Structural Analysis Of Nucleoprotein And Study On Prediction Of Protein Subnuclear Localization

Nucleoprotein is correlative with many genetic diseases and biological heritages of humans,the domains and the motifs play an important role in the protein sequence,so it is very important to study the domain and the motif.And the study on the nucleoprotein domain and motif can help us understand the mechanism of subnuclear localization.In this paper,nucleoprotein has been researched.The dataset NP1505 of 6 subnuclear localization proteins was constructed based on protein sequence database UniProtKB/ Swiss-Prot,the domain and motif information of each nucleoprotein was extracted,it was found that the unique domains of chromosome region proteins include H15,BUB1 n-terminal,Ku etc,the unique domains of nucleolar regional proteins include S1 motif,PUM-HD,Brix,the unique domains of nucleus membrane regional proteins include SUN,MIR,IQ etc,and the domain RRM were shared among proteins in 6 regions.By referring to a large number of literatures and protein databases,we find the functions of these domains and the secondary structure and tertiary structure diagrams in the PDB database.In addition,the information content map of the motif was generated by the Weblogo online server,and the characteristic motif of the nucleoprotein was analyzed.These characteristics information can be used to further predict the subnuclear localization of proteins,and provide help for the study of the nucleoprotein action mechanism.Based on the above studies of the domains and motifs of nucleoprotein,we reestablished a new dataset NP1118 of nine classes nucleoprotein with sequence similarity of 30%,the domain features of nucleoprotein were used to predict the subnuclear localization of proteins,and the characteristic parameters of amino acid composition and dipeptide component,protein blocks,gene ontology,stickiness of amino acid were extracted,then support vector machine algorithm was used to classified the nucleoprotein of nine subnuclear regions.The prediction results of single feature gene ontology information is the best,and the overall prediction accuracy is 66.91% with Jackknife test.In the end,the features were combined,the prediction results of the hybrid feature are higher than the single feature,the total prediction accuracy is up to 70.39% in the Jackknife test.