Other links:

Other links:

Event Calendar

Loading Events

Machine Learning based approach for identification of novel resistance associated mutations in

Mycobacterium tuberculosis and analysis of the structural basis of drug resistance.

  • This event has passed.

Title: Machine Learning based approach for identification of novel resistance associated mutations in Mycobacterium tuberculosis and analysis of the structural basis of drug resistance.


Whole genome sequences (WGS) are now available for thousands of resistant and susceptible strains of M.tb.  Since, AST (Antibiotic susceptibility testing) for identifying antibiotic resistance in bacteria is a time consuming process, it is necessary to develop computational methods which can not only identify antibiotic resistant strains faster from whole genome sequences, but also can help in deciphering resistance associated mutations.  This can provide insights into mechanistic details of drug resistance. We have developed Machine Learning based models for classifying antibiotic resistant and susceptible M.tb strain using whole genome sequences of 48,218 M.tb isolates. For 13 antibiotics, 13 individual ML models as well as DL models have been built using the XGBoost classifier algorithm and ANN respectively.  Comparison with results from popular antibiotic resistant prediction tools like TB-profiler revealed that,  first-line drugs such as Isoniazid, Rifampicin & Ethambutol, and for second-line drugs such as Kanamycin, Streptomycin and Ethionamide ML based models are performing better than TB-Profiler, whereas in case of other first and second-line drugs such as Ofloxacin, moxifloxacin, cycloserin, PAS, ML model performs almost similar to TB-profiler.  Current study also shows the ability of AI/ML models to identify new mutations associated with drug resistance in coding as well as non-coding regions along with already known resistance associated mutations. Availability of 3D structural models for entire TB proteome from AlphaFold2 has provided opportunity to decipher mechanistic details of drug resistance by mapping these new resistance associated mutations onto 3D structures.

About Speaker:

Debasisa Mohanty earned a post graduate degree (MSc) in physics from the Indian Institute of Technology, Kanpur in 1988 and did his doctoral studies at the Molecular Biophysics Unit of the Indian Institute of Science to secure a PhD in computational biophysics in 1994.Subsequently, he completed his post-doctoral work, first the Hebrew University of Jerusalem and, later, at the Scripps Research Institute. On his return to India, he joined the National Institute of Immunology, India (NII) where he continued his pioneering research in the interface of bioinformatics and disease biology. Currently, he is the Director of NII, and he also supervises the activities of RiPPMiner, (Bioinformatics Resource for Deciphering Chemical Structures of RiPPs) and the Bioinformatics Centre. For his contributions to biosciences, the Department of Biotechnology of the Government of India awarded him the National Bioscience Award for Career Development, one of the highest Indian science awards in 2009.


Study at Ashoka

Study at Ashoka