Omnigenic Pattern Identification Method And Its Application In Cancer

With the development of genome-wide association studies and high-throughput sequencing technology,the big data in biological networks grows rapidly,and the disease data is increasingly abundant,in which to identify complex diseases patterns have become a research hotspot in systems biology.The polygenic pattern believes that disease genes are clustered in the disease pathways or modules,but the multi-omics,heterogeneity,and multiscale characteristics of the disease data will affect the effect of the polygenic pattern in biomedical applications.The latest research results focus on the omnigenic pattern of disease.It is believed that disease genes are distributed in the entire genome and can be divided into directly acting core genes and indirect acting peripheral genes.Peripheral genes affect diseases by regulating core genes.The omnigenic pattern depicts the complete network neighborhood of the disease,and models the core and peripheral structure patterns.Based on the big data of biological network,the algorithm research of omnigenic pattern recognition has become the key to analyzing complex diseases.Cancer is a typical complex disease.Its pathogenesis involves multi-omics features such as gene expression,DNA methylation,somatic mutation,copy number variation,etc.Based on the omnigenic pattern to study the structured patterns of cancer multi-omics data,it is an important means to fully understand the molecular mechanism of cancer and is also a major challenge that biomedicine poses to pattern recognition and computational science.Based on the big data of biological network,this thesis studies the omnigenic pattern recognition method of complex diseases.Firstly,inspired by cancer multi-omics data,the core and peripheral structural patterns of the omnigenic pattern were defined;then,combined with the knowledge of network science,graph theory and statistics,the omnigenic pattern recognition method was developed.The algorithm framework extracts connectivity significance pattern of nodes in a large-scale network,a improved dynamic time warping algorithm is used to map the connectivity significance patterns,further,multi-omics core nodes and peripheral nodes are identified,and the omnigenic pattern is constructed through the correlation analysis of core nodes and peripheral nodes;finally,based on human interaction network data,the omnigenic pattern recognition method is applied to the molecular mechanism analysis of cancer and the prediction of cancer relationship.This thesis uses statistical analysis methods to verify the biological significance of cancer omnigenic pattern.Firstly,functional enrichment analysis and biological data sets verification are performed on the identified cancer omnigenic pattern.The results show that the core genes and peripheral genes in the cancer omnigenic pattern are significantly enriched in known cancer pathways;then based on the gene regulation relationship described by the expression quantitative trait loci data,it verifies that the peripheral genes in the omnigenic pattern have a significant regulatory effect on the core genes;finally,the omnigenic pattern is applied to the cancer relationship quantification,and it is found that the addition of peripheral genes improves the accuracy of cancer similarity prediction.In summary,the definition and identification method of the omnigenic pattern provides an effective calculation tool for systematic analysis of disease mechanisms.It can be applied to biological network big data and cancer multi-omics data to identify the complete network neighborhood of cancer,which can provide a key basis for support the precise diagnosis and treatment of cancer.