Significant genetic correlations among Caucasians at forensic DNA loci

Abstract

Although the effect of population differentiation on the forensic use of DNA profiles has been the subject of controversy for some years now, the debate has largely failed to focus on the genetical questions directly relevant to the forensic context. We re-analyse two published data sets and find that they convey much the same message for forensic inference, in contrast with the dramatically differing conclusions of the original authors. The analysis is likelihood-based and combines information across loci and across populations without assuming constant genetic differentiation. Our results suggest that the relevant genetic correlation coefficients are too large to be ignored in forensic work: although DNA profile evidence is typically very strong, the effect of genetic correlations can be important in some cases. Such correlations can, however, be accommodated in an appropriate assessment of evidential strength so that population genetic issues should not present a barrier to the efficient and fair use of DNA profile evidence.

Empirical Bayes Inference of Pairwise FST and Its Distribution in the Genome

“…as given by Wright (1969), Rannala and Hartigan (1996), Balding and Nichols (1997), Lockwood et al (2001), Balding (2003), and Kitada and Kishino (2004).…”

A Reassessment of Genetic Diversity in Icelanders: Strong Evidence from Multiple Loci for Relative Homogeneity Caused by Genetic Drift

“…Nicholson et al (2002) developed a simple statistical model, based on a Bayesian framework using Markov chain Monte Carlo, for the analysis of such data and reported their application of this to data on 83 SNPs typed in samples from Iceland, and French and Utah CEPH individuals (with sample sizes of around 70, 30, and 60 chromosomes, respectively), thus allowing comparisons between these three groups of European descent. Using these data, Nicholson et al (2002) estimated a single parameter, their c, which quantifies the amount of drift, or variation in allele frequencies, observed in each population relative to common ancestral populations (see also Balding & Nichols, 1997). An advantage to the approach introduced by Nicholson et al (2002) is that it permits evaluation of whether the underlying model is a good description of the data.…”

Recent developments in genetic data analysis: what can they tell us about human demographic history?

“…For more recent moment-based approaches to the problem, see Beaumont and Nichols (1996), , and Schlö tterer (2002). Again, as with the method of Rannala and Hartigan (1996), other than in studies by the authors (primarily in a forensic setting), there appear to have been no applications of the methodology of Balding and Nichols (1997) in human population genetics.…”

“…However, given the strong assumptions in the former methods that gene flow is only between the populations used in the study, this latter more modest approach may be more robust. Another useful feature of the method of Balding and Nichols (1997) implemented in the program FSTMET (http:/ /www.reading.ac.uk/ Statistics/genetics/software/fstmet/) is that it allows a hierarchical approach to modelling where immigration rates are allowed to vary among loci (see also Balding 2003). As discussed above, in relation to models of demographic growth, such an approach is useful to identify loci that might be under selection.…”

“…For given subpopulation gene frequencies, the frequency distribution in a sample is given by a multinomial distribution, and the multinomial-Dirichlet is the multinomial distribution integrated over all the possible subpopulation frequency distributions, given by the Dirichlet. Although the likelihood can be compactly expressed, computerintensive methods are required to integrate over unknown baseline gene frequencies (Balding and Nichols, 1997;Holsinger, 1999;Holsinger et al, 2002).…”