Dataset Information

ClassicalGSG: Prediction of log P using classical molecular force fields and geometric scattering for graphs.

ABSTRACT: This work examines methods for predicting the partition coefficient (log P) for a dataset of small molecules. Here, we use atomic attributes such as radius and partial charge, which are typically used as force field parameters in classical molecular dynamics simulations. These atomic attributes are transformed into index-invariant molecular features using a recently developed method called geometric scattering for graphs (GSG). We call this approach "ClassicalGSG" and examine its performance under a broad range of conditions and hyperparameters. We train ClassicalGSG log P predictors with neural networks using 10,722 molecules from the OpenChem dataset and apply them to predict the log P values from four independent test sets. The ClassicalGSG method's performance is compared to a baseline model that employs graph convolutional networks. Our results show that the best prediction accuracies are obtained using atomic attributes generated with the CHARMM generalized force field and 2D molecular structures.

SUBMITTER: Donyapour N

PROVIDER: S-EPMC8062296 | biostudies-literature | 2021 May

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

ClassicalGSG: Prediction of log P using classical molecular force fields and geometric scattering for graphs.

Donyapour Nazanin N Hirn Matthew M Dickson Alex A

Journal of computational chemistry 20210330 14

This work examines methods for predicting the partition coefficient (log P) for a dataset of small molecules. Here, we use atomic attributes such as radius and partial charge, which are typically used as force field parameters in classical molecular dynamics simulations. These atomic attributes are transformed into index-invariant molecular features using a recently developed method called geometric scattering for graphs (GSG). We call this approach "ClassicalGSG" and examine its performance und ...[more]

PMID: 33786857

Similar Datasets

Project description:For the successful development and application of lubricants, a full understanding of their complex nanoscale behavior under a wide range of external conditions is required, but this is difficult to obtain experimentally. Nonequilibrium molecular dynamics (NEMD) simulations can be used to yield unique insights into the atomic-scale structure and friction of lubricants and additives; however, the accuracy of the results depend on the chosen force-field. In this study, we demonstrate that the use of an accurate, all-atom force-field is critical in order to; (i) accurately predict important properties of long-chain, linear molecules; and (ii) reproduce experimental friction behavior of multi-component tribological systems. In particular, we focus on n-hexadecane, an important model lubricant with a wide range of industrial applications. Moreover, simulating conditions common in tribological systems, i.e., high temperatures and pressures (HTHP), allows the limits of the selected force-fields to be tested. In the first section, a large number of united-atom and all-atom force-fields are benchmarked in terms of their density and viscosity prediction accuracy of n-hexadecane using equilibrium molecular dynamics (EMD) simulations at ambient and HTHP conditions. Whilst united-atom force-fields accurately reproduce experimental density, the viscosity is significantly under-predicted compared to all-atom force-fields and experiments. Moreover, some all-atom force-fields yield elevated melting points, leading to significant overestimation of both the density and viscosity. In the second section, the most accurate united-atom and all-atom force-field are compared in confined NEMD simulations which probe the structure and friction of stearic acid adsorbed on iron oxide and separated by a thin layer of n-hexadecane. The united-atom force-field provides an accurate representation of the structure of the confined stearic acid film; however, friction coefficients are consistently under-predicted and the friction-coverage and friction-velocity behavior deviates from that observed using all-atom force-fields and experimentally. This has important implications regarding force-field selection for NEMD simulations of systems containing long-chain, linear molecules; specifically, it is recommended that accurate all-atom potentials, such as L-OPLS-AA, are employed.

Project description:Self-assembled cyclic peptide nanotubes with alternating D- and L-amino acid residues in the sequence of each subunit have attracted a great deal of attention due to their potential for new nanotechnology and biomedical applications, mainly in the field of antimicrobial peptides. Molecular dynamics simulations can be used to characterize these systems with atomic resolution at different time scales, providing information that is difficult to obtain via wet lab experiments. However, the performance of classical force fields typically employed in the simulation of biomolecules has not yet been extensively tested with this kind of highly constrained peptide. Four different classical force fields (AMBER, CHARMM, OPLS, and GROMOS), using a nanotube formed by eight D,L-α-cyclic peptides inserted into a lipid bilayer as a model system, were employed here to fill this gap. Significant differences in the pseudo-cylindrical cavities formed by the nanotubes were observed, the most important being the diameter of the nanopores, the number and location of confined water molecules, and the density distribution of the solvent molecules. Furthermore, several modifications were performed on GROMOS54a7, aiming to explore acceleration strategies of the MD simulations. The hydrogen mass repartitioning (HMR) and hydrogen isotope exchange (HIE) methods were tested to slow down the fastest degrees of freedom. These approaches allowed a significant increase in the time step employed in the equation of the motion integration algorithm, from 2 fs up to 5-7 fs, with no serious changes in the structural and dynamical properties of the nanopores. Subtle differences with respect to the simulations with the unmodified force fields were observed in the concerted movements of the cyclic peptides, as well as in the lifetime of several H-bonds. All together, these results are expected to contribute to better understanding of the behavior of self-assembled cyclic peptide nanotubes, as well as to support the methods tested to speed up general MD simulations; additionally, they do provide a number of quantitative descriptors that are expected to be used as a reference to design new experiments intended to validate and complement computational studies of antimicrobial cyclic peptides.

Project description:When good structural templates can be identified, template-based modeling is the most reliable way to predict the tertiary structure of proteins. In this study, we combine template-based modeling with a realistic coarse-grained force field, AWSEM, that has been optimized using the principles of energy landscape theory. The Associative memory, Water mediated, Structure and Energy Model (AWSEM) is a coarse-grained force field having both transferable tertiary interactions and knowledge-based local-in-sequence interaction terms. We incorporate template information into AWSEM by introducing soft collective biases to the template structures, resulting in a model that we call AWSEM-Template. Structure prediction tests on eight targets, four of which are in the low sequence identity "twilight zone" of homology modeling, show that AWSEM-Template can achieve high-resolution structure prediction. Our results also confirm that using a combination of AWSEM and a template-guided potential leads to more accurate prediction of protein structures than simply using a template-guided potential alone. Free energy profile analyses demonstrate that the soft collective biases to the template effectively increase funneling toward native-like structures while still allowing significant flexibility so as to allow for correction of discrepancies between the target structure and the template. A further stage of refinement using all-atom molecular dynamics augmented with soft collective biases to the structures predicted by AWSEM-Template leads to a further improvement of both backbone and side-chain accuracy by maintaining sufficient flexibility but at the same time discouraging unproductive unfolding events often seen in unrestrained all-atom refinement simulations. The all-atom refinement simulations also reduce patches of frustration of the initial predictions. Some of the backbones found among the structures produced during the initial coarse-grained prediction step already have CE-RMSD values of less than 3 Å with 90% or more of the residues aligned to the experimentally solved structure for all targets. All-atom structures generated during the following all-atom refinement simulations, which started from coarse-grained structures that were chosen without reference to any knowledge about the native structure, have CE-RMSD values of less than 2.5 Å with 90% or more of the residues aligned for 6 out of 8 targets. Clustering low energy structures generated during the initial coarse-grained annealing picks out reliably structures that are within 1 Å of the best sampled structures in 5 out of 8 cases. After the all-atom refinement, structures that are within 1 Å of the best sampled structures can be selected using a simple algorithm based on energetic features alone in 7 out of 8 cases.

Dataset Information

ClassicalGSG: Prediction of log P using classical molecular force fields and geometric scattering for graphs.

Publications

ClassicalGSG: Prediction of log P using classical molecular force fields and geometric scattering for graphs.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets