ToxicBlend: Virtual Screening of Toxic Compounds with Ensemble Predictors

TL;DR

ToxicBlend is an ensemble machine learning model that combines multiple architectures to improve the accuracy of virtual screening for toxic compounds, aiding faster and more reliable toxicity assessment in drug discovery.

Contribution

The paper introduces a novel ensemble approach combining gradient boosting and neural networks trained on diverse molecular representations for toxicity prediction.

Findings

Outperforms existing models on ToxCast and Tox21 datasets

Leverages multiple molecular representations for improved accuracy

Provides free online tool for toxicity prediction

Abstract

Timely assessment of compound toxicity is one of the biggest challenges facing the pharmaceutical industry today. A significant proportion of compounds identified as potential leads are ultimately discarded due to the toxicity they induce. In this paper, we propose a novel machine learning approach for the prediction of molecular activity on ToxCast targets. We combine extreme gradient boosting with fully-connected and graph-convolutional neural network architectures trained on QSAR physical molecular property descriptors, PubChem molecular fingerprints, and SMILES sequences. Our ensemble predictor leverages the strengths of each individual technique, significantly outperforming existing state-of-the art models on the ToxCast and Tox21 toxicity-prediction datasets. We provide free access to molecule toxicity prediction using our model at http://www.owkin.com/toxicblend.