Multi-body effects in a coarse-grained protein force field

Abstract

The use of coarse-grained (CG) models is a popular approach to study complex biomolecular systems. By reducing the number of degrees of freedom, a CG model can explore long time-and length-scales inaccessible to computational models at higher resolution. If a CG model is designed by formally integrating out some of the system's degrees of freedom, one expects multi-body interactions to emerge in the effective CG model's energy function. In practice, it has been shown that the inclusion of multi-body terms indeed improves the accuracy of a CG model. However, no general approach has been proposed to systematically construct a CG effective energy that includes arbitrary orders of multi-body terms. In this work, we propose a neural network based approach to address this point and construct a CG model as a multibody expansion. By applying this approach to a small protein, we evaluate the relative importance of the different multi-body terms in the definition of an accurate model. We observe a slow convergence in the multi-body expansion, where up to five-body interactions are needed to reproduce the free energy of an atomistic model.

Gaussian Process Regression for Materials and Molecules

“…Other examples are the direct prediction of the lattice energy of molecular crystals, using as training and as inputs only the geometries optimized with an empirical force field (which is a simpler learning task than training a fully general potential for the same class of systems), and the estimation of free-energy surfaces, which involve finite-temperature sampling with a (traditional or machine-learning) force field. Although the present review focuses on fully atomistic models, the construction of ML-based coarse-grained force fields is a burgeoning research field where initial progress has been made with GPR-based and other ML methods. − …”

Multiscale molecular kinetics by coupling Markov state models and reaction-diffusion dynamics

“…The scheme might also become less effective in the presence of long-ranged interactions, though it might be possible to incorporate them into the dynamics of the non-interacting-regime using coarse-grained potentials. 75,76 Finally, in its current form, the MSM/RD multiparticle implementation only takes into account pair interactions, and thus the scheme is not yet adequate for crowded multi-molecular environments.…”

Collective variable-based enhanced sampling and machine learning

“…Moreover, a few-body interaction potential is needed for the CG potential energy function. In a recent study, machine learning-based CG potential energy functions with few-body (two-body to five-body) interactions have been proposed [ 170 ]. Samples from CG potential and samples from all-atom simulations have been projected onto TICA CVs to generate a CG FES and an all-atom FES.…”