Epigenetic biomarkers, particularly DNA methylation at CpG sites, have emerged as powerful predictors of forensically relevant traits, including tissue origin, lifestyle-associated factors such as smoking, and chronological age. Existing models for age estimation, tissue identification, and smoking inference have demonstrated robust performance, but their reliance on separate assays limits practical application. To address this gap, we developed COSA (a Consolidated prediction panel for Origin, Smoking, and Age), an integrated methylation-based assay implemented through amplicon-based massively parallel sequencing (MPS). COSA consolidates 126 previously reported CpG markers from multiple validated models into 67 amplicons, thereby enabling the simultaneous prediction of tissue source, smoking status, and chronological age from a single analysis. By unifying these established markers, COSA provides a scalable and streamlined solution for comprehensive forensic epigenetic profiling.

The panel comprises three functional modules. First, body fluid identification incorporates 9 CpG markers specific to blood, semen, saliva, menstrual blood, and vaginal fluid, supporting accurate determination of sample origin in forensic framework. Additionally, tissue source inference extends to internal organs through 18 CpGs targeting blood, liver, skeletal muscle, heart, brain, epidermis, dermis, kidney, and lung. Second, lifestyle inference is supported by 13 CpGs, including the well-characterized cg05575921 locus in the AHRR gene for smoking prediction. Third, age estimation is incorporated through three tissue-specific models optimized for blood, saliva, and semen, which are the fluids most frequently encountered in forensic investigations.

Methodological refinements were essential to achieve balanced multiplex amplification. Multiplex PCR for bisulfite-converted DNA is challenged by issues related to primer compatibility and GC-content variation. To overcome this, we implemented a touchdown PCR strategy that improved amplification balance and coverage uniformity across multiple loci. Several primer sets were newly designed or modified to optimize amplicon length and annealing temperature, ensuring robust co-amplification within the 67-amplicon panel. Importantly, using as little as 20 ng of bisulfite-converted DNA, the COSA panel generated consistent coverage and prediction performance comparable to that observed in single-trait models.

Overall, COSA integrates three major forensic prediction modules, including origin classifiers for body fluids and organ tissues, tissue-specific age estimators, and a smoking-status predictor within a single DNA workflow. This panel represents a practical and scalable tool for forensic laboratories seeking to maximize information yield from limited DNA, advancing the application of epigenetics in human identification and investigative intelligence.

Authors

• Soobin Yang (Seoul National University College of Medicine, South Korea)
• Ga Hyung Kim (Seoul National University College of Medicine, South Korea)
• Moon Hyun So (Seoul National University College of Medicine, South Korea)
• Soong Deok Lee (Seoul National University College of Medicine, South Korea)
• Hwan Young Lee (Seoul National University College of Medicine, South Korea)