Improving Molecular Design with Direct Inverse Analysis of QSAR/QSPR Model

Abstract

Recent advances in machine learning have significantly impacted molecular design, notably the molecular generation method combining the chemical variational autoencoder (VAE) with Gaussian mixture regression (GMR). In this method, a mathematical model is constructed with X as the latent variable of the molecule and Y as the target properties and activities. Through direct inverse analysis of this model, it is possible to generate molecules with the desired target properties. However, this approach outputs many strings that do not follow the simplified molecular input line entry system grammar and generates unrealistic chemical structures in which the properties and activity do not satisfy the target values. In this study, we focus on hierarchical VAE using molecular graphs to address these issues. We confirm that the combination of hierarchical VAE and GMR does not generate invalid outputs and returns molecules that simultaneously satisfy multiple target values. Moreover, we use this method to identify several molecules that are predicted to exhibit activity against drug targets.

Keywords: autoencoder; cheminformatics; drug design; machine learning; virtual screening.

Figure 1

Basic concept of direct inverse analysis through the integration of VAE and GMR.

Figure 2

Diagram of VAE training with joint property prediction.

Figure 3

(a)Actual logP vs predicted value plots of test data. (b)Actual QED vs predicted value plots of test data. (c)Actual SAS vs predicted value plots of test data.

Figure 4

A molecule generated by the proposed method that achieve the target values.

Figure 5

(a) Crystal structure of target protein (DRD2). (b) Docking pose of interaction between DRD2 and risperidone. (c) Docking pose of interaction between DRD2 and the novel molecule with the highest binding affinity.

Figure 6

Binding affinity energies of generated molecules; black points indicate initial samples, blue points indicate generated samples, red points indicate samples meeting the target, grey area indicates target range (target value: −12.2).