Function Prediction of Proteins in Yeast Networks Based on the MCL Algorithm

Large-scale Protein-Protein interaction data sets exist in Saccharomyces cerevisiae due to many interaction detection methods such as yeast two-hybrid assay, mass spectrometry of purified complexes, correlated mRNA expression profile and so on. How to make use of these data sets to understand the protein function is very important. We use the algorithm [17] developed by Stijn van Dongen to describe the functional modules in PPI networks.We analyze four protein-protein networks from Saccharomyces cerevisiae, and our results suggest that the functional modules detected are consistent with the biology knowledge. ProteinProtein interaction network was separated into clusters using MCL algorithm. Based on the clusters resulted from MCL algorithm, we assign the function annotations using Pvalue and majority methods. The majority method is based on the majority rule [15]. The predicted function of proteins provide clue to biology experiments. Two methods are used to assign function annotations for the known clusters and unknown proteins, we compare the two predicted results, the results show that the two methods are consistent with each other.