Semantic similarities as discriminative features of protein complexes

Biological data stored in databases may be associated to biological information (knowledge) such as experiments, properties and functions, response to drugs etc. Such knowledge is often organized and stored in biological ontologies. Gene Ontology (GO) is one of the main resource of biological knowledge. It provides both a categorization of terms and a source of annotation for genes and proteins. This enables the use of ontology-based methodologies for the analysis of proteins and their functions. One methodology uses semantic-based similarity measures, for the calculation of the similarity of two or more proteins starting from their annotations. Recently there is a growing interest in the use of semantic-based methodologies for the analysis of protein interaction data such as the prediction of protein complexes based on semantic similarity measures. Despite this interest, there is the need for an assessment of semantic measures as well as a deep study on the impact of the chosen measure in the obtained results. This paper treats the problem of using semantic similarity measure to analyze protein complexes and to improve protein complexes prediction frameworks. Tests have been performed in yeast and human protein complexes. Results indicate that there exists a bias among different measures as well as an higher value of semantic similarity within proteins that participate in the same complex, proving a possible use of semantic similarity measures within protein complexes prediction algorithms and a way to choose the best one among them.