Title: QRNN: Transferable Neural Network for Potential Energy Surfaces of Closed-Shell Organic Molecules including Ions

Date: 29th June 2023, Thursday

Time: 2:00 PM (Central European Time)

Host: Prof. Bartosz Trzaskowski

The seminar will be in the CeNT aula hall (00.142) on the main floor.

Abstract:

Transferable high dimensional neural network potentials (HDNNPs) have shown great promise as an avenue to increase the accuracy and domain of applicability of existing atomistic force fields for organic systems relevant to life science. We have previously reported such a potential (Schrödinger-ANI) that has broad coverage of druglike molecules. We extend that work here to cover ionic and zwitterionic druglike molecules expected to be relevant to drug discovery research. We report a novel HDNNP architecture, which we call QRNN, that predicts atomic charges and uses these charges as descriptors in an energy model that delivers relative conformational energies within chemical accuracy when measured against the reference theory it is trained to. Further, we find that delta learning based on a semiempirical level of theory (GFN2-xTB) approximately further halves the errors. We test the models on torsion energy profiles, relative conformational energies, geometric parameters, and relative tautomer errors.