Integration Of Protein Interaction Networks And Multi-level Research On Functional Modules In Neurodegeneration

Protein-protein interactions exist in the life processes of every cell.An in-depth understanding of their regulation and dynamic change patterns can help to reveal the essence of life phenomena.The combination of microscopic concrete physicochemical mechanisms and macroscopic abstract network theory can elucidate the structural features and functions of protein modules at multiple levels.Neurodegenerative diseases,which are usually associated with aggregation and molecular interactions of pathological proteins,are good objects of interaction network-related studies.Integrating data from biomedical research fulfill systematic mining and analysis of pathogenic proteins and their interaction networks.The research contents are as follows.1.Integration of neurodegeneration-related interaction networks: curation of protein interactions from multiple public data sources obtained specified disease,disease-related proteins and de-redundant human protein interaction networks.In this disease-associated interactions network,protein isoforms are included to find the effect of alternative splicing products on protein binding status and disease onset,and to mine possible regulatory modules.A database named NDAtlas(http://bis.zju.edu.cn/ndatlas)has been constructed to provide quick querying of proteins,disease-and pathway-related interactions networks,and visualization of docking structures.2.Prediction of potential pathogenic proteins in neurodegeneration: Based on the topological features of protein interactions networks,random forest algorithm was implemented in predicting proteins that may cause neurodegenerative diseases.The performance of this machine learning model was further demonstrated by integrating multi-omics data and accurately identifying some hub proteins that play important roles in the protein interaction network.3.Multi-protein docking visualization: the interactions of important disease-causing proteins were studied.Their structures and those of the interacting proteins were obtained.The structural affinity of the interacting proteins was calculated with a protein docking program,and a highresolution network with structural information was constructed.An intuitive and interactive view of three-dimensional molecular graphics with protein docking was provided in addition to the traditional two-dimensional network of binary interactions.A database extension m PPI(http://bis.zju.edu.cn/mppi)was developed for PPI docking structure visualization.In summary,this study expands the protein interaction network of neurodegenerative diseases,constructs a machine learning model to identify potential pathogenic proteins;and combines the multi-protein docking structure visualization tool for the first time to increase the atomic details of structural docking for protein interactions.It is helpful for the systematic study of the crossdisease mechanisms of neurodegenerative diseases,the search for the common pathways of multiple diseases across the network,and the exploration of potential drug targets,so as to provide guidance for the diagnosis and treatment of diseases.