Generating Focused Molecule Libraries for Drug Discovery with Recurrent Neural Networks

Abstract

In <i>de novo</i> drug design, computational strategies are used to generate novel molecules with good affinity to the desired biological target. In this work, we show that recurrent neural networks can be trained as generative models for molecular structures, similar to statistical language models in natural language processing. We demonstrate that the properties of the generated molecules correlate very well with the properties of the molecules used to train the model. In order to enrich libraries with molecules active toward a given biological target, we propose to fine-tune the model with small sets of molecules, which are known to be active against that target. Against <i>Staphylococcus aureus</i>, the model reproduced 14% of 6051 hold-out test molecules that medicinal chemists designed, whereas against <i>Plasmodium falciparum</i> (Malaria), it reproduced 28% of 1240 test molecules. When coupled with a scoring function, our model can perform the complete <i>de novo</i> drug design cycle to generate large sets of novel molecules for drug discovery.