Research On Dynamic Protein Complexes And Key Protein Recognition Algorithms Based On Group Intelligence Optimization

In the post-genomic era,with the development of bioinformatics,high throughput technology produced large amounts of biological data,which made it become one of hottest research that systematically analyze and understand protein as the undertaker of the life activities through study the function,structure and interaction of proteins by computational methods.Especially,the identification of protein complexes and essential protein from the protein-protein interaction(PPI)network is of great significance to ex-plain particular biological process and reveal the characteristics of life.In recent years,although there are many breakthroughs in the research of protein complexes and essential protein,precision of detecting protein complexes and essential proteins is still very low because of the small world scale-free feature,big noise of PPI network,and the limitations of identification methods.Swarm intelligence optimization algorithm is a heuristic imitation biological algorithm,which can solve the optimization problem,and it also provides a new perspective for the identification of protein function module and essential proteins.In this article,the precision of protein complexes identi-fication was improved by using swarm intelligence algorithm to optimize the parame-ters of other clustering methods or constructing new clustering model based on the op-timization procedure of swarm intelligence algorithm,and the precision of essential protein identification was improved by simulating the optimization process of swarm intelligence algorithm.The main studies of this article are as follows:Firstly,aiming at the difficulty of parameter selection of density clustering algo-rithm DBSCAN,a modified density-based clustering method was proposed to detect protein complexes.Pigeon optimization algorithm is used to dynamic adjust DBSCAN's two parameters,which are parameter of neighborhood radius and parameter minimum of neighbor nodes.The most appropriate parameters of DBSCAN improved the performance of protein complexes identification.The experiments show that the modified DBSCAN can get better protein complexes.Secondly,aiming at the defect of the core-attachments clustering algorithms that there are no uniform or standard definition for the combination of core and attachments,a new protein complexes identification method was proposed by simulating the process of fruit flies foraging.This new method combines the fruit flies' foraging optimiza-tion behavior and the core-attachment structure of protein complexes.The optimization process made every attachment protein find the most appropriate core and then formed more accurate protein complexes.The experiments show that the new clustering method has a better performance than the traditional core-attachment protein complexes identi-fication methods.Thirdly,aiming at the limitation of traditional clustering methods that can not con-sider PPI network from local and global at the same time,a new method was proposed by simulating the pigeon optimization algorithm.This method based on the superiority of pigeon' s global search and local search and formed protein complexes by global and local searching in the PPI network.The experimental results shown that the new clus-tering method can get more accurate protein complexes than traditional methods.Finally,most of the essential protein identification methods analyze the essential protein based on the isolated or piecemeal characteristics,and they lack a comprehen-sive and overall grasp for the protein node.Aimed at this shortage,a new essential pro-tein identification method based on the artificial bee colony optimization algorithm was proposed.This method combined the priori knowledge of essential protein,topological properties of PPI network,gene expression characteristics and the optimizing characte-ristics of artificial bee colony.To begin with,the mining bees and following bees search the new nectar source by a secondary-search.If they could not find the best new nectar source in the first stage,scout bees were followed to find the new nectar source through a global research.The experimental results show that the new identification method have a better performance than traditional methods.