Machine Learning Models Combined with Virtual Screening and Molecular Docking to Predict Human Topoisomerase I Inhibitors

Abstract

In this work, random forest (RF), support vector machine, k-nearest neighbor and C4.5 decision tree, were used to establish classification models for predicting whether an unknown molecule is an inhibitor of human topoisomerase I (Top1) protein. All these models have achieved satisfactory results, with total prediction accuracies from 89.70% to 97.12%. Through comparative analysis, it can be found that the RF model has the best forecasting effect. The parameters were further optimized to generate the best-performing RF model. At the same time, features selection was implemented to choose properties most relevant to the inhibition of Top1 from 189 molecular descriptors through a special RF procedure. Subsequently, a ligand-based virtual screening was performed from the Maybridge database by the optimal RF model and 596 hits were picked out. Then, 67 molecules with relative probability scores over 0.7 were selected based on the screening results. Next, the 67 molecules above were docked to Top1 using AutoDock Vina. Finally, six top-ranked molecules with binding energies less than −10.0 kcal/mol were screened out and a common backbone, which is entirely different from that of existing Top1 inhibitors reported in the literature, was found.

AI in drug development: a multidisciplinary perspective

“…The application of QSAR modelling in medicinal chemistry projects has been extensive in the last decades, helping to solve problems in a wide variety of topics. One of its major applications has been the search for active compounds against a therapeutic target, where the QSAR model itself can be applied as a filter in the screening of chemical libraries [111][112][113][114][115][116]. Another important application is the development of models capable of predicting a relationship between chemical structures and different types of toxicity, such as In vitro toxicity [117], In vivo toxicity [118], mutagenesis [119] or hepatotoxicity [120], among others.…”

Predicting New Anti-Norovirus Inhibitor With the Help of Machine Learning Algorithms and Molecular Dynamics Simulation–Based Model

“…Machine learning models could be beneficial for lead optimization and chemical compound prioritization when using computer-aided drug design ( Lavecchia, 2015 ). Statistical learning algorithms, namely, Naïve Bayesian ( Murakami and Mizuguchi, 2010 ; Fang et al, 2013 ) random forests (RFs) ( Jayaraj et al, 2016 ; Wei et al, 2016 ; Li et al, 2019a ; Wei et al, 2020 ), support vector machines (SVMs) ( Han et al, 2008 ; Mahé and Vert, 2009 ; Fang et al, 2013 ; Jayaraj and Jain, 2019 ; Wei et al, 2019 ), decision stump ( Nand et al, 2020 ), artificial neural networks (ANNs) ( Lobanov, 2004 ; Li et al, 2019b ), and k nearest neighbors (kNNs) ( Mahé and Vert, 2009 ), have been used to build models and effectively employed in virtual screening, prediction of protein–protein interactions, ADMET prediction, and pharmacokinetic studies with substantial outputs. Kadioglu and co-workers applied a workflow of combined virtual drug screening, molecular docking, and supervised machine learning algorithms to identify novel drug candidates against COVID-19 ( Kadioglu et al, 2021 ).…”

“…The best model was further used to virtually screen the SPECS database for NA inhibitors ( Zhang et al, 2017a ). In the reported work of Li et al, RF, SVM, kNN, and C4.5 decision tree models were used to discriminate inhibitors of the human topoisomerase I (Top1) protein from the non-inhibitors with total prediction accuracies ranging between 89.70 and 97.12% ( Li et al, 2019a ). Among machine learning algorithms, the RF model was detailed as the best model and was used to virtually screen the Maybridge database for Top1 inhibitors.…”

Prediction of Premature Termination Codon Suppressing Compounds for Treatment of Duchenne Muscular Dystrophy Using Machine Learning

“…In another study, Li and colleagues [ 25 ] conducted classical ML-based elucidation of new inhibitors of topoisomerase I. They prepared an input set containing 481 inhibitors and 480 non-inhibitor compounds.…”