Study On Synergistic Regulation Patterns And Functional Characteristics Of NcRNA In Malignant Tumor

Malignant tumors heavily threaten human life and health,and they are mainly divided into two categories(i.e.cancer and sarcoma)depending on the type of malignant tumor cells.Compared with normal cells,the structure and function of genes in malignant tumor cells are changed.Generally,the occurrence and development of malignant tumors do not conform to Mendelian inheritance.Moreover,malignant tumors have the characteristics of polygenic synergy,disease heterogeneity and individual specificity,etc.Essentially,most of malignant tumors are polygenic disorders,and their pathogenesis is caused by polygenic co-regulation.Therefore,for revealing the pathogenesis of malignant tumors,it has become an important breakthrough to investigate gene regulation patterns,biological processes,signaling pathways and pathological processes of malignant tumors from the perspective of gene molecular level.In the human genome,～98% of transcripts are non-coding RNAs(nc RNAs).They usually do not encode proteins,but act as regulators to modulate gene expression levels.The regulator nc RNA gives a new vision for exploring the occurrence and development of malignant tumors,and provides theoretical basis and new technology for the diagnosis and preventive treatment of human complex diseases including malignant tumors.However,the study of nc RNA involved in the regulatory mechanism of malignant tumors and its role as biomarkers for malignant tumors is still in the preliminary stage.Driven by heterogeneous data sources in malignant tumors,this dissertation focused on breast cancer,leukemia and pan-cancer as research objects,and carried out the research of new bioinformatics methods and technologies for identifying synergistic regulation patterns of nc RNA in malignant tumors,including both synergistic interaction and synergistic competition patterns.Meanwhile,an R package was developed to identify and analyse nc RNA synergistic competition patterns from heterogeneous data,including multiple gene expression profiles and mi RNA-target binding information.Furthermore,mi RSM was applied into malignant tumor data,and will be a useful tool for exploring nc RNA synergistic competition patterns in malignant tumors.The main research contents and contributions are as follows:(1)At the bulk level,mi Rsyn(mi RNA synergism)was proposed to identify mi RNA synergism in breast cancer.By integrating bulk transcriptome data and mi RNA-target binding information,mi Rsyn estimated causal effects of mi RNAs on m RNAs using the causal inference method,and further infered mi RNA synergism.At the synergistic interaction network layer,the identified mi RNA synergistic interaction network was not a scale-free but a small-world network,and was closely associated with 19 biological processes,pathways and diseases related to breast cancer.In addition,71.08% of mi RNA-mi RNA synergistic interaction pairs showed similar expression patterns,and 46.53% of mi RNA-mi RNA synergistic interaction pairs at the sequence level were not working synergistically at the expression level.At the synergistic interaction module layer,out of the identified 361 mi RNA synergistic interaction modules,72 modules were significantly associated with breast cancer.The comparison results indicated that mi Rsyn(mimicking multiple gene knockouts)was more suitable than the other method(e.g.mir SRN by mimicking single gene knockout)in studying mi RNA synergism.(2)At the single-cell level,CSmi R(Cell-Specific mi RNA regulation)was presented to identify cell-specific mi RNA synergism in leukemia.CSmi R combined single-cell transcriptome data and mi RNA-target binding information to identify cell-specific mi RNA synergism.When applied to small-scale single-cell transcriptome data,CSmi R had investigated the mi RNA synergism of 19 K562 cells.In terms of mi RNA-mi RNA synergistic interaction pairs,hub mi RNAs and mi RNA synergistic interaction modules,the similarity between any pair of K562 cells was less than 90%.Moreover,the percentage of the rewired mi RNA-mi RNA synergistic interaction pairs(38.91%),hub mi RNAs(21.88%)and mi RNA synergistic interaction modules(64.35%)was higher than that of the conserved mi RNA-mi RNA synergistic interaction pairs(14.41%),hub mi RNAs(14.58%)and mi RNA synergistic interaction modules(0%)respectively.By constructing cell-cell similarity matrix,CSmi R offered a new strategy for clustering single-cells and provided a new insight for understanding cell-cell crosstalk.Network topology and functional analysis had revealed that the identified 19cell-specific mi RNA synergistic interaction networks were not scale-free but small-world networks,and the cell-specific mi RNA synergistic interaction networks and modules may be closely related to the occurrence and development of leukemia.(3)Scomp(Synergistic competition)was proposed to identify nc RNA synergistic competition network in pan-cancer.Based on the nc RNA synergistic competition hypothesis,Scomp infered nc RNA synergistic competition network by integrating multiple gene expression data and putative competing endogenous RNA network.Network topology and functional analysis displayed that the identified nc RNA synergistic competition network was scale-free and small-world network,and was significantly enriched in many malignant tumors,cancer hallmarks and cancer phenotypes.Furthermore,the identified 13 hub nc RNAs in the nc RNA synergistic competition network were closely associated with malignant tumors,and the performance of them was higher than baseline’s performance in classifying 32 malignant tumor types.Functional analysis of network modules had revealed that all of3 network modules in the nc RNA synergistic competition network were functional modules,and 2 network modules were potential pan-cancer biomarkers.Multi-label classification analysis further indicated that all of network modules performed better than baseline in classifying 32 malignant tumor types.Finally,the consistent results showed that Scomp was robust in identifying nc RNA synergistic competition network.(4)A novel method LMSM(Lnc RNA-related Mi RNA Sponge Modules)was proposed to identify lnc RNA synergistic competition modules in breast cancer.Based on the mi RNA sponge modular competition hypothesis,LMSM infered lnc RNA synergistic competition modules by combining heterogeneous data sources.As a result,90.16% of the mediating mi RNAs acted as crosslinks across 17 different lnc RNA synergistic competition modules.All of lnc RNA synergistic competition modules were statistically significant,and 94.12% of the lnc RNA synergistic competition modules were functionally associated with breast cancer.In addition,9 lnc RNA synergistic competition modules were breast cancer subtype-specific,and the performance of lnc RNA synergistic competition modules was significantly higher than baseline’s performance in classifying breast cancer subtypes(the significance p-value was less than 2.20E-16).Moreover,the comparison result showed that LMSM outperformed a graph clustering-based strategy in identifying lnc RNA synergistic competition modules.Finally,the consistent results suggested that LMSM was robust in identifying lnc RNA synergistic competition modules.(5)By combining multiple gene expression data and mi RNA-target binding information,an R package mi RSM(mi RNA Sponge Module)was developed to identify and analyse nc RNA synergistic competition modules.In mi RSM,there are 50co-expression analysis methods to identify gene co-expression modules,4 module identification methods to predict nc RNA synergistic competition modules,and 7modular analysis methods to analyse nc RNA synergistic competition modules.In the two case studies(breast cancer and pan-cancer),mi RSM was generally superior to network clustering or graph clustering methods,and contributed to exploring the synergistic competition patterns of nc RNAs in malignant tumors.In summary,from the perspective of computational biology,mi RSM can simplify the research process of nc RNA synergistic competition patterns.Altogether,driven by multi-source omics data,this dissertation has conducted the research around mi RNA synergistic interaction patterns,nc RNA synergistic competition patterns and their functional characteristic,and has provided novel methods and tools for investigating nc RNA synergistic regulation patterns in malignant tumors.The research results lay a foundation for establishing the relationship between nc RNA and its synergistic regulation patterns and malignant tumors,and are of great significance to explore the pathogenesis of malignant tumors,malignant tumor subtype classification,and personalized diagnosis and treatment.