A preliminary investigation of DNA methylation-based multiplex system for semen identification and age estimation.

Abstract

The identification and age estimation of body fluid stains found at crime scenes provide crucial information for reconstructing events and narrowing down the potential suspect pool. In this study, we developed a multiplex methylation SNaPshot system containing seven age-related CpGs (AR-CpGs) and two semen-specific CpGs that enables simultaneous acquisition of information on donor age and sample type from a single reaction. Four candidate AR-CpGs derived from previous in-house screening (cg17791346, cg06880412, cg25093603, and cg06966678) were incorporated into system and showed Pearson correlation coefficients (R) ranging from 0.527 to 0.717 with chronological age. The performance of the multiplex system was evaluated using 40 semen samples and 19 non-semen samples for semen identification and age estimation. The system showed high specificity for semen identification within the evaluated sample set and remained effective with 50 ng DNA input. Five types of machine learning models were developed using a non-overlapping historical dataset of 136 semen samples and were independently evaluated using 34 semen samples with complete 7 CpG methylation profiles from the present test cohort, with performance evaluated using Mean Absolute Deviation (MAD) and Root Mean Square Error (RMSE). The best-performing linear kernel Support Vector Machine (SVM) model achieved a MAD of 4.648 years based on valid semen age prediction profiles. In addition, 20 simulated semen-containing mixture samples were analyzed to evaluate the performance of the system in semen-containing mixtures. Based on the established semen detection criteria, 12 mixtures were classified as semen detected, whereas 8 were classified as semen not detected. Age estimation was performed using evaluable mixture samples, achieving a MAD of 8.865 years. This study presents an integrated approach for semen detection and donor age estimation, offering a more efficient and practical tool for forensic casework involving suspected semen samples.