| Embryonic stem cells(ESC) have pluripotent which is the ability to develop to any type of cell. DNA methylation plays crucial roles in self-renewal and differentiation of embryonic stem cells. The systematic identification and characterization of methylation marks across cell types is crucial to understand the complex regulatory network for cell fate determination. As bisulfite sequencing enables profiling of developmental methylomes on an unprecedented scale and genome wide scanning of methylation markers in embryonic stem cells, however, experimental biologists face challenges in integrating and mining these data. Thus, accurate and efficient bioinformatics database and software based on integrated bisulfite sequencing data for experimental scientists are needed for identification of novel methylation markers in embryonic stem cells and analysis of their roles in complex regulatory networks of embryonic development. These novel markers are valuable for further understanding of pluripotency maintenance and directed differentiation in embryonic stem cells.In this thesis, we proposed an entropy-based algorithm for quantification of methylation specificity. Random data and real data were used to evaluate the performation of this algorithm in exactness, applicability to different sample numbers, and resource utilization. The algorithm for genome segmentation was developed based on distance-dependent methylation similarity between neighbouring Cp Gs. Further, the statistical method based on t test was proposed for identification of cell-type-specific methylation markers. Then we developed a specific methylation analysis and report tool SMART by Python. Comparison with other software suggetsted that SMART can be used to integrate whole genome bisulfite sequencing methylomes in large number of samples and de novo identify the cell-type-specific methylaton markers.To facilitate experimental scientists we integrated high-throughput DNA methylation data and developed mouse development methylation database Dev Mouse focusing on storing the mouse developmental methylomes centered with embryonic stem cells in temporal order. Meanwhile, we integrated human DNA methylomes and build the human methylation database Human Methy DB centered with embryonic stem cells. These two development methylation databases not only provide enough dataset for research in this thesis, but also enable experimental scientists to carry out bioinformatics analysis of DNA methylation in development.The entropy-based algorithm developed in this thesis was used to analyze the DNA methylome and epigenome data from mouse embryonic stem cells to brain which were obtained from our developed database Dev Mouse. We found DNA methylation variation was positively correlated with H3K27me3 variation. Further we identified 1 341 Cp G islands which were differentially methylated during development. It was revealed that these Cp G islands were significantly overlaped with those differentially modified by H3K27me3. Further, we identified 429 differentially expressed genes and confirmed the co-regulation of DNA methylation and other epigenetic modifications in these genes including core transcription factors and imprinted genes. Then we found intergenic Cp G islands may also be markers or regulatory elements for novel genes.Application of SMART to whole genome bisulfite sequencing methylomes in Human Methy DB de novo identified 757 887 function segments. Nearly 75% of all identified segments were uniformly methylated across all cell-types, indicating the stability of global DNA methylation in human genome. Further analysis revealed that the uniformly hypermethylated segments were prone to be located in repeat elements, while the uniformly hypomethylated segments were more likely Cp G islands in gene promoter. The analysis of segments with high methylation specificity revealed that embryonic stem cells specific hypomethylation are stable methylation markers and participate in regulation of key development genes. SMART was used to identify 3 758 DNA methylation markers of embryonic stem cells. We found various pluripotent stem cells shared more hypermethylation markers.Characterization of these markers based on high-throughput epigenomic data revealed some interesting findings. The hypomethylation markers in embryonic stem cells were significantly enriched in functions related to development. Some long methylation markers longer than 3 500 bp showed specific methylation in embryonic stem cells. By genome wide analysis, we found hypomethylation markers in embryonic stem cells were significantly enriched by two important marks for super-enhancer including H3K27 ac and transcription factor binding sites in cell-type-specific manner. Further colocalization analysis directly confirmed the significant overlap between hypomethylation markers and super-enhancers in embryonic stem cells. Futher we identified the hypomethylation markers overlapped by super-enhancers in embryonic stem cells and 71 related genes. Futher analysis revealed that the knockdowns of these genes may lead to abnormal development. Comparative analysis between human and mouse revealed the species conservation of the ESC methylation markers.Taken together, we developed methylation analysis tools and databases based entropy. The databases and software developed in this thesis would help experimental scientists for identification and function analysis of novel methylation markers. Based on these database and software, we identified novel DNA methylation markers of embryonic stem cells in mouse and human, and systematically analyzed their roles in co-regulation of developmental genes together with other epigenetic elements. These results provides novel important reference for further understanding of the mechanisms behind pluripotency maintenance and directed differentiation. |
PDF Full Text Request