Computational and Experimental Modeling of Epigenetic DNA Methylation

描述（由申请人提供）：DNA甲基化是一种影响基因组组织和功能的表观遗传修饰，在正常发育和疾病中起着关键作用。基于亚硫酸氢盐的将未甲基化的Cs转化为Ts，然后进行深度测序（BS-seq）已经成为以单核苷酸分辨率研究全基因组DNA甲基化的金标准。虽然下一代测序（NGS）的进展允许越来越负担得起的全基因组BS-seq（WGBS），但对由此产生的大量数据的解释需要高效的生物信息学方法。在这个提议中，我们将开发一系列新的生物信息学方法用于BS-seq数据分析。首先，在我们的BSMAP项目早期成功的基础上，我们将开发下一代亚硫酸氢盐校准器。我们将构建一个亚硫酸氢盐和SNP“感知”基因组索引，用于使用IUPAC代码和动态Burrows-Wheeler变换（DBWT）进行读取映射。我们还将区分CpG甲基化和C/T SNP，并使用GPU硬件加速来提高映射速度。其次，我们将开发一种强大的差异甲基化分析算法，可以兼顾两者 来自测序的取样变异和重复之间的生物学变异。我们还将介绍一种新的指标，用于评估差异甲基化的统计学和生物学意义。该模型将具有足够的能力来检测低CpG密度调控区（如增强子）中的单CpG分辨率差异甲基化，其测序深度低至5-10倍。第三，我们将使用Galaxy Web界面和云计算开发全面的BS-seq数据分析管道。我们将根据表观遗传学社区的新需求，持续整合我们正在开发的所有BS-seq工具和其他公共算法。这条管道将使实验生物学家能够自己进行大部分分析。这些生物信息学方法将由我们的合作者进行广泛的测试和实验验证。虽然在本提案中使用常规BS-seq专注于CpG甲基化，但我们的生物信息学方法可以立即用于其他修改的BS-seq方案，例如最近分别针对5 mC和5 hmC开发的oxBS-Seq和TAB-Seq。最后，作为一个案例研究，我们将应用这些新的方法来解开在体内的作用，DNA甲基化在造血系统恶性肿瘤。这些实验和后续验证也将使我们能够提高我们的生物信息学方法的有效性。

DESCRIPTION (provided by applicant): DNA methylation, an epigenetic modification affecting the organization and function of the genome, plays a critical role in both normal development and disease. Bisulfite based conversion of unmethylated Cs to Ts followed by deep sequencing (BS-seq) has emerged as the gold standard to study the genome-wide DNA methylation at single-nucleotide resolution. While progress in next-generation sequencing (NGS) allows increasingly affordable whole-genome BS-seq (WGBS), interpretation of the resulting massive amount of data requires efficient bioinformatics methods. In this proposal, we will develop a series of novel bioinformatics methods for BS-seq data analysis. First, building on the early success of our BSMAP program, we will develop the next generation of bisulfite aligner. We will construct a bisulfite- and SNP-"aware" genome indexing for read mapping with IUPAC code and dynamic Burrows-Wheeler transformation (DBWT). We will also distinguishing CpG methylation from C/T SNP and use GPU hardware acceleration to improve the mapping speed. Second, we will develop a powerful differential methylation analysis algorithm that can take into account both sampling variation from sequencing and biological variation between replicates. We will also introduce a novel metric for evaluating both the statistical and biological significance of differential methylation. This model will have enough power to detect single-CpG resolution differential methylation in low-CpG-density regulatory regions, such as enhancers, with as low as 5-10 fold sequencing depth. Third, we will develop a comprehensive BS-seq data analysis pipeline using the Galaxy web interface and cloud computing. We will integrate all the BS-seq tools we are developing and other public algorithms on a continuous basis according to the emerging needs of the epigenetic community. This pipeline will empower experimental biologists to perform most analyses on their own. These bioinformatics methods will undergo extensive testing and experimental validation by our collaborators. Although focused on CpG methylation using conventional BS-seq in this proposal, our bioinformatics methods can be immediately used in other modified BS-seq protocols, such as oxBS-Seq and TAB-Seq recently developed for 5mC and 5hmC, respectively. Finally, as a case study, we will apply these new methods to unravel the in vivo role of DNA methylation in hematopoietic malignancies. These experiments and follow-up validations will also enable us to improve the efficacy of our bioinformatics methods.