Application of machine learning in understanding bioactivity of beta-lactamase AmpC

Abstract

Abstract The ability of microorganisms like bacteria to develop mechanisms against the treatment is becoming a concern globally. This topic of concern is called Antimicrobial Resistance aka AMR. In this study, with the help of machine learning algorithms we are trying to evaluate the activity of molecules that have been tested experimentally either to bind or not bind the beta lactamases. Machine learning is a technique for analysis of data which teaches the computers what naturally comes to living organisms. Beta lactamases are diverse family of microbial enzymes that hydrolyse the cyclic amide bond of susceptible to beta-lactam antibiotics. Studying the effects and functioning of beta lactamases enzymes can provide better insights into the AMR mechanism adopted by the microorganisms. AMR is one of the top 10 global public health threats facing humanity in this era. Therefore, finding potential compounds that can combat these microorganisms is very important. Here, we have considered few plant-based flavonoids and terpenoids and checked the bioactivity against these beta lactamases containing microorganisms by using machine learning algorithms. A large dataset having more than 62,000 compounds and their pPotency values against beta lactamase AmpC was obtained from ChEMBL and employed in QSAR (quantitative structure activity relationship) model in order to understand the origin of their bioactivity. Several set of fingerprint descriptors and predictive models were constructed and results are obtained.