Bio-Inspired Data Mining for Optimizing GPCR Function Identification

GPCRs are the largest family of cell surface receptors; many of them remain orphans. The GPCR functions prediction represents a very important bioinformatics task. It consists in assigning to the protein the corresponding functional class. This classification step requires a good protein representation method and a robust classification algorithm. However, the complexity of this task could be increased because of the great number of GPCRs features in most databases, which produce combinatorial explosion. In order to reduce complexity and optimize classification, the authors propose to use bio-inspired metaheuristics for both the feature selection and the choice of the best couple (feature extraction strategy [FES], data mining algorithm [DMA]). The authors propose to use the BAT algorithm for extracting the pertinent features and the genetic algorithm to choose the best couple. They compared the results they obtained with two existing algorithms. Experimental results indicate the efficiency of the proposed system.