Rodriguez Perez, Raquel and Bajorath, Jürgen (2021) Feature importance correlation from machine learning indicates functional relationships between proteins and similar compound binding characteristics. Scientific reports, 11.

Abstract

Machine learning is widely applied in drug discovery research to predict molecular properties and aid in the identification of new active compounds. Herein, we introduce a new approach that uses model-internal information from compound activity predictions to uncover relationships between target proteins. On the basis of a large-scale analysis comparing machine learning models for more than 200 proteins, feature importance correlation analysis is shown to detect similar compound binding characteristics. Furthermore, rather unexpectedly, the analysis also reveals functional relationships between proteins that are independent of binding characteristics. The underlying concept does not depend on specific representations, algorithms, or metrics and is generally applicable as long as predictive models can be derived. On the basis of our findings, feature importance correlation represents a new facet of machine learning in drug discovery with potential for practical applications.

Item Type:	Article
Keywords:	Machine learning, active compounds, target proteins, feature importance, binding characteristics, functional relationships.
Date Deposited:	23 Jul 2021 00:45
Last Modified:	23 Jul 2021 00:45
URI:	https://oak.novartis.com/id/eprint/44761