reduction strategies, which have been proposed. [94][95] The weighted ensemble method was chosen over other rigorous path sampling approaches 10,[27][28][29][30][31][47][48][49][50][51][52] and standard (history-independent) Markov state models (MSMs). [96][97] Compared to other path sampling methods, WE offers fully scalable parallelization and does not require hard-coding within the dynamics engine in order to "catch" trajectories as they cross interfaces.…”

Section: Discussionmentioning

confidence: 99%

Computational estimation of ms-sec atomistic folding times

Adhikari

Mostofian

Copperman

et al. 2018

Preprint

View full text Add to dashboard Cite

Despite the development of massively parallel computing hardware including inexpensive graphics processing units (GPUs), it has remained infeasible to simulate the folding of atomistic proteins at room temperature using conventional molecular dynamics (MD) beyond the µs scale.Here we report the folding of atomistic, implicitly solvated protein systems with folding times f ranging from ~100 µs to ~1s using the weighted ensemble (WE) strategy in combination with GPU computing. Starting from an initial structure or set of structures, WE organizes an ensemble of GPU-accelerated MD trajectory segments via intermittent pruning and replication events to generate statistically unbiased estimates of rate constants for rare events such as folding; no biasing forces are used. Although the variance among atomistic WE folding runs is significant, multiple independent runs are used to reduce and quantify statistical uncertainty. Folding times are estimated directly from WE probability flux and from history-augmented Markov analysis of the WE data. Three systems were examined: NTL9 at low solvent viscosity (yielding f = 0.8 − 9.0 s), NTL9 at water-like viscosity ( f = 0.2 − 1.9 ms), and Protein G at low viscosity ( f = 3.3 − 200 ms). In all cases the folding time, uncertainty, and ensemble properties could be estimated from WE simulation; for Protein G, this characterization required significantly less overall computing than would be required to observe a single folding event with conventional MD simulations. Our results suggest that the use and calibration of force fields and solvent models for precise estimation of kinetic quantities is becoming feasible.

show abstract

“…To this end we employ weighted ensemble simulation for the α subset: trajectories initiated at A which subsequently arrive at the absorbing sink at B are regenerated at A. WE provides an unbiased representation of the α reactive trajectory ensemble 14,15 .…”

Section: Theory and Proceduresmentioning

confidence: 99%

Accelerated Estimation of Long-timescale Kinetics by Combining Weighted Ensemble Simulation with Markov Model “Microstates” using Non-Markovian Theory

Copperman

Zuckerman

2020

Biophysical Journal

View full text Add to dashboard Cite

The weighted ensemble (WE) simulation strategy can provide unbiased sampling of nonequilibrium processes, such as molecular folding or binding. Unbiased kinetic rates can be extracted from any discrete clustering of the configuration space based on a historyaugmented Markov state model (haMSM) at any lag time, in the steady-state. However, the convergence of WE to steady-state may require unaffordably long simulations in complex systems. Here we show that by clustering molecular configurations into many (thousands of) microbins using methods developed in the Markov State Modeling (MSM) community, unbiased kinetics can be obtained from WE data using history-augmented Markov State Models (haMSMs) before SS convergence of the WE simulation itself. Because arbitrarily small lag times can be used within the exact haMSM formulation, accurate kinetics can be obtained with significantly less trajectory data than traditional MSMs, while bypassing the often prohibitive convergence requirements of the non-equilibrium weighted ensemble. We validate the method in a simple diffusive process on a 2D random energy landscape, and apply the method to atomistic protein folding simulations using WE molecular dynamics.We report significant progress towards the unbiased estimation of protein folding times and pathways, though key challenges remain.

show abstract

Analysis and optimization of weighted ensemble sampling

Cited by 19 publications

References 28 publications

Computational Estimation of Microsecond to Second Atomistic Folding Times

Computational Estimation of Microsecond to Second Atomistic Folding Times

Computational estimation of ms-sec atomistic folding times

Accelerated Estimation of Long-timescale Kinetics by Combining Weighted Ensemble Simulation with Markov Model “Microstates” using Non-Markovian Theory

Contact Info

Product

Resources

About