Mixture evidence remains one of the most challenging issues in forensic DNA typing. In sexual assault cases, differential extraction can be used to separate sperm from other cells and obtain semen-specific genotypes from mixtures of semen and vaginal fluid. However, recovering sufficient quantity and quality of male DNA using this method remains difficult. Recent studies have highlighted the potential of combining microhaplotypes (MHs) with tissue-specific DNA methylation markers for body fluid-specific genotyping. In this study, we developed a tissue- and body fluid-specific genotyping method using target-specific CpG and adjacent MH (CpG–MH) markers for the analysis of forensic mixtures. We designed ten CpG–MH markers, comprising four epidermis-specific and six semen-specific CpG sites, each including one CpG and two to four SNPs. Due to the indistinguishable nature of C/T or G/A alleles after bisulfite conversion, two multiplex panels were constructed to separately amplify the top and bottom DNA strands. Using bisulfite-converted DNA, a two-step PCR strategy was applied to amplify target regions and incorporate sequencing adapters. A total of 90 samples were analyzed, including 80 mixture samples (with semen or epidermis as the minor contributor mixed with blood, vaginal fluid, or saliva at ratios of 1:1, 3:1, 5:1, and 10:1) and 10 single-source samples. Stable sequencing data were generated using the Illumina MiSeq v3 platform. After data processing and mapping, reads containing tissue-specific methylation patterns at CpG sites were selected. Subsequently, allele count-based filtering was applied to remove non-target reads by setting a threshold of 20% of total clustered allele counts for each marker. The remaining reads were used to determine target tissue-specific genotypes. In semen mixtures, target tissue-specific genotypes at each marker perfectly matched the reference profiles at 1:1 and 3:1 ratios, while the average accuracy was 84.17% at 5:1 and 10:1 ratios. In epidermis mixtures, the average accuracy was 81.25% at 1:1 and 3:1 ratios, but performance gradually decreased as the proportion of epidermis decreased. In conclusion, this method demonstrated effective performance for target tissue-specific genotyping, particularly in balanced mixtures which remain challenging even for probabilistic genotyping approaches. However, its performance was reduced in highly imbalanced mixtures, and individual-specific methylation variation may affect accurate profile recovery. Future studies should address these limitations and expand marker sets to accommodate more complex forensic scenarios.

Authors

• Soyi Park (Department of Forensic Medicine, Seoul National University College of Medicine, South Korea)
• Sang Un Park (Department of Forensic Medicine, Seoul National University College of Medicine, South Korea)
• Jong-Lyul Park (Aging Convergence Research Center, KRIBB, South Korea)
• Jeong Min Lee (Department of Forensic Medicine, Seoul National University College of Medicine, South Korea)
• Kyun-Hee Lee (Aging Convergence Research Center, KRIBB, South Korea)
• Hae Ri Park (Aging Convergence Research Center, KRIBB, South Korea)
• Hwan Young Lee (Department of Forensic Medicine, Seoul National University College of Medicine, South Korea)