Search academic texts

Information × Registration Number 0222U000067, 0120U103195 , R & D reports Title Biologically active compounds against tuberculosis: molecular basis, targeted search, biological screening. popup.stage_title Head Tukalo Mykhailo A., Доктор біологічних наук Registration Date 02-01-2022 Organization Institute of Molecular Biology and Genetics of NAS of Ukraine popup.description2 The models of artificial neural networks (ANNs) have been developed for the search of inhibitors for Mycobacterium tuberculosis leucyl-tRNA synthetase (LeuRS). The ROC-score of the models after cross-validation was in the range from 0.91 to 0.95 (mean value of 0.93) and the Matthews correlation coefficient was in the range from 0.6 to 0.87 (mean value of 0.76). After validation toward test set the ROC-score was in the range from 0.84 to 0.95 (mean value of 0.9) and the Matthews correlation coefficient was in the range from 0.83 to 0.91 (mean value of 0.89). The model of support vector machine (SVM) was constructed for the search of M. tuberculosis LeuRS inhibitors. The ROC-score for the cross-validated model was 0.9958, and after validation toward test set was 0.86. According to the results of machine learning screening, 85 compounds were selected for investigation of their inhibitory activity toward recombinant M. tuberculosis LeuRS in aminoacylation assay. According to the results of in vitro screening 6 inhibitors of LeuRS have been identified which decrease the activity of enzyme by more than 50% at the concentration of 100 μM. The most active compound – (1H-indol-3-yl)-piperidin-3-yl-methanone, inhibits M. tuberculosis LeuRS by 84% at the concentration of 100 μM. Product Description popup.authors popup.nrat_date 2022-03-09 Close