Dynamic And Conservative Protein Complexes Mining Research Based On Multi Omics Data Integration

Protein complexes perform specific life activities in all biological processes,and play the crucial role in complicated life systems.The proteins between with each other in different space-time and their homologous mapping among different organisms show the characteristics of dynamic change and conservatism evolution in life course.Thanks to different and complementary biological information included in different protein interaction data,it is necessary to integrate multi-source protein data for studying the relationship,function,dynamic change and conservative evolution of the interaction.It is also a hot spot in research of protein complex in the post genome era.This paper proposed a method based on group dynamic decision idea for mining sequential protein complexes,and constructed the conservative similarity interaction network for excavating conserved protein complexes on the basis of fusing multi-source omics biological data.The main research work is introduced as follows:Protein complexes mining based on dynamic group decision.Traditional strategy for clustering the static protein-protein interaction networks often ignore the dynamic properties existing between complexes,this paper combined the gene ontology(GO)function annotation data,the time-sequenced gene expression data and the topological characteristics of static interaction network to construct the multi-source dynamic protein-protein interaction network.Meanwhile,a new identified dynamic protein complexes algorithm--IPC-DGD(Identifying Protein Complexes based on Dynamic Group Decision)in sequential network was proposed,which was used for referencing the idea of intelligent group decision.Based on the local density and the relative distance of proteins in network,the suitable solution space of the cluster center was determined,and then the primary clustering of the dynamic protein-protein interaction network was realized.Dynamically simulating the discussion process of human group decision in the process of optimizing initial clusters,the reciprocal rule between individual preference and adjustment strategies between group preference was designed,realized the two dynamic updating schemes for the migration of protein node within the complexes and the integration between complexes,which in order to obtain the better complexes.The experimental results showed that:the proposed IPC-DGD method performed better on the Matching Degree,Sensitivity and F-score index comparing with other state-of-the-art protein complex mine algorithms,and could also efficiently overcome the shortcomings of cluster center's sensitivity,topological limitations and local optimization.Protein complexes mining based on conservative information fusion.Most of the existing protein complex mining is conducted on a single species network,with little consideration for the conservative and evolutionary of interspecies complexes.Therefore,the GO function annotation data,the orthologs data between patterns species,and the known human protein interaction data was fused in this paper,A new framework for calculating functional similarity and homology similarity between proteins was proposed,constructed the conservative similarity network.Through the properties of conserved complexes mined from network and the level of proteins,the conserved characteristics between protein-protein interactions were analyzed to supplement the components of conserved complexes.The experimental results showed that:in the Maximum Matching Rate,Sensitivity and F-score,this method had a more advantageous evaluation value.It could identify more accurate and larger conservative protein complexes,and successfully matched up to 80%of the known conserved complexes,helped to explain the evolutionary divergences of life and the conservative rules of species,digged out some biologically meaningful knowledge.This paper integrated the multi omics biological data,constructed the multi-source dynamic protein network and the conservative protein network.Based on computational biology idea,the degree of interaction was predicted and the mining of dynamic protein complex and conservative protein complex was realized.Results showed that the influence of noise data could be overcome effectively,the topological limitations of network could be compensated and the potential biological information could be found.