Computing committors in collective variables via Mahalanobis diffusion maps

Evans, Luke; Cameron, Maria K.; Tiwary, Pratyush

doi:10.1016/j.acha.2023.01.001

Cited by 8 publications

(9 citation statements)

References 62 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Equation ( 31) is approximately the average time the systems spends to commute between x k and x l , and the associated commute map is given by: where we can see the difference between the diffusion and commute distances are in the coefficients λ n and t n /2, respectively. Various methods exploit the relation of the effective timescale with eigenvalues [9,61,94,137,138,[145][146][147][148][149].…”

Section: Diffusion and Commute Distancesmentioning

confidence: 99%

Manifold learning in atomistic simulations: a conceptual review

Rydzewski

Chen

Valsson

2023

Mach. Learn.: Sci. Technol.

View full text Add to dashboard Cite

Analyzing large volumes of high-dimensional data requires dimensionality reduction: finding meaningful low-dimensional structures hidden in their high-dimensional observations. Such practice is needed in atomistic simulations of complex systems where even thousands of degrees of freedom are sampled. An abundance of such data makes gaining insight into a specific physical problem strenuous. Our primary aim in this review is to focus on unsupervised machine learning methods that can be used on simulation data to find a low-dimensional manifold providing a collective and informative characterization of the studied process. Such manifolds can be used for sampling long-timescale processes and free-energy estimation. We describe methods that can work on datasets from standard and enhanced sampling atomistic simulations. Unlike recent reviews on manifold learning for atomistic simulations, we consider only methods that construct low-dimensional manifolds based on Markov transition probabilities between high-dimensional samples. We discuss these techniques from a conceptual point of view, including their underlying theoretical frameworks and possible limitations.

show abstract

Section: Diffusion and Commute Distancesmentioning

confidence: 99%

Manifold learning in atomistic simulations: a conceptual review

Rydzewski

Chen

Valsson

2023

Mach. Learn.: Sci. Technol.

View full text Add to dashboard Cite

show abstract

“…where we can see the difference between the diffusion and commute distances are in the coefficients λ n and t n /2, respectively. Various methods exploit the relation of the effective timescale with eigenvalues [9,[117][118][119][120][121][122][123].…”

Section: Diffusion and Commute Distancesmentioning

confidence: 99%

“…defines a pairwise reweighting factor required to unbiased the transition probabilities estimated from biased data [51,124]. Other reweighting methods for DMAP are also available [120,123,125].…”

Section: Dmap Can Learn Low-dimensional Cvs From Unbiased Atomistic S...mentioning

confidence: 99%

Manifold Learning in Atomistic Simulations: A Conceptual Review

Rydzewski¹,

Chen²,

Valsson³

2023

Preprint

View full text Add to dashboard Cite

Analyzing large volumes of high-dimensional data requires dimensionality reduction: finding meaningful low-dimensional structures hidden in their high-dimensional observations. Such practice is needed in atomistic simulations of dynamical systems where even thousands of degrees of freedom are sampled. An abundance of such data makes it strenuous to gain insight into a specific physical problem. Our primary aim in this review is to focus on unsupervised machine learning methods that can be used on simulation data to find a low-dimensional manifold providing a collective and informative characterization of the studied process. Such manifolds can be used for sampling longtimescale processes and free-energy estimation. We describe methods that can work on data sets from standard and enhanced sampling atomistic simulations. Compared to recent reviews on manifold learning for atomistic simulations, we consider only methods that construct low-dimensional manifolds based on Markov transition probabilities between high-dimensional samples. We discuss these techniques from a conceptual point of view, including their underlying theoretical frameworks and possible limitations.

show abstract

“…This feature not only makes the algorithm adaptive, but offers future scope of improvement for applications requiring advanced decision-making, either based on inferencing [12][13][14]54,55 or neural network based machine learning algorithms. [56][57][58][59][60] Finally, the application converges to yield a refined ensemble (7), which exit the R-MDFF workflow and downloads results to the end-user's working directory. The R-MDFF API is implemented as a Python module, loaded into the workflow application's code.…”

Section: Introductionmentioning

confidence: 99%

Adaptive Ensemble Refinement of Protein Structures in High Resolution Electron Microscopy Density Maps with Radical Augmented Molecular Dynamics Flexible Fitting

Sarkar

Lee

Vant

et al. 2021

Preprint

View full text Add to dashboard Cite

Recent advances in cryo-electron microscopy (cryo-EM) has enabled modeling macromolecular complexes that are essential components of life. The density maps obtained from cryo-EM experiments is often integrated with ab-initio, knowledge-driven or first principles-based computational methods to build, fit and refine protein structures inside the electron density maps. Going beyond a single stationary-structure determination scheme, it is becoming more common to interpret the experimental data with a set of multiple physical models all of which contributes to the average observation seen by the experiment. Hence, there is a need to decide on the quality of an ensemble of protein structures on-the-fly, while refining them against the density maps. In this work, we demonstrate such adaptive decision making capabilities during flexible fitting of biomolecules. Our solution uses RADICAL tools (RCT) and we test this new implementation in exascale high performance computing environment for two proteins, Adenylate Kinase (ADK) and Carbon Monoxide Dehydrogenase (CODH). Our results indicate that using multiple replicas in flexible fitting with adaptive decision making improves the overall quality of fit and model by 40 % improvement when compared against the traditional flexible fitting approach. These advances are agnostic to system-size and computing environments.

show abstract

Computing committors in collective variables via Mahalanobis diffusion maps

Cited by 8 publications

References 62 publications

Manifold learning in atomistic simulations: a conceptual review

Manifold learning in atomistic simulations: a conceptual review

Manifold Learning in Atomistic Simulations: A Conceptual Review

Adaptive Ensemble Refinement of Protein Structures in High Resolution Electron Microscopy Density Maps with Radical Augmented Molecular Dynamics Flexible Fitting

Contact Info

Product

Resources

About