Forensic age prediction based on DNA methylation using next generation sequencing

Age estimation basing on DNA methylation technique has been on increase in the past years. In this study, we assessed age-associated methylation for 12 CpG sites, (EDARADD, E LO VL2 RASSF5, MIR29B2CHG, FH L2, SM C4, T RI M59, L D B2, ELOVL2-AS1, CNTNAP2, ARHGAP22 and CCDC102B) in 25 buccal samples of living individuals aged 4 -99 years using Next Generation sequencing platform. Selection of the CpG sites used in our model was based on short-lists published by other research groups. Linear regression and multiple linear regression were used to develop an age prediction model. A multiple linear regression validation model with 10 CpG sites (EDARADD, ELOVL2 RASSF5, MIR29B2CHG, FHL2, SMC4, LDB2, CNTNAP2, ARHGAP22 and CCDC102B) revealed a p-value of 0.02976 which is statistically significant. R2 of 0.736 was obtained indicating a strong correlation between chronological age and predicted age with a Mean Absolute Deviation (MAD) of 5.1 years between chronological age and predicted age and an adjusted R2 of 0.5199. In the cross validated model, the variables(CpG sites) RASSF5 (p = 0.033), MIR29B2CHG (p = 0.034) and SMC4 (p =0.0323) had a significant influence on the predictor variable (Age) at a 95% confidence level. The Multiple R square value of 0.9878 with this training set of data was obtained indicating a strong positive correlation between predicted age with a Mean Absolute Deviation (MAD) of 2.3 years between chronological age and predicted age and an adjusted R square value of 0.9332. Seven methylation standards of 0%,10%, 25%, 50%, 75%and 100% percentage methylation were included in this study in order to evaluate the accuracy of methylation levels obtained by bisulphite sequencing. R square value of 0.9887 was obtained indicating a strong correlation between expected methylation and observed methylation. This demonstrated the reliability and precision of the technique in measuring methylation patterns.