Protein molecules play essential roles in the living organisms. The knowledge about their functions is very important in order to design new drugs that could be used to control various processes in the organisms. The determination of the protein functions could be performed by detecting the binding sites where interactions between proteins occur. In this paper we focus on predicting the protein binding sites. First, several characteristics of the amino acid residues are extracted. Then, prediction methods are induced. In this research paper we consider several classification methods for inducing models. In order to enhance the predictions, we use ensembles, which combine several classification models. The results show that using ensembles, the prediction power is increased.
protein function, protein interaction, protein binding site, BIND database, ensembles.