Fully continuous probabilistic genotyping (PG) software implements mathematical models that account for a wide range of complexities when assessing the weight of evidence of DNA mixtures in forensic casework. Among these, EuroForMix (EFM) stands as an open access software capable of analysing samples with multiple contributors, degradation and stutter artifacts. DNA mixtures are complex samples, as many alleles from different contributors overlap and intermingle, hampering interpretation. This complexity increases when there are multiple donors, imbalanced contributions, and degradation. In these scenarios, matches are more likely to occur, and a non-contributor may be stochastically assigned a positive log10(LR) value.

Related individuals present similar genetic profiles due to shared ancestry, with first degree relatives. Indeed, and unless mutation, parent-child duos share at leas one pair of alleles at each locus, while full-siblings may share two pairs, one or none. In any case, this similarity would, in theory, make them especially prone to adventitiously matching to a DNA mixture.

In this work, the rate of adventitious matches when the person of interest (POI) is either parent or full sibling is compared to the one when an unrelated individual is analysed. The LRs were reported using four different workflows in EFM v4.2.2, ordered from least to most conservative: i) Maximum Likelihood Estimation (MLE), ii) MLE with a reporting threshold of log(LR) > 3, iii) Sensitivity analysis and iv) Sensitivity analysis with a reporting threshold of log10(LR) > 3. In addition, the estimation of the number of contributors (NOC) through the Automatic Model Search option was assessed and the percentage of alleles shared by a relative or unrelated individual with the true contributor, along with the resulting log10(LR) value was analysed. A set of 35 in-house mixtures with imbalanced contributions of two and three donors, and six samples from the PROVEDit database under six different degradation treatments (36 samples in total) were used for this study.

NOC estimation proved to be correct in all but the more complex cases and showed to impact the rate of adventitious matches. More conservative workflows in EFM resulted in a reduction of up to 100% of adventitious matches compared to the MLE approach. Finally, although allele similarity with the true contributor was positively correlated with the log10(LR) result, it has also proved to be mixture-dependent.

Authors

• Javier González-Bao (Forensic Genetics Unit, Institute of Forensic Sciences, Universidade de Santiago de Compostela, Spain)
• Camila Costa (FCUP – Faculdade de Ciências, Universidade do Porto, Portugal)
• Ana Mosquera-Miguel (Forensic Genetics Unit, Institute of Forensic Sciences, Universidade de Santiago de Compostela, Spain)
• Lucía Casanova-Adán (Forensic Genetics Unit, Institute of Forensic Sciences, Universidade de Santiago de Compostela, Spain)
• María Victoria Lareu (Forensic Genetics Unit, Institute of Forensic Sciences, Universidade de Santiago de Compostela, Spain)
• Nádia Pinto (i3S – Instituto de Investigação e Inovação em Saúde, Universidade do Porto, R. Alfredo Allen 208, 4200-135, Portugal)
• María de la Puente (Forensic Genetics Unit, Institute of Forensic Sciences, Universidade de Santiago de Compostela, Spain)