Research On The Methods Of Identifying Complexes Based On Protein-Protein Interaction Networks

With the deepening of proteomics and systems biology,identifying protein complexes in protein-protein interaction(PPI)networks is a fundamental problem in computational biology.High-throughput experimental techniques have generated large,experimentally detected PPI datasets.These interactions represent a rich source of data that can be used to detect protein complexes;however,such interactions contain much noise.Therefore,these interactions should be validated before they could be applied to detect protein complexes.Focuses on the topic of identifying protein complexes,the paper makes intensive studies on the graph clustering algorithms based on PPI networks.The followings are main innovations and research achievements:(1)In order to reduce the high-throughput experimental techniques of protein interaction data generated noise,this paper analyzes the characteristics of protein interaction networks from the topology,and proposes a new method PPIR(Protein-Protein Interaction Reliability)to reduce noise in the protein-protein interaction data.Two protein pairs and their common neighbor nodes are considered as a single module,and the PPIR value in the percentage of this module mainly corresponds to these common neighbor nodes.PPIR value was then calculated as weight between proteins,and then build a weighted interaction networks.Results show that,the paper compared PPIR with a representative set of other topology-based weighting schemes methods,PPIR method for identifying protein complexes exhibit a very good experimental results.In addition,PPIR method can improve the performance of the graph clustering algorithms based on PPI networks.(2)PPIRU,a novel protein complex clustering algorithm based on PPIR,is proposed.The PPIRU algorithm consists of three steps.Firstly,each binary protein interaction is assigned a weight by using PPIR,showing a true positive interaction.Secondly,the results are then clustered using RRW,MCL and ClusterONE algorithms.Finally,the integration the results of these three graph clustering algorithms.Thus,experiments obtained clustering results of PPIRU algorithm for identification of protein complexes.In order to evaluate the effectiveness of PPIRU algorithm,this paper use PPIRU algorithm and several classical algorithms for comparison,and using DIP and BioGRID protein data sets and the results are compared with the standard MIPS and SGD complex sets respectively.The paper compare PPIRU with several efficient,existing clustering algorithms and reveal that the accuracy values of PPIRU clusters are much higher than those of other algorithms.