Development of novel algorithm to detect enzyme active-sites considering cofactors

点击翻译按钮获取中文摘要

Prediction of active sites in enzyme proteins is extremely essential not only for protein sciences but also for practical applications. Because enzyme reaction mechanisms are based on the local structures of enzyme active sites, a simple measurement, mean square deviation, has been used to compare such local structures in proteins so far. In order to improve the ability of such a simple measurement, various kinds of template-based methods that compare the local sites have been developed to date. In this work, parameters for the deviation was introduced. Moreover, the Bregman Divergence Regularized Machine was also employed to develop a new machine learning algorithm that determines the parameters of the square deviation. Experimental results showed that the proposed methods possess promising search performance.