Thyroid hormone receptor β (ThrB) is a ligand-depended nuclear receptor responsible for the mediation of biological thyroid hormone sensitivity. The present work aims to develop machine learning-based prediction models for ThrB activity of small molecules, using public-domain bioactivity data available in PubChem.  Binary classification models that predict ThrB active/inactive compounds were constructed with a quantitative high-throughput screening (qHTS) data set of 10,000 compounds obtained from the Tox21 screening library, archived in PubChem.  Nine fingerprints and six machine learning algorithms were used for model development.  The general applicability of the developed models was evaluated on two external data sets, which were also stored in PubChem: one containing literature-extracted bioactivity data obtained from ChEMBL and the other containing qHTS screening data from the NCATS Chemical Genomics Center (NCGC).  Fragment enrichment analysis was performed to characterize the structural characteristics of the active compounds. This work showcases the use of public-domain bioactivity data and open-source software packages.

Thyroid hormone receptor β activity prediction of small molecules by using machine learning and public-domain 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.