南方医科大学学报 ›› 2020, Vol. 40 ›› Issue (12): 1831-1837.doi: 10.12122/j.issn.1673-4254.2020.12.21

• • 上一篇    下一篇

焦磷酸测序和 MassARRAY 定量检测 DNA 甲基化在年龄推断中的差异

王 玲,彭付端,赵 慧,李姗飞,孙晓萌,刘天资,丰 蕾   

  • 出版日期:2020-12-20 发布日期:2020-12-28

Quantitative analysis of DNA methylation by pyrosequencing and MassARRAY technique for age estimation: a comparative study

  • Online:2020-12-20 Published:2020-12-28

摘要: 目的 利用MassARRAY和焦磷酸测序两种方法检测的DNA甲基化数据在年龄推断中的差异,探讨两种检测方法的年龄推断计算方法。方法 应用MassARRAY和焦磷酸测序,分别对65份和62份外周血样本的9CpG位点的甲基化水平进行测定,运用多元线性回归模型预测年龄,对比预测年龄与实际年龄之间的差异;对比Z-score转化前后两种方法之间的年龄预测差异。结果 MassARRAY法,65样本集数据Z-score转化前,平均绝对误差(MAD=2.49岁,Z-score转化后,MAD=2.44岁;62样本集数据Z-score转化前,MAD=3.36岁,Z-score转化后,MAD=3.42岁。焦磷酸测序法,65样本集数据Z-score转化前,MAD=4.20岁,Z-score转化后,MAD=2.76岁;62样本集数据Z-score转化前,MAD=3.92岁,Z-score转化后,MAD=3.63岁。结论 Zscore转化方法能够有效的消除MassARRAY和焦磷酸测序数据之间的系统性批次效应;MassARRAY数据可以直接使用进行样本的年龄预测;使用焦磷酸测序数据进行年龄预测结果误差较大,但可以通过多样本积累进行Z-score转化之后预测年龄。

关键词: 年龄推断;DNA甲基化;MassARRAY;焦磷酸测序;逐步多元线性回归模型;Z-score

Abstract: Objective To study the difference in age estimation based on quantitative analysis of DNA methylation by MassARRAY and pyrosequencing techniques. Methods The methylation levels of 9 CpG sites from two independent whole blood sample sets (containing 65 and 62 samples) were detected using MassARRAY and pyrosequencing techniques. Z-score transformation was used to remove the batch effects of different techniques, and a linear regression model was used for age prediction. Results For age prediction using the MassARRAY system, the 65 samples showed a mean absolute difference (MAD) of 2.49 years before Z-score transformation of the data and 2.44 years after the transformation, similar to the results in the 62 samples (MAD of 3.36 years before and 3.42 years after Z-score transformation). For data typed from pyrosequencing, the 65 samples showed a MAD of 4.20 years before and 2.76 years after data Z-score transformation, also similar to the results in the 62 samples (MAD of 3.92 years before and 3.63 years after data transformation). Conclusion Z-score transformation can effectively reduce the system batch effect between MassARRAY and pyrosequencing. Data from the MassARRAY system allows direct age estimation without further data processing, while the pyrosequencing data may increase the error in age estimation, which can be corrected by Z-score transformation of the data.

Key words: age estimation; DNA methylation; MassARRAY detection; Pyrosequencing; stepwise multivariate linear regression model; Z-score