Prediction of matrix metal proteinases-12 inhibitors by machine learning approaches

Abstract

Matrix metal proteinases-12 (MMP-12) is a hot pharmaceutical target on the treatment of many human diseases. There's a crying need for designing and finding new MMP-12 inhibitors. In this work, four machine learning approaches, support vector machine, k-nearest neighbor, C4.5 decision tree, and random forest, were employed to derive statistical models from datasets with well distributed biological activities and predict a compound whether it is a MMP-12 inhibitor. The prediction accuracies of the models are in the range of 96.15-98.08% for sensitivity, 87.23-100.00% for specificity, 91.92-98.99% for the overall prediction accuracy and 0.8401-0.9800 for Matthews correlation coefficient, all producing satisfactory results. By means of diverse feature selection methods, several sets of critical descriptors with key information of inhibitory properties were selected by different models, accelerating the classification for MMP-12 inhibitors and non-inhibitors.

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

“…In this study, the computed D ( A ) values are 0.4716, 0.4824 and 0.4494 for the whole data set, the training set and the testing set, respectively, which are much higher than that of the external validation set in recent literature [29], and also superior to that of the data set in our previous work [30], as shown in Table 1. These results reflect considerable structural diversity for our data sets.…”

Virtual Screening and Discovery of Matrix Metalloproteinase‐12 Inhibitors by Swarm Intelligence Optimization Algorithm‐Based Machine Learning

“…In the present study, the model parameters of the LibSVM toolkit were optimized by swarm intelligence optimization algorithms, and six models (PSO_SVM, ABC_SVM, CS_SVM, GWO_SVM, SA_SVM, and GSA_SVM) were established based on the same training and test sets of 189 molecular descriptors used in a previous study . Table shows the parameter settings used for the six algorithms.…”

“…We fixed the training set of the “RF+opt” model developed previously and sorted the test set of 1,277,331 compounds in the ZINC database into 10 sub‐sets . After de‐weighting and energy optimization, 1,268,418 compounds remained, and the “RF+opt” tool was used for prediction, which screened out 184,226 possible inhibitors.…”

“…In this study, we applied virtual screening technology based on swarm intelligence optimization algorithms to search for potential MMP‐12is with new structural properties and expand on previously published work . The action patterns of MMP‐12 and selected inhibitors were studied using molecular docking techniques.…”

Machine Learning Approaches for Metalloproteins

“…Judging from amino acid frequencies at different positions across 4000 peptide inhibitor samples, substantial new knowledge on what signature amino acid sequences should inhibitors of specific MMP types have and how specific the binding would be was added. Aside from Song et al, there was another work on a particular MMP by Li et al [64]. Inhibitors for the MMP-12 enzyme were predicted using k-nearest neighbor (k-NN), random forest, C4.5 decision tree, and SVM.…”