The Research Of The Interaction Between Epigenetic Modifications

The study of epigenetic feature is a hot spot of the field of epigenetics. In whichthe feature extracting is one import step for predicting the methylation status of DNAsequence, however every proposed method was just devised on some features and thecalculation of quantitative feature is complicated. Meanwhile, due to the specificity inDNA sequence pattern calculating, the common pattern matching algorithms are notideal. Moreover, in previous studies, different features have been used for predictionand analysis of DNA methylation and there is no unified comparison and evaluationof feature to date. And few studies of nonCpG island sequence have been performed,the import feature of DNA methylation is not indeterminacy. Furthermore, therelationship between histone modifications, DNA methylation and gene expression,and the mechanism underlying how they regulate gene expression remains largelyunknown.In this paper, in view of the problems appears in epigenetics mentioned above, weperform a research on the epigenetic feature correlated algorithm&tool and therelationship between epigenetic feature and gene expression. The main researchcontents and innovations are as follows:We collected the features used in published papers, then designed andimplemented the software of extracting these features for DNA sequence. Thissoftware can easily process a batch of target sequences and facilitate the subsequentdata analysis and mining.We analyzed the specificity of DNA feature calculation, proposed an algorithmbased on “space for time” idea, which designed a scheme of using map data structureto store the intermediate calculation results, so that each DNA sequence needs to bescanned just once to calculate all pattern features. Experiment results andanalysis show that, the algorithm can effectively improve the efficiency of calculatingDNA feature, and better solve the problems common pattern calculating algorithmshave in calculating DNA feature.We collected the primary features of published paper to perform a unified andextensive comparison, evaluated the feature importance for the sequence of CpGisland and nonCpG island and utilized feature selection to identify the compact set of import feature. This analysis demonstrated that sequence pattern and histonemodification are important feature for DNA methylation but not independent and bothare responsible for determing and (or) maintaining the methylation pattern of genomicDNA. In detail, H3K4me3, which has the maximum selection frequency, is the importhistone feature for CpG island and nonCpG island sequence. The identified keyfeatures of DNA methylation in this analysis have important biological function andmay be landmarks for the mysteries among DNA methylation, histone methylationand gene regulation.By combining partial correlation coefficient with Pearson correlation coefficient,an epigenetic interaction network(EIN) of histone modification, DNA methylation andgene expression is constructed, and we mine the key modules in the EIN. Though theanalysis of the mining results, we deduce some mechanism underlying how theepigenetic modifications regulate gene expression. In detail, H3K27me3, H3K27me2,DNA methylation, H3K9me2and H3K9me3mainly play an repressive role in geneexpression regulation, but have a weak relationship with gene activation. Theapproach would be very helpful in uncovering inherent relationship betweenepigenetic modifications and gene expression regulation.