Artificial intelligence to predict dynamic properties of amino acids of small hydrolases using molecular dynamic approaches

Abstract

Molecular dynamics (MD) is a really useful tool in the hands of the modern scientist of computational expert. It is possible to find the macroscopic properties of a system through microscopic simulations. The study of protein dynamics in the lab is a very complicated, expensive and time-consuming process. Therefore, a lot of effort and hope lies with the computers and the in silico study of protein structure using molecular dynamics. Present study uses MD simulation to explore relation between sequence conservation and dynamism of amino-acid proteins. A total of 50 AI models were developed. After development, all AI models were evaluated and compared on the basis of performance, ROC, AUC and confusion matrix. AdaBoostM1 classifiers models were found most promising on the basis of Accuracy. In order to predict dynamic or static part of proteins, all 50 AI models showed accuracy from 66.05% to 89.26%. newline