Pattern Discovery Of Competing Endogenous RNA Regulation With Applications

Bioinformatics and computational biology aim to comprehensively utilize the theories and methods of biology,mathematics and informatics to study the collection,storage,analysis and interpretation of large-scale biological data.They are powerful tools to unlock the mysteries of lives.With the widespread application of next-generation high-throughput sequencing technologies,a large number of non-coding RNAs have been discovered.How to explain the functions of these non-coding RNAs has become one of the current research hotspots.Recent studies have shown that these non-coding RNAs can affect m RNA expression levels by competitively binding micro RNAs.This post-transcriptional regulation mechanism is called competing endogenous RNA(ceRNA)regulation.A large number of studies have shown that ceRNA regulation plays an important role in a variety of physiological and pathological processes.Therefore,mining the ceRNA regulation pattern of organisms has positive significance for analyzing the process of gene regulation in diseases.This dissertation focuses on the mining of ceRNA regulation patterns and carries out relevant research work from two aspects: the identification of ceRNA regulation pattern and the application of ceRNA regulation pattern.Specifically,the main research contents and contributions of this dissertation are as follows:This dissertation builds the most complete long non-coding RNAs(lnc RNA)subcellular localization database,lnc SLdb.The ceRNA regulation is affected by the RNA subcellular localization patterns,and lnc RNAs are important players of ceRNA regulation,whose subcellular localization patterns are still unclear.Therefore,this dissertation collects the qualitative and quantitative experimental data of lnc RNA subcellular localization from previous researches by literature mining.The current version of lnc SLdb contains subcellular location information of more than 11,000 lnc RNA transcripts from 9494 genes and 3 major species(human,mouse and fruit fly).According to the results of biological experiments,lnc SLdb divides the subcellular locations of lnc RNA into three basic types(nucleus,cytoplasm and nucleus/cytoplasm)and three subtypes(ribosome,chromatin and nucleolus).This dissertation also uses the data in lnc SLdb to analyze and discuss the sequence features that may affect the subcellular localization pattern of lnc RNAs.The analysis finds that k-mer features have a partial predictive ability for lnc RNA subcellular location,and some k-mer features may be one of the signals of lnc RNA nucleus retention.Sequence motifs are widely found in lnc RNAs accumulated in the cytoplasm,and it is likely that sequence motifs are important signals for lnc RNAs to transport out of nucleus.This dissertation proposes a method,LAce Module,for identifying ceRNA regulation modules,which integrates the traditional correlation and the dynamic correlation measurement by multi-view non-negative matrix factorization.In response to the problem of false positives in the identification of ceRNA regulation relationships,liquid association(LA)is introduced as a dynamic correlation measurement for identifying ceRNA regulation relationships.Experimental results show that LA can effectively predict the ceRNA regulation relationship,and LAce Module can effectively identify ceRNA regulation modules compared with traditional methods.Further analysis of the ceRNA regulation modules of breast tumor patients finds that ceRNA regulation plays a very important role in breast tumors,especially in cell adhesion,cell transfer,and cell communication.In addition,the analysis finds that ceRNA genes and associated micro RNAs may be potential drug targets and biomarkers for tumor treatment and prognosis.This dissertation analyzes the changes in the ceRNA regulation networks of various cell types in the peripheral blood mononuclear cells(PBMCs)of COVID-19 patients.This dissertation analyzes four aspects of the ceRNA regulation networks,including network topological features,topological properties of COVID-19 related nodes,differential and conserved network modules and differentially expressed sub-networks.The results find that the overall topological features of the ceRNA regulation networks in each cell type do not change much after the virus infection,but the hub nodes of virus infected network for each cell type change significantly.A large number of antiviral-related genes become the hub nodes in the virus infected ceRNA regulation networks,especially the four genes related to measles virus infection,which are found in natural killer cells and placed at the center of the network.The topological property analysis of the nodes associated with COVID-19 finds that the topology properties of the nodes associated with cytokine genes in the PBMCs do not change significantly.On the contrary,the topological properties of nodes associated with differentially expressed genes in various cell types change significantly,especially the lnc RNA MALAT1 in ?/? T cells.Further analysis reveals that the specific neighbor nodes of MALAT1 in the infected network are closely related to a variety of pathogen infection and immune signaling pathways.The analysis of the differential network modules finds that the COVID-19-specific networks form modules enriched with antiviral genes.However,the analysis of the conserved modules finds that the functions of the conserved modules of innate immune cells do not change substantially,but the adaptive immune cells produce antiviral immune responses after the virus infection.Finally,the differentially expressed sub-network analysis shows that the differentially expressed genes in cells may be related to the differential expression of micro RNAs.People can use existing drugs to control and influence the states of PBMCs of COVID-19 patients.In view of the lack of comprehensive,user-friendly,convenient and unified software platform in the current ceRNA regulation research field,this dissertation designs and implements a ceRNA regulation network construction and analysis platform based on the Shiny framework,Ce Net Omnibus.Ce Net Omnibus is divided into 5 different components,which accomplish the functions of data uploading,data preprocessing,network construction,network visualization and network analysis,respectively.Users can group ceRNA genes,filter samples and genes of ceRNAs and micro RNAs,construct ceRNA regulation networks based on single or multiple measurements,analyze the topological properties of ceRNA regulation networks,identify ceRNA regulation modules,and analyze the biological and medical significance of ceRNA regulation networks or sub-networks.