Epigenome-wide association studies (EWAS) have been widely applied to identify methylation CpG sites associated with human disease. To date, the Infinium MethylationEPIC array (EPIC) is commonly used for high-throughput DNA methylation profiling. However, the EPIC array covers only 30% of the human methylome. Methylation Capture bisulfite sequencing (MC-seq) captures target regions of methylome and has advantages of extensive coverage in the methylome at an affordable price. Epigenome-wide DNA methylation in four peripheral blood mononuclear cell samples was profiled by using SureSelectXT Methyl-Seq for MC-seq and EPIC platforms separately. CpG site-based reproducibility of MC-seq was assessed with DNA sample inputs ranging in quantity of high (> 1000 ng), medium (300–1000 ng), and low (150 ng–300 ng). To compare the performance of MC-seq and the EPIC arrays, we conducted a Pearson correlation and methylation value difference at each CpG site that was detected by both MC-seq and EPIC. We compared the percentage and counts in each CpG island and gene annotation between MC-seq and the EPIC array. After quality control, an average of 3,708,550 CpG sites per sample were detected by MC-seq with DNA quantity > 1000 ng. Reproducibility of DNA methylation in MC-seq-detected CpG sites was high among samples with high, medium, and low DNA inputs (r > 0.96). The EPIC array captured an average of 846,464 CpG sites per sample. Compared with the EPIC array, MC-seq detected more CpGs in coding regions and CpG islands. Among the 472,540 CpG sites captured by both platforms, methylation of a majority of CpG sites was highly correlated in the same sample (r: 0.98–0.99). However, methylation for a small proportion of CpGs (N = 235) differed significantly between the two platforms, with differences in beta values of greater than 0.5. Our results show that MC-seq is an efficient and reliable platform for methylome profiling with a broader coverage of the methylome than the array-based platform. Although methylation measurements in majority of CpGs are highly correlated, a number of CpG sites show large discrepancy between the two platforms, which warrants further investigation and needs cautious interpretation.

中文翻译：

---

外周血单个核细胞甲基化捕获测序与Infinium MethylationEPIC阵列的比较

表观基因组关联研究 (EWAS) 已被广泛应用于鉴定与人类疾病相关的甲基化 CpG 位点。迄今为止，Infinium MethylationEPIC 阵列 (EPIC) 常用于高通量 DNA 甲基化分析。然而，EPIC 阵列仅覆盖了人类甲基化组的 30%。甲基化捕获亚硫酸氢盐测序 (MC-seq) 可捕获甲基化组的目标区域，并具有以可承受的价格在甲基化组中广泛覆盖的优势。分别使用用于 MC-seq 和 EPIC 平台的 SureSelectXT Methyl-Seq 对四个外周血单核细胞样本中的表观基因组范围的 DNA 甲基化进行了分析。MC-seq 基于 CpG 位点的可重复性是通过 DNA 样本输入量来评估的，其数量范围为高 (> 1000 ng)、中 (300–1000 ng) 和低 (150 ng–300 ng)。为了比较 MC-seq 和 EPIC 阵列的性能，我们对 MC-seq 和 EPIC 检测到的每个 CpG 位点进行了 Pearson 相关性和甲基化值差异。我们比较了 MC-seq 和 EPIC 阵列之间每个 CpG 岛和基因注释的百分比和计数。质量控制后，通过 MC-seq 检测到每个样本平均 3,708,550 个 CpG 位点，DNA 量 > 1000 ng。在具有高、中和低 DNA 输入的样本中，MC-seq 检测到的 CpG 位点中 DNA 甲基化的重现性很高（r > 0.96）。EPIC 阵列每个样本平均捕获 846,464 个 CpG 位点。与 EPIC 阵列相比，MC-seq 在编码区和 CpG 岛中检测到更多的 CpG。在两个平台捕获的 472,540 个 CpG 位点中，大多数 CpG 位点的甲基化在同一样本中高度相关（r：0.98-0.99）。然而，两个平台之间一小部分 CpG（N = 235）的甲基化差异显着，β 值的差异大于 0.5。我们的结果表明，与基于阵列的平台相比，MC-seq 是一个有效且可靠的甲基化组分析平台，对甲基化组的覆盖范围更广。尽管大多数 CpG 中的甲基化测量高度相关，但许多 CpG 位点显示两个平台之间存在很大差异，这需要进一步调查并需要谨慎解释。我们的结果表明，与基于阵列的平台相比，MC-seq 是一个有效且可靠的甲基化组分析平台，对甲基化组的覆盖范围更广。尽管大多数 CpG 中的甲基化测量高度相关，但许多 CpG 位点显示两个平台之间存在很大差异，这需要进一步调查并需要谨慎解释。我们的结果表明，与基于阵列的平台相比，MC-seq 是一个有效且可靠的甲基化组分析平台，对甲基化组的覆盖范围更广。尽管大多数 CpG 中的甲基化测量高度相关，但许多 CpG 位点显示两个平台之间存在很大差异，这需要进一步调查并需要谨慎解释。