基于甲基化的年龄推断模型构建与效能评估

Methylation-Based Age Estimation Model Construction and Its Effectiveness Evaluation

目的筛选与中国北方汉族男性个体年龄相关的DNA甲基化位点并构建年龄推断模型。方法筛选21个候选甲基化位点,使用EpiTYPER技术平台在476份汉族男性血液DNA中检测包含上述位点在内的21个扩增片段,共获得153个DNA甲基化位点数据。结果通过相关性分析,最终筛选出8个与年龄高度相关的DNA甲基化位点,其中CpG1、CpG2、CpG4、CpG7和CpG8分别位于TRIM59、RASSF5、Clorf132、CSNK1D和ELOVL2上,CpG5、CpG6均位于PDE4C上,CpG3位于17号染色体21452808位置上。其中352份样本用于模型构建,基于这8个位点构建的多元线性回归年龄推断模型,R^2=0.93,平均绝对偏差(mean absolute deviation,MAD)为2.69岁;109份样本用于模型验证,MAD为3.80岁。在15份新的样本中重复检测三次,MAD分别为4.08岁、4.68岁和3.93岁。结论本研究构建的中国北方汉族男性个体年龄的推断模型,可用于公安实战中受害者和嫌疑人的年龄推断,缩小案件的侦查范围,具有实际应用价值。

Objective To screen the DNA methylation loci associated with the age of Han males in northern China and to construct an age estimation model. Methods Twenty-one candidate methylation loci were screened. The DNA methylation levels of 476 blood samples from Chinese Han males were detected for 21 amplicons using EpiTYPER technology platform, and data on 153 DNA methylation loci were obtained. Results After correlation analysis, 8 age-related DNA methylation loci were finally screened. CpG1, CpG2, CpG4, CpG7, CpG8 were located on TRIM59, RASSF5, Clorf132, CSNK1D, ELOVL2, CpG5, CpG6 on PDE4C, and CpG3 on chr17:21452808. Based on the 8 loci, 352 samples were used for model construction. A multivariate linear regression age estimation model was constructed (R2=0.93), with mean absolute deviation (MAD) of 2.69 years old. When 109 samples were used for model validation, the MAD was 3.80 years old. The test was repeated 3 times in 15 new samples, with MADs of 4.08, 4.68 and 3.93 years old, respectively. Conclusion The age estimation model on Han males in northern China constructed in this study can be used to estimate the age of victims and suspects and to narrow the scope of investigation, and therefore has practical application value.