How reliable is the sub-population model in DNA testimony?

A universal panel of STR loci for the study of polymorphism of the species Canis lupus and forensic identification of dog and wolf

Commercial panels of microsatellite (STR) loci are intended for DNA analysis of the domestic dog (Canis lupus familiaris) and, therefore, when genotyping the Grey wolf (Canis lupus lupus), most markers reveal significant deviations from the Hardy-Weinberg equilibrium and have a low informative value, which complicates their use in a forensic examination. The aim of this study was to select STR markers that equally effectively reflect population polymorphism in the wolf and the dog, and to create a universal panel for the identification of individuals in forensic science. Based on the study of polymorphisms of 34 STR loci, a CPlex panel of 15 autosomal loci and two sex loci was developed, which is equally suitable for identifying wolfs and dogs. Analysis of molecular variance (AMOVA) between samples revealed significant differentiation values (FST = 0.0828, p < 0.05), which allows the panel to be used for differentiating between wolf and dog samples. For the first time in the forensic examination of objects of animal origin in the Republic of Belarus, population subdivision coefficients (θ-values) were calculated for each of the 15 STR loci of the test system being reported. It was shown that the values of the genotype frequency, when averaged over all studied animals without and with considering the θ-value, differ by three orders of magnitude (3.39 · 10-17 and 4.71 · 10-14, respectively). The use of population subdivision coefficients will provide the researcher with the most relevant results of an expert identification study. The test system was validated in accordance with the protocol of the Scientific Working Group on DNA Analysis Methods. A computational tool was developed to automate the analysis of genetic data on the wolf and dog in the forensic examination; two guides were approved for practicing forensic experts. This methodology is being successfully used in expert practice in investigating cases of illegal hunting, animal abuse and other offenses in the Republic of Belarus.

How can courts take into account the uncertainty in a likelihood ratio?

As legal practitioners and courts become more aware of scientific methods and evidence evaluation, they are demanding measures of the reliability of expert opinion. In particular, there are calls for error rates to accompany opinion evidence in comparative forensic sciences. While error rates or confidence intervals can be useful for those disciplines that claim to identify the source of a trace, the call for these statistical tools has extended to sciences that present opinions in the form of a likelihood ratio. In this article we argue against presenting both a likelihood ratio and numerical measures of its uncertainty. We explain how the LR already encapsulates uncertainty. Instead we consider how sensitivity analyses can be used to guide the presentation of LRs that are informative to the court and not unfair to defendants.

The factor of 10 in forensic DNA match probabilities

An update was performed of the classic experiments that led to the view that profile probability assignments are usually within a factor of 10 of each other. The data used in this study consist of 15 Identifiler loci collected from a wide range of forensic populations. Following Budowle et al. [1], the terms cognate and non-cognate are used. The cognate database is the database from which the profiles are simulated. The profile probability assignment was usually larger in the cognate database. In 44%-65% of the cases, the profile probability for 15 loci in the non-cognate database was within a factor of 10 of the profile probability in the cognate database. This proportion was between 60% and 80% when the FBI and NIST data were used as the non-cognate databases. A second experiment compared the match probability assignment using a generalised database and recommendation 4.2 from NRC II (the 4.2 assignment) with a proxy for the matching proportion developed using subpopulation allele frequencies and the product rule. The findings support that the 4.2 assignment has a large conservative bias. These results are in agreement with previous research results.

The multivariate Dirichlet-multinomial distribution and its application in forensic genetics to adjust for subpopulation effects using the θ-correction

In this paper, we discuss the construction of a multivariate generalisation of the Dirichlet-multinomial distribution. An example from forensic genetics in the statistical analysis of DNA mixtures motivates the study of this multivariate extension. In forensic genetics, adjustment of the match probabilities due to remote ancestry in the population is often done using the so-called θ-correction. This correction increases the probability of observing multiple copies of rare alleles in a subpopulation and thereby reduces the weight of the evidence for rare genotypes. A recent publication by Cowell et al. (2015) showed elegantly how to use Bayesian networks for efficient computations of likelihood ratios in a forensic genetic context. However, their underlying population genetic model assumed independence of alleles, which is not realistic in real populations. We demonstrate how the so-called θ-correction can be incorporated in Bayesian networks to make efficient computations by modifying the Markov structure of Cowell et al. (2015). By numerical examples, we show how the θ-correction incorporated in the multivariate Dirichlet-multinomial distribution affects the weight of evidence.

Evaluating methods to correct for population stratification when estimating paternity indexes

The statistical interpretation of the forensic genetic evidence requires the use of allelic frequency estimates in the reference population for the studied markers. Differences in the genetic make up of the populations can be reflected in statistically different allelic frequency distributions. One can easily figure out that collecting such information for any given population is not always possible. Therefore, alternative approaches are needed in these cases in order to compensate for the lack of information. A number of statistics have been proposed to control for population stratification in paternity testing and forensic casework, Fst correction being the only one recommended by the forensic community. In this study we aimed to evaluate the performance of Fst to correct for population stratification in forensics. By way of simulations, we first tested the dependence of Fst on the relative sizes of the sub-populations, and second, we measured the effect of the Fst corrections on the Paternity Index (PI) values compared to the ones obtained when using the local reference database. The results provide clear-cut evidence that (i) Fst values are strongly dependent on the sampling scheme, and therefore, for most situations it would be almost impossible to estimate real values of Fst; and (ii) Fst corrections might unfairly correct PI values for stratification, suggesting the use of local databases whenever possible to estimate the frequencies of genetic profiles and PI values.

Analysis of global variability in 15 established and 5 new European Standard Set (ESS) STRs using the CEPH human genome diversity panel

The CEPH human genome diversity cell line panel (CEPH-HGDP) of 51 globally distributed populations was used to analyze patterns of variability in 20 core human identification STRs. The markers typed comprised the 15 STRs of Identifiler, one of the most widely used forensic STR multiplexes, plus five recently introduced European Standard Set (ESS) STRs: D1S1656, D2S441, D10S1248, D12S391 and D22S1045. From the genotypes obtained for the ESS STRs we identified rare, intermediate or off-ladder alleles that had not been previously reported for these loci. Examples of novel ESS STR alleles found were characterized by sequence analysis. This revealed extensive repeat structure variation in three ESS STRs, with D12S391 showing particularly high variability for tandem runs of AGAT and AGAC repeat units. The global geographic distribution of the CEPH panel samples gave an opportunity to study in detail the extent of substructure shown by the 20 STRs amongst populations and between their parent population groups. An assessment was made of the forensic informativeness of the new ESS STRs compared to the loci they will replace: CSF1PO, D5S818, D7S820, D13S317 and TPOX, with results showing a clear enhancement of discrimination power using multiplexes that genotype the new ESS loci. We also measured the ability of Identifiler and ESS STRs to infer the ancestry of the CEPH-HGDP samples and demonstrate that forensic STRs in large multiplexes have the potential to differentiate the major population groups but only with sufficient reliability when used with other ancestry-informative markers such as single nucleotide polymorphisms. Finally we checked for possible association by linkage between the two ESS multiplex STRs closely positioned on chromosome-12: vWA and D12S391 by examining paired genotypes from the complete CEPH data set.

A universal strategy to interpret DNA profiles that does not require a definition of low-copy-number

In this paper we critically examine the causes of the underlying confusion that relates to the issue of low-template (LT) DNA profile interpretation. Firstly, there is much difficulty in attempting to distinguish between LT-DNA vs. conventional DNA because there is no discrete 'cut-off' point that can be reasonably defined or evaluated. LT-DNA is loosely characterised by drop-out (where alleles may be missing) and drop-in (where additional alleles may be present). We have previously described probabilistic methods that can be used to incorporate these phenomena using likelihood ratio (LR) principles. This is preferred to the random man not excluded (RMNE) method, because we cannot identify a coherent way forward within the restrictions provided by this framework. Most LT-DNA profiles are interpreted using a 'consensus' profile method, we called this the 'biological model', where only those alleles that are duplicated in consecutive tests are reported. We recognise that there is an increased need for probabilistic models to take precedence over the biological model. These models are required for all kinds of DNA profiles, not just those that are believed to be low-template. We also recognise that there is a need for education and training if the methods we recommend are to be widely introduced.

Object-oriented Bayesian networks for paternity cases with allelic dependencies

This study extends the current use of Bayesian networks by incorporating the effects of allelic dependencies in paternity calculations. The use of object-oriented networks greatly simplify the process of building and interpreting forensic identification models, allowing researchers to solve new, more complex problems. We explore two paternity examples: the most common scenario where DNA evidence is available from the alleged father, the mother and the child; a more complex casewhere DNA is not available from the alleged father, but is available from the alleged father's brother. Object-oriented networks are built, using HUGIN, for each example which incorporate the effects of allelic dependence caused by evolutionary relatedness.

Use of subpopulation data in Australian forensic DNA casework

DNA profiling evidence presented in court should be accompanied by a reliable estimate of its evidential weight. In calculating such statistics, allele frequencies from commonly employed autosomal microsatellite loci are required. These allele frequencies should be collected at a level that appropriately represents the genetic diversity that exists in the population. Typically this occurs at broadly defined bio-geographic categories, such as Caucasian or Asian. Datasets are commonly administered at the jurisdictional level. This paper focuses on Australian jurisdictions and assesses whether this current practice is appropriate for Aboriginal Australian and Caucasian populations alike. In keeping with other studies we observe negligible differences between Caucasian populations within Australia when segregated geographically. However segregation of Aboriginal Australian population data along contemporary State and Territory lines appears to mask the diversity that exists within this subpopulation. For this reason datasets collated along more traditional lines may be more appropriate, particularly to distinguish the most genetically differentiated populations residing in the north of the continent.

Population Substructure Can Significantly Affect Reliability of a DNA-led Process of Identification of Mass Fatality Victims

Aiming to evaluate the effects of population substructure on the reliability of a DNA correspondence in the process of human identification, we used the model of "in silico" constructed populations with and without substructure. Effects of population substructure were evaluated at the level of locus heterozygosity, Hardy-Weinberg equilibrium and mini-haplotype distribution. Inbreeding in a subpopulation of 100 individuals through 10 generations did not significantly alter the level of heterozygosity and Hardy-Weinberg equilibrium. However, analysis of mini-haplotype distribution revealed a significant homogenization in separated subpopulations. Average observed mini-haplotype frequency (f(o)) increased to threefold from expected values (f(e)), and the number of mini-haplotypes with f(o)/f(e) above 10 increased over sixfold, suggesting that the effects of population substructure on calculated likelihood ratios (LR) might be larger than previously estimated. In most criminal cases, this would not represent a problem, whereas for identifications in large-scale mass fatality events, population substructure might considerably increase the risk of false identification.