PubChem (https://pubchemncbi.nlm.nih.gov) is a large-scale chemical information resource at the U.S. National Institutes of Health. Among various kinds of data contained in PubChem are the descriptions and test results of biological assay experiments, archived in the PubChem BioAssay data collection. This presentation showcases how these data can be used to develop machine-learning-based bioactivity prediction models for small molecules, using the bioactivity data for three nuclear receptors: retinoid X receptor-α (RXRA), farnesoid-X receptor (FXR), and glucocorticoid receptor (GR). The quantitative high-throughput screening (qHTS) data from the Tox21 project, stored in PubChem, was used as a training set for the model development. The models were built using several supervised machine learning methods (support vector machine, random forest, neural network, k-nearest neighbors, decision tree, and naïve Bayes). The general applicability of the developed models was evaluated with external validation data sets, also archived in PubChem. In addition, this presentation discusses some important aspects of bioactivity data archived in PubChem in the context of bioactivity prediction using artificial intelligence and machine learning (AI/ML).

Developing small-molecule bioactivity prediction models using publicly available bioactivity data in PubChem

Fall 2022 abstracts, posters, and presentations will launch in mid-September.

ACS Fall 2022

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.