Author:
Moukheiber Lama,Mangione William,Maleki Saeed,Falls Zackary,Gao Mingchen,Samudrala Ram
Abstract
AbstractHumans are exposed to numerous compounds daily, some of which have adverse effects on health. Computational approaches for modeling toxicological data in conjunction with machine learning algorithms have gained popularity over the last few years. Machine learning methods have been used to predict toxicity-related biological activities using chemical structure descriptors. However, toxicity-related proteomic features have not been fully investigated. In this study, we construct a computational model using machine learning for selecting the most important proteins representing features in predicting the toxicity of the compounds in the Tox21 dataset using the multiscale Computational Analysis of Novel Drug Opportunities (CANDO) platform for therapeutic discovery. Tox21 is a highly imbalanced dataset consisting of twelve in vitro assays, seven from the nuclear receptor (NR) signaling pathway and five from the stress response (SR) pathway, for more than 10,000 compounds. For our computational model, we employed a random forest (RF) with the combination of Synthetic Minority Oversampling Technique (SMOTE) and Edited Nearest Neighbor (ENN) method, aka SMOTE+ENN, which is resampling method to balance the activity class distribution. Within the NR and SR pathways, the activity of the aryl hydrocarbon receptor (NR-AhR), toxicity mediating transcription factor, and mitochondrial membrane potential (SR-MMP) were two of the top-performing twelve toxicity endpoints with AUROCs of 0.90 and 0.92, respectively. The top extracted features for evaluating compound toxicity were passed into enrichment analysis to highlight the implicated biological pathways and proteins. We validated our enrichment results for the activity of the AhR using a thorough literature search. Our case study showed that the selected enriched pathways and proteins from our computational pipeline are not only correlated with NR-AhR toxicity but also form a cascading upstream/downstream arrangement. Our work elucidates significant relationships between protein and compound interactions computed using CANDO and the associated biological pathways to which the proteins belong, with twelve toxicity endpoints. This novel study uses machine learning not only to predict and understand toxicity but also elucidates therapeutic mechanisms at a proteomic level for a variety of toxicity endpoints.
Publisher
Cold Spring Harbor Laboratory