Rodriguez Perez, Raquel, Trunzer, Markus, Schneider, Nadine, Faller, Bernard and Gerebtzoff, Gregori (2022) Multi-species machine learning predictions of in vitro intrinsic clearance with uncertainty quantification analyses. Molecular Pharmaceutics. ISSN 1543-83841543-8392

Abstract

In pharmaceutical research, compounds are optimized for metabolic stability to avoid a too fast elimination of the drug. Intrinsic clearance (CLint) measured in liver microsomes or hepatocytes is an important parameter during lead optimization. In this work, machine learning models were developed to relate compound structure to microsomal metabolic stability and predict CLint for new compounds. A multitask (MT) learning architecture was introduced to model the CLint of six species simultaneously, giving as a result a multi-species machine learning model. MT graph neural network (MT-GNN) regression was identified as the top-performing method and an ensemble of ten MT-GNN models was evaluated prospectively. Geometric mean fold errors were consistently smaller than 2-fold. Moreover, high precision values were obtained in the prediction of ‘high’ (>300µL/min/mg) and ‘low’ (<100µL/min/mg) CLint compounds. Precision values ranged from 80% to 94% for low CLint predictions and from 75% to 97% for high CLint predictions, depending on the species. Uncertainty on experimental values and model predictions was systematically quantified. Experimental variability (aleatoric uncertainty) of all historical Novartis in vitro clearance experiments was analyzed. Interestingly, MT-GNN models’ performance approached assay’s experimental variability. Moreover, uncertainty estimation in predictions (epistemic uncertainty) enabled identifying predictions associated to lower and higher error. Taken together, our manuscript combines a multi-species deep learning model and large-scale uncertainty analyses to improve CLint predictions and facilitate early informed decisions for compound prioritization.

Item Type:	Article
Date Deposited:	20 Dec 2022 00:45
Last Modified:	20 Dec 2022 00:45
URI:	https://oak.novartis.com/id/eprint/48115