You are here:
In silico study of in vitro GPCR assays by QSAR modeling
Citation:
Mansouri, K. AND R. Judson. In silico study of in vitro GPCR assays by QSAR modeling. Chapter 16, Emilio Benfenati (ed.), In Silico Methods for Predicting Drug Toxicity - vol. 1425 of Methods in Molecular Biology . Springer Science + Business Media, New York, NY, 1425:361-381, (2016).
Impact/Purpose:
Book chapter to describe a method to fill date gaps with in silico methods via QSAR models for GPCR receptor assay data and demonstrate the ability of QSAR models to predict bioactivity.
Description:
The U.S. EPA is screening thousands of chemicals of environmental interest in hundreds of in vitro high-throughput screening (HTS) assays (the ToxCast program). One goal is to prioritize chemicals for more detailed analyses based on activity in molecular initiating events (MIE) of adverse outcome pathways (AOPs). However, the chemical space of interest for environmental exposure is much wider than this set of chemicals. Thus, there is a need to fill data gaps with in silico methods, and quantitative structure-activity relationships (QSARs) are a proven and cost effective approach to predict biological activity. ToxCast in turn provides relatively large datasets that are ideal for training and testing QSAR models. The overall goal of the study described here was to develop QSAR models to fill the data gaps in a larger environmental database of ~32k structures. The specific aim of the current work was to build QSAR models for 18 G-Protein Coupled Receptor (GPCR) assays, part of the aminergic category. Two QSAR modeling strategies were adopted: classification models were developed to separate chemicals into active/non-active classes, and then regression models were built to predict the potency values of the bioassays for the active chemicals. Multiple software programs were used to calculate constitutional, topological and substructural molecular descriptors from two-dimensional (2D) chemical structures. Model-fitting methods included PLSDA (partial least squares discriminant analysis), SVM (support vector machines), kNN (k-nearest neighbors) and PLS (partial least squares). Genetic algorithms (GAs) were applied as a variable selection technique to select the most predictive molecular descriptors for each assay. N-fold cross-validation (CV) coupled with multi-criteria decision making fitting criteria were used to evaluate the models. Finally, the models were applied to make predictions within the established chemical space limits. The most accurate model was for the bovine non-selective dopamine receptor (bDR_NS) GPCR assay, for which the classification balanced accuracy reached 0.96 in fitting and 0.95 in 5-fold CV, with only 2 latent variables. These results demonstrate the ability of QSAR models to predict bioactivity.
