druGAN: An Advanced Generative Adversarial Autoencoder Model for de Novo Generation of New Molecules with Desired Molecular Properties in Silico

TLDR

Deep generative adversarial networks and variational autoencoders are emerging deep learning models used in drug discovery and biomarker development. The study aims to develop an advanced AAE model for molecular feature extraction and compare its performance to VAE. The authors implement an advanced AAE that offers adjustable fingerprint generation, handles large datasets, and enables efficient unsupervised pretraining for regression. The proposed AAE significantly enhances capacity and efficiency for developing new anticancer molecules using deep generative models.

Abstract

Deep generative adversarial networks (GANs) are the emerging technology in drug discovery and biomarker development. In our recent work, we demonstrated a proof-of-concept of implementing deep generative adversarial autoencoder (AAE) to identify new molecular fingerprints with predefined anticancer properties. Another popular generative model is the variational autoencoder (VAE), which is based on deep neural architectures. In this work, we developed an advanced AAE model for molecular feature extraction problems, and demonstrated its advantages compared to VAE in terms of (a) adjustability in generating molecular fingerprints; (b) capacity of processing very large molecular data sets; and (c) efficiency in unsupervised pretraining for regression model. Our results suggest that the proposed AAE model significantly enhances the capacity and efficiency of development of the new molecules with specific anticancer properties using the deep generative models.