The evaluation of seven age-related CpGs for forensic purpose in blood from Chinese Han population

Abstract

Age prediction of biological samples is one of the important tasks in forensic DNA phenotyping, and DNA methylation is regarded as the most promising biomarker for forensic age prediction. To date, numerous CpG sites have been reported to be age-related based on the changes in methylation. In this study, seven age-related CpG (AR-CpG) sites, cg02228185 (ASPA), cg09809672 (EDARADD), cg19283806 (CCDC102B), cg04208403 (ZNF423), chr17: 44,390,358 of GRCh38/hg38 (ITGA2B), cg14361627 (KLF14), and cg06639320 (FHL2), were selected and analyzed in 310 blood samples using a multiplex methylation SNaPshot assay to evaluate the value of selected AR-CpGs in age prediction in blood from Chinese Han population. The study confirmed the correlation of all the investigated markers with human age, and the correlation of cg19283806 with age is the highest while cg04208403 is the lowest in the Chinese Han population. Two different age prediction models, stepwise regression and support vector regression (SVR), were established based on the methylation SNaPshot data using 230 blood samples from donors aged 2–86 years old. The stepwise regression model included six AR-CpGs (except cg09809672) and enabled age prediction with R2 = 0.85, mean absolute deviation (MAD) = 4.22, while the SVR model enabled age prediction with R2 = 0.86, MAD = 4.01. An independent set of 80 samples was used to test the two models′ performance and the prediction MAD for the validation set was 4.71 and 4.56 for the stepwise regression and SVR models, respectively. The number of correct predictions for ±5 years achieved a high level of 67.50 % and 73.75 %, respectively for the stepwise regression and SVR models. In general, the SVR model was superior to the linear regression model in age prediction. These results suggest that these seven CpG sites would be useful for age prediction in blood samples from the Chinese Han population.