In a forensic context, determining the most likely population of origin of a crime-scene DNA sample can provide investigative leads when comparative DNA profiling fails to reach an identification. Various software programs were developed to perform population assignment using SNP data. However, none of these tools support microhaplotype (MH) profiles. This is mostly due to the multi-allelic nature of MHs, that requires extensive databases to obtain robust frequency estimates, particularly for rare and population specific haplotypes. Decomposable graphical models (DGMs) were suggested as a possible approach to tackle this problem. DGMs allow for evaluation of the underlying dependencies among the SNPs within the MHs, and for division of the MH into smaller independent subsets of SNPs. Allele frequencies are computed per subset instead of the entire MH configuration, which enables better estimates for rare and unobserved haplotypes.

The objective of this work was to apply DGMs on MH data and fully exploit the MH multi-allelic nature while reducing the dependence on large databases for reliable haplotype frequency estimation. This computational approach will be integrated into the GenoGeographer interface, with the goal of providing the forensic community with a tool that supports custom MH panels for population assignment.

The MHappaMundi custom AmpliSeq panel consists of 82 ancestry informative MHs and 434 allele-defining SNPs. MHappaMundi data for seven metapopulations from publicly available datasets (Sub-Saharan Africans, Europeans, Middle Easterners, South Asians, East Asians, African Americans, and Admixed Americans) were used to evaluate the tool’s performance. The population assignment of MH profiles consisted of two steps: 1) an outlier detection test for each of the defined reference populations 2) calculations of the evidential weights under user-defined hypotheses in the form of likelihood ratios (LRs), similar to the GenoGeographer analysis of ancestry informative SNP panels. The performance obtained for MH data was compared to GenoGeographer assignments based on SNP data from the Precision ID ancestry panel for the same individuals.

Overall, this approach may be used for population assignment at both the inter and intra-continental level. For metapopulations, the concordant assignment rates were similar for both MH (94.9%) and SNP (94.7%) data. However, for subpopulations, MHs surpassed SNPs. The proportion of concordant assignments (LR > 1) was 54.5% for SNP data while it reached more than 80% for MHs.

Our results demonstrated that MHappaMundi outperformed the Precision ID ancestry panel for population assignment, particularly for assignments at an intra-continental level.

Authors

• Pedro Rodrigues (Section of Forensic Genetics, Department of Forensic Medicine, Faculty of Health and Medical Sciences, University of Copenhagen;, Denmark)
• Torben Tvedebrink (Department of Mathematical Sciences, Aalborg University, Denmark)
• Nádia Pinto (Instituto de Investigação e Inovação em Saúde (i3S), Portugal)
• Maria João Prata (Instituto de Investigação e Inovação em Saúde (i3S), Portugal)
• Claus Børsting (Section of Forensic Genetics, Department of Forensic Medicine, Faculty of Health and Medical Sciences, University of Copenhagen;, Denmark)
• Leonor Gusmão (DNA Diagnostic Laboratory (LDD), State University of Rio de Janeiro (UERJ), Brazil)
• Vania Pereira (Section of Forensic Genetics, Department of Forensic Medicine, Faculty of Health and Medical Sciences, University of Copenhagen;, Denmark)