年龄预测可以提供有关犯罪现场生物证据贡献者的重要信息。DNA甲基化被认为是最有前途的年龄预测生物标志物。在全基因组范围内测量甲基化水平是筛选用于法医年龄预测的特定标记的重要步骤。在目前的研究中,我们从公开的EPICBeadChip数据中筛选出5个与年龄相关的CpG位点,并通过多重甲基化SNaPshot测定法在训练集(115血液)中进行评估.通过完整子集回归,五个标记缩小到三个,即cg10501210(C1orf132),cg16867657(ELOVL2),和cg13108341(DNAH9),其中最后一个是新发现的与年龄相关的CpG位点。基于这三个指标建立了年龄预测模型,解释86.8%的年龄变化,平均绝对偏差(MAD)为4.038岁。然后,根据年龄预测模型调整多重甲基化SNaPshot测定。考虑到血迹是实际案例中最常见的生物样本之一,由30个血液组成的三个验证集,使用30种新鲜血迹和30种老化血迹来评估年龄预测模型。每套的MAD估计为4.734、4.490和5.431年,分别,表明我们的年龄预测模型适用于11-71岁中国汉族人群的血液和血迹的年龄预测。总的来说,这项研究描述了从公共芯片数据中筛选CpG标记的工作流程,并提出了用于法医年龄预测的3-CpG标记模型.
Age prediction can provide important information about the contributors of biological evidence left at crime scenes. DNA methylation has been regarded as the most promising age-predictive biomarker. Measuring themethylation level at the genome-wide scaleis an important step to screen specific markers for forensic age prediction. In present study, we screened out five age-related CpG sites from the public EPIC BeadChip data and evaluated them in a training set (115 blood) by multiplex methylation SNaPshot assay. Through full subset regression, the five markers were narrowed down to three, namely cg10501210 (C1orf132), cg16867657 (ELOVL2), and cg13108341 (DNAH9), of which the last one was a newly discovered age-related CpG site. An age prediction model was built based on these three markers, explaining 86.8% of the variation of age with a mean absolute deviation (MAD) of 4.038 years. Then, the multiplex methylation SNaPshot assay was adjusted according to the age prediction model. Considering that bloodstains are one of the most common biological samples in practical cases, three validation sets composed of 30 blood, 30 fresh bloodstains and 30 aged bloodstains were used for evaluation of the age prediction model. The MAD of each set was estimated as 4.734, 4.490, and 5.431 years, respectively, suggesting that our age prediction model was applicable for age prediction for blood and bloodstains in Chinese Han population of 11-71 age. In general, this study describes a workflow of screening CpG markers from public chip data and presents a 3-CpG markers model for forensic age prediction.