Implications of Additivity and Nonadditivity for Machine Learning and Deep Learning Models in Drug Design

Abstract

Matched molecular pairs (MMPs) is nowadays a commonly applied concept in drug design. It is used in many computational tools for structure activity relationship analysis, biological activity prediction or optimization of physicochemical properties. However, up to date it has not been shown in a rigorous way that MMPs, i.e. changing only one substituent between two molecules, can be predicted with high accuracy and precision in contrast to any other chemical compound pair. It is expected that any model should be able to predict such a defined change with high accuracy and reasonable precision. In this study, we examine the predictability of four classical properties relevant for drug design ranging from simple physicochemical parameters (logD and solubility) to more complex cell based ones (permeability and clearance), using different data sets and machine learning algorithms. Our study confirms that additive data is the easiest to predict which highlights the importance of recognition of nonadditivity events and the challenging complexity of predicting properties in case of scaffold hopping. Despite of deep learning being well suited to model non-linear events, these methods do not seem to be an exception of this observation. Though, they are in general performing better than classical machine learning methods, this leaves the field with a still standing challenge.

Exposing the limitations of molecular machine learning with activity cliffs.

“…Similar results were obtained when comparing graph networks for (a) feature attribution with activity cliffs 84 , and (b) bioactivity prediction 30 . A recent analysis on physicochemical-property cliffs highlights an opposite trend, with deep learning methods performing better than simpler machine learning approaches 85 -potentially due to the higher number of training samples (approx. 20,000 molecules).…”