Research On Protein Complex Accurate Recognition Based On Machine Learning

In the context of the completion of the human genome sequencing project,the life sciences are shifting their focus to the protein category.In recent years,with the boom of machine learning algorithms,its application in various fields has attracted much attention,and the field of proteomics has also been greatly affected.As the PPI network continues to grow,the content gradually improved,a large amount of data and high complexity of the PPI network challenges related work.At present,the recognition of PPIN complex is mainly based on the application of clustering algorithm.It is found that these algorithms have limitations in varying degrees.Therefore,based on the existing problems in protein network clustering,this paper proposes an improved PSO algorithm--KPSO algorithm for protein network complex detection algorithm.In this paper,we focus on the speed and accuracy of clustering algorithm in protein-protein interaction network,and make related investigations,research,experiments and analysis.It has important academic significance based on improved particle swarm algorithm,fusion topology,function annotation and biological evolution information to the detection of protein module.Specific description is as follows:(1)Analyze and compare the performance advantages and disadvantages of several solutions in PPIN clusteringAt present,the algorithms used in PPIN clustering are varied.However,with the development of protein network,the complexity and scale of protein network are changing,so some algorithms are not suitable or the solution to the problem is not good.In this paper,several common solutions are introduced,analyzed and compared,and the advantages and disadvantages of each strategy are summarized.Hoping to learn from it,for the strategy of this article provides a valuable reference.(2)To discuss the problems that need attention in the clustering research of PPIN functional modulesAt present,clustering algorithm has made some remarkable achievements in the detection of functional modules of PPIN.However,the clustering algorithm is a highly subjective method,and because researchers in the field of computer science lack of knowledge of the biological field,it is often easy to ignore the biological characteristics of protein data,resulting in unreasonable clustering results.Therefore,we need to pay attention to the problems in the process of protein-protein interaction network clustering,such as the selection of algorithms,the acquisition and processing of data,the definition of similarity,the modeling of network and the setting of algorithm parameters.(3)An efficient,stable and accurate PPIN clustering strategy is proposedThis paper proposes an efficient and robust PPIN detection strategy to overcome the shortcomings of current algorithms by studying and comparing existing algorithms,and discussing algorithms selection,data acquisition and processing,similarity definition and so on.The basic principle,process description and experimental verification of the proposed detection strategy are given.