Classification of Acetylcholinesterase Inhibitors and Decoys by a Support Vector Machine

Abstract

Acetylcholinesterase has long been considered as a target for Alzheimer disease therapy. In this work, several classification models were built for the purpose of distinguishing acetylcholinesterase inhibitors (AChEIs) and decoys. Each molecule was initially represented by 211 ADRIANA.Code and 334 MOE descriptors. Correlation analysis, F-score and attribute selection methods in Weka were used to find the best reduced set of descriptors, respectively. Additionally, models were built using a Support Vector Machine and evaluated by 5-, 10-fold and leave-one-out cross-validation. The best model gave a Matthews Correlation Coefficient (MCC) of 0.99 and a prediction accuracy (Q) of 99.66% for the test set. The best model also gave good result on an external test set of 86 compounds (Q=96.51%, MCC=0.93). The descriptors selected by our models suggest that H-bond and hydrophobicity interactions are important for the classification of AChEIs and decoys.

Identification of Potent and Selective Acetylcholinesterase/Butyrylcholinesterase Inhibitors by Virtual Screening

“…The best models showed robust performance with AUC-ROC values ranging from 0.83 ± 0.03 to 0.87 ± 0.01 (Figure A). Consistent with our results, several previous studies have reported that machine learning models can successfully predict inhibitors of AChE/BChE activity, such as NB and SVM models for the prediction of BChE inhibitors and AChE inhibitors. − However, these machine learning models were intended to predict generic inhibitors of either AChE or BChE with no consideration for target selectivity. Molecular docking methods were mainly used for virtual screens of compound libraries to identify AChE/BChE selective inhibitors. , A major challenge in applying these molecular docking methods to virtual screening is that the docking score may not be a reliable indicator of compound activity .…”

Predictions of BuChE Inhibitors Using Support Vector Machine and Naive Bayesian Classification Techniques in Drug Discovery

“…The SVM prediction accuracy of the best model is up to 88.0% for AChEIs and 79.6% for non-AChEIs. Yan and colleagues 36 built SVM models using a larger number of compounds and data than previous work, and the best model gave a Matthews correlation coefficient of 0.99 and a prediction accuracy (Q) of 99.66% for the test set. However, up to now, there is limited research on classification predictions of the BuChE inhibitors and noninhibitors.…”

Impact of polyphenols on receptor–ligand interactions by NMR: the case of neurotensin (NT)–neurotensin receptor fragment (NTS1) complex

“…Emerging evidence has shown that neurotensin (NT), a tridecapeptide neurotransmitter, could play a key role in the excessive release of glutamate via its neurotensin receptor 1 (NTS1) (Ferraro et al, 2008;Ferraro et al, 2009;Wang et al, 2014). Moreover, several authors have postulated that antagonists of NTS1 could be a possible new therapeutic strategy for these neurodegenerative diseases (Antonelli et al, 2007;Ferraro et al, 2008;Ferraro et al, 2009;Wang et al, 2014).…”

“…Emerging evidence has shown that neurotensin (NT), a tridecapeptide neurotransmitter, could play a key role in the excessive release of glutamate via its neurotensin receptor 1 (NTS1) (Ferraro et al, 2008;Ferraro et al, 2009;Wang et al, 2014). Moreover, several authors have postulated that antagonists of NTS1 could be a possible new therapeutic strategy for these neurodegenerative diseases (Antonelli et al, 2007;Ferraro et al, 2008;Ferraro et al, 2009;Wang et al, 2014). However, studying in vitro the release of glutamate via NTS1 is very complicated and even more so when attempting to examine inhibitors of this metabolic pathway (Antonelli et al, 2008;Chen et al, 2006;Ferraro et al, 2000;Matsuyama et al, 2003).…”