R语言包InfiniumPurify在肿瘤纯度估计和差异甲基化分析中的应用

发布时间：2018-05-18 01:02

本文选题：DNA甲基化 + 表观遗传学　；参考：《上海师范大学》2017年硕士论文

【摘要】：DNA甲基化与人类发育以及肿瘤疾病密切相关,对肿瘤细胞纯度的估计以及差异甲基化分析都是表观遗传学研究的重要内容。但是,基于DNA甲基化芯片数据来研究肿瘤细胞纯度以及差异甲基化分析的方法还不完善。由于肿瘤样本中含有正常细胞,而肿瘤纯度会给混合肿瘤-正常样本的差异甲基化分析带来偏差甚至是产生错误预测。现有的方法还没有完全实现对肿瘤样本的“矫正”,估计纯肿瘤样本甲基化水平的方法有待研究。本文通过InfiniumPurify包来研究上述问题。InfiniumPurify包含如下三个模型:第一,估计肿瘤细胞纯度的getPurity函数。getPurity首先基于混合肿瘤-正常样本的?值矩阵得到差异最显著的CpG位点(记为iDMCs),再通过iDMCs属于超甲基化位点还是低甲基化位点来转换iDMCs的甲基化水平,最后对这些iDMC利用核密度估计得到肿瘤样本的纯度。预测的结果与ABSOLUTE以及其他方法的结果高度一致;第二,考虑肿瘤纯度进行差异甲基化分析的InfiniumDMC函数。由于Infinium DMC考虑了肿瘤样本的纯度,避免了因为肿瘤样本不纯而导致的差异甲基化分析中误差的出现,与其它现有的方法相比得到的差异甲基化位点更准确;在没有正常样本控制时InfiniumDMC也可以进行差异甲基化分析,大大扩展了对TCGA数据的分析与应用;第三,InfiniumPurify函数,其基于肿瘤、正常样本以及肿瘤纯度值通过线性回归模型来估计纯的肿瘤样本的甲基化水平,经过纯度的矫正,使得差异甲基化位点处肿瘤样本与正常样本的甲基化水平的分布有了非常显著的差异。
[Abstract]:DNA methylation is closely related to human development and tumor disease. Estimation of tumor cell purity and differential methylation analysis are important in epigenetics. However, it is not perfect to study tumor cell purity and differential methylation analysis based on DNA methylation chip data. Due to the presence of normal cells in the tumor samples, the purity of the tumor can cause deviation or even false prediction for differential methylation analysis of mixed tumor-normal samples. The existing methods have not completely achieved the "correction" of tumor samples, and methods to estimate the methylation level of pure tumor samples need to be studied. In this paper, we use InfiniumPurify package to study the above problem. Infinium purify contains the following three models: first, the getPurity function of estimating tumor cell purity. GetPurity is based on mixed tumor-normal sample? The value matrix obtained the most significant difference of CpG sites (denoted as iDMCsN, then converted the methylation level of iDMCs by iDMCs belonging to hypermethylation site or low methylation site). Finally, the purity of tumor samples was estimated by nuclear density estimation for these iDMC. The predicted results are highly consistent with those of ABSOLUTE and other methods. Secondly, the InfiniumDMC function for differential methylation analysis of tumor purity is considered. Because Infinium DMC takes into account the purity of tumor samples and avoids errors in differential methylation analysis caused by the impurity of tumor samples, the differential methylation sites obtained by Infinium DMC are more accurate than those obtained by other existing methods. InfiniumDMC can also perform differential methylation analysis without normal sample control, greatly expanding the analysis and application of TCGA data. Normal samples and tumor purity values were estimated by linear regression model to estimate the methylation level of pure tumor samples and corrected by purity. The distribution of methylation level in tumor samples at differential methylation sites is significantly different from that in normal samples.
【学位授予单位】：上海师范大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：R73-3;O212.1

【相似文献】