“Computational Investigation of Pyrazole Derivatives as Antitubercular Agents: A QSAR-Based Approach”

Introduction: Tuberculosis is one of the leading global problems, and the appearance of drug-resistant forms of Mycobacterium tuberculosis makes it urgent to find new anti-tuberculosis drugs. DNA gyrase is an enzyme that plays a key role in the replication of bacterial DNA and can be a target for the action of new antibiotics. In silico QSAR and molecular docking studies can help to determine the features of hit compounds. Objectives: The current problem is the development of quantitative structure-activity relationship (QSAR) multiple linear regression (MLR) models for anti-tuberculosis compounds and identifying molecular descriptors with a high correlation to determine the activity of new potential DNA gyrase inhibitors. Methods: Using the MLR method, a QSAR model was developed based on the data set of 14 anti-tuberculosis compounds. Fourteen compounds were used for regression analysis, excluding one compound with a missing molecular descriptor. After calculating the molecular descriptors, the most optimal four descriptors were selected for the MLR-QSAR model. Statistical characteristics of the model were determined: coefficient of determination (R 2 ) and adjusted coefficient of determination (adj. R 2 ), mean square error (MSE), root mean square error (RMSE), F-test, p-value, residual analysis, and leave-one-out cross-validation. Results: A QSAR model was obtained that showed a good correlation between the predicted and actual values of biological activity of the compounds according to R² = 0.9438 and adjusted R² = 0.9188 , respectively. The model’s MSE was 0.02444, RMSE – 0.1563, p-value – 3.77×10 -109 , F-test – 37.766(p < 0.0001 ), and average absolute error – 2.823%. The following equation was found for the MLR-based QSAR model; Activity=−139.3936+2.3916(SM11_EA)−3.6094(VE1_A)+41.1803(SpMAD_L)+302.7458(X4A).Among the molecular descriptors used, VE1_A has the highest standardized regression coefficient in absolute value. Conclusions: The QSAR model based on the MLR algorithm demonstrates a significant correlation between molecular descriptors and biological activity of a substance. The developed equation allows for making initial estimates of the activity of related chemical compounds and thus for selecting promising substances for further preclinical and clinical trials. However, it is vital to keep in mind that the QSAR model works only within the range of the sample used.