Imputation of sensory properties using deep learning

• Published in: Journal of computer-aided molecular design Vol. 35; no. 11; pp. 1125 - 1140
• Main Authors: Mahmoud, Samar, Irwin, Benedict, Chekmarev, Dmitriy, Vyas, Shyam, Kattas, Jeff, Whitehead, Thomas, Mansley, Tamsin, Bikker, Jack, Conduit, Gareth, Segall, Matthew
• Format: Journal Article
• Language: English
• Published: Cham Springer International Publishing 01.11.2021; Springer Nature B.V
• Subjects: Algorithms; Animal Anatomy; Artificial neural networks; Chemistry; Chemistry and Materials Science; Computer Applications in Chemistry; Deep Learning; Histology; Humans; Morphology; Physical Chemistry; Quantitative Structure-Activity Relationship; Sensory Receptor Cells - physiology; Uncertainty; Deep learning; Sensory properties; Imputation; Quantitative structure–activity relationship; In silico model
• ISSN: 0920-654X; 1573-4951
• DOI: 10.1007/s10822-021-00424-3

Predicting the sensory properties of compounds is challenging due to the subjective nature of the experimental measurements. This testing relies on a panel of human participants and is therefore also expensive and time-consuming. We describe the application of a state-of-the-art deep learning method, Alchemite™, to the imputation of sparse physicochemical and sensory data and compare the results with conventional quantitative structure–activity relationship methods and a multi-target graph convolutional neural network. The imputation model achieved a substantially higher accuracy of prediction, with improvements in R 2 between 0.26 and 0.45 over the next best method for each sensory property. We also demonstrate that robust uncertainty estimates generated by the imputation model enable the most accurate predictions to be identified and that imputation also more accurately predicts activity cliffs, where small changes in compound structure result in large changes in sensory properties. In combination, these results demonstrate that the use of imputation, based on data from less expensive, early experiments, enables better selection of compounds for more costly studies, saving experimental time and resources.