Identification of protein complexes algorithm based on random walk model

Advanced technologies are producing large-scale protein-protein interaction data at an ever increasing pace. Finding protein-protein interaction complexes from large PPI networks is a fundamental problem in bioinformatics. In order to solve the set contains false negative and false positive noise data of protein data, we develop the graph model for limitation of structural protein complex. We proposed a RWSPFinder protein complexes identification algorithm based on random walk model, Based on the random walk algorithm to predict protein network that exists on the network of protein interaction data, the false negative or false-positive noise data can be removed. According to the GO ontology for computing semantic similarity between protein complexes, ultimately determine the identification of protein complexes. The algorithm is not sensitive to the input parameters, the experiments verify the effectiveness of the proposed algorithm.