Comparison of two online algorithm methods for forensic ancestry inference

Abstract

The forensic prediction of the biogeographic ancestry based on DNA typing has become more widespread. The search for optimized panels, consisting of a small but efficient and robust set of ancestry informative markers (AIMs), has received increased interest. Several panels published in recent literature would provide excellent information on ancestry as a useful forensic investigative tool. Relying on the genetic profiles from these panels, accurate and efficient estimation of an individual ancestry depends on excellent algorithmic methods. In the present work, a comparison of the FROG-kb (http://frog.med.yale.edu) and Bayesian classification approaches of Snipper (http://mathgene.usc.es/snipper/) was carried out to assess the ability of genetic ancestry inference for the Kidd Lab 55 AISNP panel (Kidd et al., FSI Genetics 2014 (10), 23–32). Preliminary reported data from the online SPSmart browser illustrated that both algorithmic inference methods were adaptable for forensic ancestry assignments of an individual. For a few individuals, especially originating from admixed populations, ancestral assignment was inaccurate or discordant according to the likelihood calculations in FROG-kb or the ratio from Snipper forensic ancestry analysis portal. Therefore, future improvement will require more populations adding into the reference populations for the likelihood function in FROG-kb, as well as appropriate training sets applying to the online Snipper analysis tool.