Top-down machine learning of coarse-grained protein force-fields
This addresses the need for efficient protein dynamics modeling for researchers in biophysics and computational biology, offering a novel approach but with incremental improvements in training data requirements.
The paper tackles the problem of developing accurate coarse-grained protein force-fields by introducing a method that trains neural network potentials using only native protein conformations, enabling efficient simulation of folding events and prediction of native-like structures with extrapolation capabilities.
Developing accurate and efficient coarse-grained representations of proteins is crucial for understanding their folding, function, and interactions over extended timescales. Our methodology involves simulating proteins with molecular dynamics and utilizing the resulting trajectories to train a neural network potential through differentiable trajectory reweighting. Remarkably, this method requires only the native conformation of proteins, eliminating the need for labeled data derived from extensive simulations or memory-intensive end-to-end differentiable simulations. Once trained, the model can be employed to run parallel molecular dynamics simulations and sample folding events for proteins both within and beyond the training distribution, showcasing its extrapolation capabilities. By applying Markov State Models, native-like conformations of the simulated proteins can be predicted from the coarse-grained simulations. Owing to its theoretical transferability and ability to use solely experimental static structures as training data, we anticipate that this approach will prove advantageous for developing new protein force fields and further advancing the study of protein dynamics, folding, and interactions.