Analysis of a DNA mixture involving Romani reference populations

Here we present an Italian criminal case that shows how statistical methods can be used to extract information from a series of mixed DNA profiles. The case involves several different individuals and a set of different DNA traces. The case possibly involves persons of interest of a small population of Romani origin. First, a brief description of the case is provided. Secondly, we introduce some heuristic tools that can be used to evaluate the data and we also briefly outline the statistical model used for analysing DNA mixtures. Finally, we illustrate some of the findings on the case and discuss further directions of research. The results show how the use of different population database allele frequencies for analysing the DNA mixtures can lead to very different results, some seemingly inculpatory and some seemingly exculpatory. We also illustrate the results obtained from combining the evidence from different samples.

Overdispersion in allelic counts and θ-correction in forensic genetics

We present a statistical model for incorporating the extra variability in allelic counts due to subpopulation structures. In forensic genetics, this effect is modelled by the identical-by-descent parameter θ, which measures the relationship between pairs of alleles within a population relative to the relationship of alleles between populations (Weir, 2007). In our statistical approach, we demonstrate that θ may be defined as an overdispersion parameter capturing the subpopulation effects. This formulation allows derivation of maximum likelihood estimates of the allele probabilities and θ together with computation of the profile log-likelihood, confidence intervals and hypothesis testing. In order to compare our method with existing methods, we reanalysed FBI data from Budowle and Moretti (1999) with allele counts in six US subpopulations. Furthermore, we investigate properties of our methodology from simulation studies.