Uncover protein complexes in E.coli network

Recent advances in proteomic technologies have enabled high-throughput binary data on protein-protein interactions of E. coli to be released into public domain, and many protein complexes have been identified by experimental methods. Although it has a long study history, a large-scale analysis of protein complex in binary PPI network of E. coli is still absent. We used a novel link clustering algorithm named ELPA to infer protein complexes and functional modules in E. coli PPI network. By mapping our results to 276 gold standard protein complexes and protein function annotations offered by EcoCyc, we found that 80.2% of predicted modules mapping well with one or more complexes, while 92.8% of predicted modules tally well with certain GO terms. Furthermore, we compare our results with MCL algorithm, and evaluated our results with several accuracy measures and biological relevance, the result shows that ELPA achieved an average 18.3% improvement over MCL based on the accuracy measures, which means our method will contributes to uncover the complexes of Ecoli.