Androgen receptor binding affinity: a QSAR evaluation

Abstract

The multiparameter formulation of the COmmon REactivity PAttern (COREPA) approach has been used to describe the structural requirements for eliciting rat androgen receptor (AR) binding affinity, accounting for molecular flexibility. Chemical affinity for AR binding was related to the distances between nucleophilic sites and structural features describing electronic and hydrophobic interactions between the receptor and ligands. Categorical models were derived for each binding affinity range in terms of specific distances, local (maximal donor delocalizability associated with the oxygen atom of the A ring), global nucleophilicity (partial positive surface areas and energy of the highest occupied molecular orbital) and hydrophobicity (log Kow) of the molecules. An integral screening tool for predicting binding affinity to AR was constructed as a battery of models, each associated with different activity bins. The quality of the screening battery of models was assessed using a high value (0.9) of the Pearson contingency coefficient. The predictability of the model was assessed by testing the model performance on external validation sets. A recently developed technique for selection of potential androgenically active chemicals was used to test the performance of the model in its applicability domain. Some of the selected chemicals were tested for AR transcriptional activation. The experimental results confirmed the theoretical predictions.

Identifying potential endocrine disruptors among industrial chemicals and their metabolites – development and evaluation of in silico tools

“…Many of these models can be applied only to a limited number of compounds. Models with relatively wide applicability have been developed for predicting binding affinity to the estrogen Serafimova et al, 2007;Li and Gramatica, 2010b;Toropov et al, 2012;Yi and Zhang, 2012;Zhang et al, 2013) and androgen receptors Li et al, 2009Li et al, , 2013aLi and Gramatica, 2010a;Jensen et al, 2011;Todorov et al, 2011). To our knowledge, studies on the use of QSAR models for prioritizing potential endocrine disruptors considering multiple endpoints were limited to two mechanisms of action (Li and Gramatica, 2010b) or to one group of compounds sharing similar structural features (Juberg et al, 2013), for example brominated flame retardants (Kovarich et al, 2011).…”

Cheminformatic Analysis and Machine Learning Modeling to Investigate Androgen Receptor Antagonists to Combat Prostate Cancer

“…Their model has already been validated by external experimental evaluation. Apart from that, there are a number of studies focusing on QSAR/machine learning modelings of AR agonists/antagonists. − Among the previous machine learning modeling studies, the project of CoMPARA, namely, the collaborative modeling project for androgen receptor activity, initiated by the U.S. Environmental Protection Agency (EPA) to explore endocrine-disrupting chemicals (EDC)­s, is the most prominent project with large amounts of high-quality data sets of various origins, rigorous consensus model based on a total of 91 externally submitted models, and reliable model performance of 80% accuracy . However, the CoMPARA project is focused on evaluating environmental chemicals and toxicants that may play the roles of EDCs and is not applicable to drug discovery.…”

Optimization of molecular docking scores with support vector rank regression

“…To train an SVR model, the log‐transformed binding affinities were used as the target values for binding conformations. As in many previous studies that used affinity data to optimize docking scores, the affinity data, K d and K i , were treated equally . Two training datasets were constructed for the SVR models.…”

“…A number of studies have attempted to improve the correlation between docking scores and binding affinities using statistical learning approaches that integrate additional variables that indicate the fitness of a computed binding conformation . These works all evaluated their results in a library‐screening context.…”