Abstract:
Mixture deconvolution involves the ability to reliably decipher and separate component genotypes of individual contributors at each tested genetic marker. The ultimate objective of this study is to develop an understanding of the integrated framework for attesting the value of using known samples when appropriate to decrease uncertainty in mixture deconvolution by leveraging more of the available genotyping data and observing the impact genotype conditioning has on multiple-contributor mixtures and resulting LRs. In this study known mixtures containing two, three, four, and five contributors were separated in iterative analyses through the assumption of contributors using provided known reference samples, a process referred to as genotype peeling or genotype conditioning.To direct the order of genotype conditioning, contributor mixture weights were estimated as all contributors to the mixture were assumed by mixture weight. Conditioning by match statistic was directed without genotype assumptions, where all contributor genotypes were inferred solely on STR peak height data. Subsequent analyses of each mixture item were conducted, in which, the order of contributors was assumed from highest to lowest based on mixture weight as well as match statistics by utilizing a probabilistic program, TrueAllele®, developed by Cybergenetics. The study demonstrates how genotype conditioning effects mixture deconvolution and resulting match statistics by also considering mixture weight and the number of contributors to a mixture. The results of this study demonstrate that it is possible to generate more informative statistics by refocusing probability distributions for each contributor to the original mixture, leading to refined LRs and reduced uncertainty.
