Nutmeg and SPICE: Models and Data for Biomolecular Machine Learning
This work addresses the need for accurate and efficient machine learning potentials in computational chemistry, particularly for simulating small molecules, though it appears incremental as it builds on existing datasets and architectures.
The authors tackled the problem of training machine learning potentials for biomolecular simulations by expanding the SPICE dataset with broader chemical sampling and non-covalent interaction data, and introduced Nutmeg models based on TensorNet that incorporate precomputed partial charges to improve performance on charged and polar molecules, achieving excellent reproduction of energy differences and stable molecular dynamics trajectories.
We describe version 2 of the SPICE dataset, a collection of quantum chemistry calculations for training machine learning potentials. It expands on the original dataset by adding much more sampling of chemical space and more data on non-covalent interactions. We train a set of potential energy functions called Nutmeg on it. They are based on the TensorNet architecture. They use a novel mechanism to improve performance on charged and polar molecules, injecting precomputed partial charges into the model to provide a reference for the large scale charge distribution. Evaluation of the new models shows they do an excellent job of reproducing energy differences between conformations, even on highly charged molecules or ones that are significantly larger than the molecules in the training set. They also produce stable molecular dynamics trajectories, and are fast enough to be useful for routine simulation of small molecules.