Extraction and Visualization of Potential Pharmacophore Points Using Support Vector Machines:  Application to Ligand-Based Virtual Screening for COX-2 Inhibitors

Abstract

Support vector machines (SVM) were trained to predict cyclooxygenase 2 (COX-2) and thrombin inhibitors. The classifiers were obtained using sets of known COX-2 and thrombin inhibitors as "positive examples" and a large collection of screening compounds as "negative examples". Molecules were encoded by topological pharmacophore-point triangles. In retrospective virtual screening, 50-90% of the known active compounds were listed within the first 0.1% of the ranked database. To check the validity of the constructed classifiers, we developed a method for feature extraction and visualization using SVM. As a result, potential pharmacophore points were weighted according to their importance for COX-2 and thrombin inhibition. Known thrombin and COX-2 pharmacophore points were correctly recognized by the machine learning system. In a prospective virtual screening study, several potential COX-2 inhibitors were predicted and tested in a cellular activity assay. A benzimidazole derivative exhibited significant inhibitory activity with an IC(50) of 0.2 microM, which is better than Celecoxib in our assay. It was demonstrated that the SVM machine-learning method can be used in virtual screening and be analyzed in a human-interpretable way that results in a set of rules for designing novel molecules.

A Virtual Screening Filter for Identification of Cytochrome P450 2C9 (CYP2C9) Inhibitors

“…Potential pharmacophore models of the inhibitors were visualized by showing to which extent a certain feature is linked to favourable or unfavourable interactions according to the final SVM model [49]. The importance R i of each 3PP feature was calculated based on the change in the SVM prediction for a molecule with this feature removed (Eq.…”

“…In addition to a standard binary classification, SVM regression models were also built to analyse the consistent IC 50 values. These models then allowed visualization of important features [49] of cytochrome-ligand interactions on SAR series from these data. This approach also led us to extract relevant features for binding of S-warfarin to CYP2C9 in agreement with its X-ray structure [25].…”

A simple representation of three-dimensional molecular structure

“…Naive Bayes Classifiers (NB) [9][10][11] , Random Forests (RF) 12,13 , Support Vector Machines (SVM) 10,14,15 , and Deep Neural Networks (DNN) [16][17][18][19][20] predict a molecule's target binding profile and other properties from the features encoded into its 2D fingerprint. SEA and methods building on it such as Optimized Cross Reactivity Estimation (OCEAN) 21 quantify and statistically aggregate patterns of molecular pairwise similarity to the same ends.…”

Design and Synthesis of Benzimidazole Analogs Endowed with Oxadiazole as Selective COX‐2 Inhibitor

“…From the reported literature, benzimidazoles indicated promising pharmacological activities, viz. anticancer [15], analgesic [16], anti-inflammatory [17], COX-2 inhibitory [18][19][20], anthelminthic (mebendazole), antihypertensive (candesartan), antipsychotic (pimozide), proton pump inhibitory (omeprazole), and iondilatory (pimobendan). In order to make high potency drug molecules, one of the strategies would be to combine two bioactive molecules belonging to a particular therapeutic category.…”