Genotype likelihood ratio distributions and random match probability: Generalization, calculation and application

Perlin, M.W. Genotype likelihood ratio distributions and random match probability: Generalization, calculation and application, Cybergenetics Report, July, 2017.

Downloads

Article

Abstract

Computers have revolutionized DNA evidence interpretation, replacing guesswork by sound statistical inference. Probabilistic reasoning resolves complex DNA mixtures, extracting contributor genotypes whose uncertainty is expressed through probability. Information theory can tell us much about these "probabilistic" genotypes, even before making a comparison to calculate a likelihood ratio (LR). A genotype’s distribution of possible LR outcomes, under prior or posterior probability assumptions, shows the power and breadth of its match possibilities.

Genotype LR distributions can be rapidly computed by convolving independent locus distributions. The tail probability of the non-contributor distribution gives the chance that the evidence against a random person is as strong as it is against the suspect. This LR error is a generalized random match probability (RMP) for uncertain genotypes. A sexual assault case example applies these LR and RMP concepts to DNA mixture evidence and database search.

While the LR summarizes evidence, the RMP estimates error. Both statistical measures assist a trier of fact in understanding DNA evidence.