Exploring the application of gene ontology semantic similarity measure for identifying protein complexes

In recent years, numerous protein-protein interaction (PPI) datasets have been generated with the development of high-throughput experimental techniques. These datasets enable researchers to uncover protein complexes on network level. However, the performance of the computational methods relies heavily on the quality of the underlying protein interaction data, and these datasets are usually quite noisy. Protein complex identification results are often affected by these datasets which have high false positive and false negative rates. To address this problem, we adopt a gene ontology (GO) semantic similarity measure to evaluate the reliability of PPI networks to reconstruct networks. We apply three protein complexes identification algorithms to these reconstructed networks. The experimental results demonstrate the effectiveness obtained by incorporating the GO semantic similarity measure.