ABSTRACT

Tuberculosis, the deadly airborne disease that has now re-emerged as Multi-drug resistant tuberculosis, poses a major health threat to mankind. It has created a global concern for researchers to find a new drug that can fight against this dreadful bacterium. The bacterial DNA Gyrase B has been identified as an important drug target for the disease. Even though high-throughput screens to identify the anti-tubercular activity of small molecules are now available, they are expensive and time-consuming. Computational methods including Machine Learning approaches, QSARmodelling, and Docking process are less timeconsuming as well as efficient. This attracts us to build classifiers for virtual high-throughput screening and to select molecules from large libraries for further analysis. We developed four supervised classification models (SMO, Random Forest, Naive Bayes, and MLR) using Weka to classify the molecules as actives and inactives and four regression models to predict the inhibitory activities of these active molecules. The filtered molecules were further docked to the Gyrase B protein (PDB ID: 4B6C ) to understand the interacting residues and the output poses were also rescored to understand the stability of the ligand-protein complex. Finally, we have screened 6 phytomolecules from a phytochemical database IMPAAT by applying these developed models and predicted their inhibitory activities. The pharmacokinetics and toxicity of these molecules were also studied. These molecules could be modified to increase the potency and thus can be used to develop drugs against Mycobacterium tuberculosis. Keywords: DNA Gyrase, Machine learning,Docking,QSAR, ADMET, Virtual Screening
Publication date: 01/12/2021

https://ijbpas.com/pdf/2021/December/MS_IJBPAS_2021_5774.pdf

Download PDF
https://doi.org/10.31032/IJBPAS/2021/10.12.5774