The interpretation of mixed DNA profiles remains a major challenge in forensic genetics, particularly in the presence of low-template DNA, degraded samples, multiple contributors, and allele sharing. Although short tandem repeats (STRs) are the current standard markers, their limitations in resolving complex mixtures have driven the exploration of alternative genetic systems. In recent years, the integration of probabilistic genotyping (PG) frameworks with novel genetic markers such as microhaplotypes (MHs) has opened new perspectives for improving mixture deconvolution. MHs, defined as short genomic regions containing multiple closely linked single nucleotide polymorphisms (SNPs), have emerged as promising markers due to their high polymorphism, low mutation rate, and absence of stutter artifacts. Combined with massively parallel sequencing (MPS), MHs enable direct haplotype determination with improved discriminatory power.

This study aimed to evaluate the applicability and performance of probabilistic genotyping models for the interpretation of forensic DNA mixtures using microhaplotype data generated by MPS. DNA samples were obtained from the National Institute of Standards and Technology (NIST) Reference Material (RGTM 10235). Three multi-contributor mixtures were analysed, including one highly unbalanced two-person mixture (90:10) and two three-person mixtures with varying contributor ratios (20:20:60 and 10:30:60). A 74-plex microhaplotype panel was employed, and samples were tested at two DNA input levels (1 ng and 0.3 ng) to assess performance under low-template conditions. Libraries were prepared using the Precision ID Library Kit and sequenced on the Ion GeneStudio™ S5 platform. Sequencing data were processed with HID_Microhaplotype_Research_PluginV1.5 to generate MH profiles.  Data interpretation was conducted using two complementary probabilistic approaches: the fully continuous model EuroForMix and the semi-continuous qualitative model LRmix Studio. Microhaplotypes demonstrated stable performance across input quantities, confirming robustness under low-template conditions. Both probabilistic models successfully distinguished contributor from non-contributor hypotheses, consistently producing high likelihood ratios (LRs) for true contributors and LR values below one for non-contributors.  EuroForMix effectively modelled quantitative data and accurately estimated mixture proportions in simpler scenarios, while limitations emerged in complex three-person mixtures due to allele sharing and reduced identifiability of contributor proportions. LRmix Studio provided a robust complementary approach, enabling conservative LR estimation and efficient computation, particularly in balanced mixtures.

Overall, the findings demonstrate that probabilistic genotyping frameworks can be reliably extended to microhaplotype MPS data, providing robust evidential evaluation even under challenging conditions. However, the study highlights the importance of combining continuous and semi-continuous models, improving population frequency databases, and addressing limitations related to mixture complexity and marker standardization.

Authors

• Alessia Bernini Di Michele (Department of Biomedical Sciences and Public Health, Section of Legal Medicine, Marche Polytechnic University, Italy)
• Federica Alessandrini (Department of Biomedical Sciences and Public Health, Section of Legal Medicine, Marche Polytechnic University, Italy)
• Filomena Melchionda (Department of Biomedical Sciences and Public Health, Section of Legal Medicine, Marche Polytechnic University, Italy)
• Valerio Onofri (Legal Medicine Unit, Azienda Ospedaliero-Universitaria delle Marche, Italy)
• Lucia Fiordelmondo (Legal Medicine Unit, Azienda Ospedaliero-Universitaria delle Marche, Italy)
• Eleonora Ciarimboli (Legal Medicine Unit, Azienda Ospedaliero-Universitaria delle Marche, Italy)
• Mauro Pesaresi (Department of Biomedical Sciences and Public Health, Section of Legal Medicine, Marche Polytechnic University, Italy)
• Claus Børsting (Section of Forensic Genetics, University of Copenhagen, Denmark)
• Chiara Turchi (Department of Biomedical Sciences and Public Health, Section of Legal Medicine, Marche Polytechnic University, Italy)