Is Preservation of Symmetry Necessary for Coarse-Graining?

Chakraborty, Maghesree; Xu, Jinyu; White, Andrew D.

doi:10.26434/chemrxiv.10728128.v1

Cited by 6 publications

(8 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Chakraborty et al recently showed that CG-mapping symmetry has a negligible impact on structural accuracy. 38 Asymmetry indeed appears to be irrelevant for SP models. However, the use of asymmetric CG mappings will hurt the transferability in the EE scheme.…”

Section: State-point Specific Modelsmentioning

confidence: 99%

“…Low performance stems from certain functional groups that promote vastly different conformational ensembles depending on the environment and molecular topology. An extended-ensemble average over the structural correlations of these functional groups does not capture the specificity of these diverging conformational states, and instead suggests the need for an improved mapping [38][39][40] or an increased force-field complexity. 33,[41][42][43] We validate the transferability of the derived potentials by running CG simulations on compounds that were not used in the extendedensemble training set and find that the accuracy of the CG RDFs is on par with that of the representative compounds.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Broad chemical transferability in structure-based coarse-graining

Kanekal¹,

Rudzinski²,

Bereau³

2022

Preprint

View full text Add to dashboard Cite

Compared to top-down coarse-grained (CG) models, bottom-up approaches are capable of offering higher structural fidelity. This fidelity results from the tight link to a higher-resolution reference, making the CG model chemically specific. Unfortunately, chemical specificity can be at odds with compound-screening strategies, which call for transferable parametrizations. Here we present an approach to reconcile bottom-up, structure-preserving CG models with chemical transferability. We consider the bottom-up CG parametrization of 3,441 C 7 O 2 small-molecule isomers. Our approach combines atomic representations, unsupervised learning, and a large-scale extended-ensemble force-matching parametrization. We first identify a subset of 19 representative molecules, which maximally encode the local environment of all gas-phase conformers. Reference interactions between the 19 representative molecules were obtained from both homogeneous bulk liquids and various binary mixtures. An extended-ensemble parametrization over all 703 state points leads to a CG model that is both structure-based and chemically transferable. Remarkably, the resulting force field is on average more structurally accurate than single-state-point equivalents. Averaging over the extended ensemble acts as a mean-force regularizer, smoothing out both force and structural correlations that are overly specific to a single state point. Our approach aims at transferability through a set of CG bead types that can be used to easily construct new molecules, while retaining the benefits of a structure-based parametrization.

show abstract

Section: State-point Specific Modelsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Broad chemical transferability in structure-based coarse-graining

Kanekal¹,

Rudzinski²,

Bereau³

2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Systematic parametrization schemes exist, such as iterative Boltzmann inversion or force matching, accompanied by convenient software platforms [95,96]. Aside from the CG potentials, bottom-up strategies can strongly benefit from a more systematic optimization of the mapping itself [97,98]. Combinations of structure-based CG and ML have recently sparked interest and are quickly enabling new avenues, see below Section 2.5.3.…”

Section: Force-field Parametrizationmentioning

confidence: 99%

Computational compound screening of biomolecules and soft materials by molecular simulations

Bereau¹

2020

Preprint

View full text Add to dashboard Cite

Decades of hardware, methodological, and algorithmic development have propelled molecular dynamics (MD) simulations to the forefront of materials-modeling techniques, bridging the gap between electronic-structure theory and continuum methods. The physics-based approach makes MD appropriate to study emergent phenomena, but simultaneously incurs significant computational investment. This topical review explores the use of MD outside the scope of individual systems, but rather considering many compounds. Such an in silico screening approach makes MD amenable to establishing coveted structure-property relationships. We specifically focus on biomolecules and soft materials, characterized by the significant role of entropic contributions and heterogeneous systems and scales. An account of the state of the art for the implementation of an MD-based screening paradigm is described, including automated force-field parametrization, system preparation, and efficient sampling across both conformation and composition. Emphasis is placed on machinelearning methods to enable MD-based screening. The resulting framework enables the generation of compound-property databases and the use of advanced statistical modeling to gather insight. The review further summarizes a number of relevant applications.

show abstract

“…The experiment will run 5 iterations. At i-th iteration (i ∈ [1,5]), the i-th split of the dataset is regarded as testing set (groundtruth partition B is not used) and rest 4 splits of the dataset is regarded as the training set (ground-truth partition B is used for training). Therefore, after training, DSGPM is evaluated on unseen molecules in the testing set.…”

Section: Evaluation Metricsmentioning

confidence: 99%

Graph Neural Network Based Coarse-Grained Mapping Prediction

Li,

Wellawatte,

Chakraborty

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

The selection of coarse-grained (CG) mapping operators is a critical step for CG molecular dynamics (MD) simulation. It is still an open question about what is optimal for this choice and there is a need for theory. The current state-of-the art method is mapping operators manually selected by experts. In this work, we demonstrate an automated approach by viewing this problem as supervised learning where we seek to reproduce the mapping operators produced by experts. We present a graph neural network based CG mapping predictor called DEEP SUPERVISED GRAPH PARTITIONING MODEL(DSGPM) that treats mapping operators as a graph segmentation problem. DSGPM is trained on a novel dataset, Humanannotated Mappings (HAM), consisting of 1,206 molecules with expert annotated mapping operators. HAM can be used to facilitate further research in this area. Our model uses a novel metric learning objective to produce high-quality atomic features that are used in spectral clustering. The results show that the DSGPM outperforms state-of-the-art methods in the field of graph segmentation. Finally, we find that predicted CG mapping operators indeed result in good CG MD models when used in simulation.

show abstract

Is Preservation of Symmetry Necessary for Coarse-Graining?

Abstract: This work investigates how preservation of molecular symmetry affects accuracy of coarse-grained (CG) molecular dynamics simulation. We studied 26 mapping operators for 7 molecules to find that it has little effect on accuracy of CG simulations.

Cited by 6 publications

References 16 publications

Broad chemical transferability in structure-based coarse-graining

Broad chemical transferability in structure-based coarse-graining

Computational compound screening of biomolecules and soft materials by molecular simulations

Graph Neural Network Based Coarse-Grained Mapping Prediction

Contact Info

Product

Resources

About