PREDICTING THE FUNCTION OF PROTEINS USING DIFFERENTIAL EVOLUTION

摘要

The proteins perform major of the most important tasks in the organism of any live being. Since the number of functions performed by them is large and there are relationships of dependency among them, knowing their function is an important challenge in bioinformatics. Thus, some methods of the Artificial Intelligence have been used for the hierarchical classification of proteins. In this work the RCMDE (Rule Construction Method using Differential Evolution) algorithm is proposed. The main objective of RCMDE is to construct a single label local hierarchical classifier using Differential Evolution. RCMDE is compared with LMNBwU, a hierarchical classifier based on Naive Bayes algorithm. RCMDE have shown good capacity of exploration of the space state and to construct good rules. In the experiments, using the enzymes datasets, RCMDE outperformed or had similar results with LMNBwU in all the experiments.