Machine learned calibrations to high-throughput molecular excited state calculations

Abstract

Understanding the excited state properties of molecules provides insights into how they interact with light. These interactions can be exploited to design compounds for photochemical applications, including enhanced spectral conversion of light to increase the efficiency of photovoltaic cells. While chemical discovery is time- and resource-intensive experimentally, computational chemistry can be used to screen large-scale databases for molecules of interest in a procedure known as high-throughput virtual screening. The first step usually involves a high-speed but low-accuracy method to screen large numbers of molecules (potentially millions) so only the best candidates are evaluated with expensive methods. However, use of a coarse first-pass screening method can potentially result in high false positive or false negative rates. Therefore, this study uses machine learning to calibrate a high-throughput technique (xTB-sTDA) against a higher accuracy one (TD-DFT). Testing the calibration model shows a ~5-fold decrease in error in-domain and a ~3-fold decrease out-of-domain. The resulting mean absolute error of ~0.14 eV is in line with previous work in machine learning calibrations and out-performs previous work in linear calibration of xTB-sTDA. We then apply the calibration model to screen a 250k molecule database and map inaccuracies of xTB-sTDA in chemical space. We also show generalizability of the workflow by calibrating against a higher-level technique (CC2), yielding a similarly low error. Overall, this work demonstrates machine learning can be used to develop a both cheap and accurate method for large-scale excited state screening, enabling accelerated molecular discovery across a variety of disciplines.

Ground truth explanation dataset for chemical property prediction on molecular graphs

“…A wide range of machine learning (ML) approaches allows for explaining the chemistry of molecules, attributing which parts of the molecules are responsible for the chemical property of interest [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19] , and lessening the black box challenge of machine learning 20,21 . Typical explainable ML approaches that provide atomwise attribution include dummy atoms 22 , classification of atoms by chemical intuition 23 , regression models 24 , graph neural network (GNN) attributions [25][26][27][28] with gradients 29 , perturbations 30 , decompositions 31 , and surrogates 32 .…”