Research On Protein Complex Recognition Algorithm Based On Core-attachment

With the development of high-throughput technology,a lot of protein databases are produced to be used to research.In the field of bioinformatics,the research hotspots are protein function,protein structure and protein-protein interaction.They have an significant impact on cell vital movements.PPI network is composed of protein and proteins' interaction.Protein complex is the base unit which can complete cell life activity.So,it is very important to identify protein complexes with biological significance from protein interaction network efficiently and accurately.Today,a great many of protein complex prediction algorithms run in the static PPI network.These methods ignore the dynamic properties of proteins.In fact,owing to time,environment and other factors,proteins and their interaction always change.In order to construct a dynamic PPI network,we integrate the static PPI data and gene expression data because gene expression level and protein expression level are consistent.In this paper,we proposed some method to identify protein complex based on the structural characteristic of protein complex.The main research work of this paper is as follows:(1)The improved ant colony optimization algorithm is used to identify protein complexes.The basic framework of this method is the core-attachment structure.After obtaining the core of complex,we mark every neighbor protein of core with unique label through picking and dropping principle of the improved ant colony optimization algorithm.Then,we get prediction protein complex set.We considered that external factors have contributed to the irregular changes of protein.Introducing the random probability makes the algorithm precise.(2)The neighbor affinity model is used to mine protein complexes.By calculating the centrality score of proteins,we can get seed protein.Then,we receive complex core according to the similarity of seed protein and its neighbor proteins.The neighbor affinity theory is used to form the complex attachment.We analyzed the experimental results and compared this algorithm with other state-of-the-art methods.It is found that the proposed algorithm can predict protein complexes accurately.Furthermore,there are more protein complexes which have strong biological significance.(3)The PageRank algorithm is used to predict protein complex.The weight of every protein is calculated in the dynamic PPI network.The weight value is compared with a threshold.And the seed protein is obtained.Taking into account the stability structure of the triangle,we get complex core.The basic unit is composed of seed protein,seed protein's two neighbor proteins and their interactions.To construct the complex attachment,the PageRank algorithm is used to measure the relationship between the complex core and its neighbor proteins.The experimental results show that this algorithm can predict protein complexes with strong biological significance accurately and efficiently.