Strategies for pairwise searches in forensic kinship analysis.

Testing kinship between pairs of individuals is central to a wide range of applications. We focus on cases where many tests are done jointly. Typical examples include cases where DNA profiles are available from a burial site, a plane crash or a database of convicted offenders. The task is to determine the relationships between DNA profiles or individuals. Our approach generalises previous methods and implementations in several respects. We model general, possibly inbred, pairwise relationships which is important for non-human applications and in archaeological studies of ancient inbred populations. Furthermore, we do not restrict attention to autosomal markers. Some cases, such as distinguishing between maternal and paternal half siblings, can be solved using X-chromosomal markers. When many tests are done, the risk of errors increases. We address this problem by building on the theory of multiple testing and show how optimal thresholds for tests can be determined. We point out that the likelihood ratios in a blind search may be dependent so multiple testing methods and interpretation need to account for this. In addition, we show how a Bayesian approach can be helpful. Our examples, using simulated and real data, demonstrate the practical importance of the methods and implementation is based on freely available software.

Joint DNA-based disaster victim identification.

We address computational and statistical aspects of DNA-based identification of victims in the aftermath of disasters. Current methods and software for such identification typically consider each victim individually, leading to suboptimal power of identification and potential inconsistencies in the statistical summary of the evidence. We resolve these problems by performing joint identification of all victims, using the complete genetic data set. Individual identification probabilities, conditional on all available information, are derived from the joint solution in the form of posterior pairing probabilities. A closed formula is obtained for the a priori number of possible joint solutions to a given DVI problem. This number increases quickly with the number of victims and missing persons, posing computational challenges for brute force approaches. We address this complexity with a preparatory sequential step aiming to reduce the search space. The examples show that realistic cases are handled efficiently. User-friendly implementations of all methods are provided in the R package dvir, freely available on all platforms.

Making decisions in missing person identification cases with low statistical power.

The present work proposes a general strategy for dealing with missing person identification cases through DNA-database search. Our main example is the identification of abducted children in the last civic-dictatorship of Argentina, known as the "Missing Grandchildren of Argentina". Particularly we focus on those pedigrees where few, or only distant relatives of the missing person are available, resulting in low statistical power. For such complex cases we provide a statistical method for selecting a likelihood ratio (LR) threshold for each pedigree based on error rates. Furthermore, we provide an open-source user friendly software for computing LR thresholds and error rates. The strategy described in the paper could be applied to other large-scale cases of DNA-based identification hampered by low statistical power.

Pairwise relatedness testing in the context of inbreeding: expectation and variance of the likelihood ratio.

In this paper we investigate various effects of inbreeding on the likelihood ratio (LR) in forensic kinship testing. The basic setup of such testing involves formulating two competing hypotheses, in the form of pedigrees, describing the relationship between the individuals. The likelihood of each hypothesis is computed given the available genetic data, and a conclusion is reached if the ratio of these exceeds some pre-determined threshold. An important aspect of this approach is that the hypotheses are usually not exhaustive: The true relationship may differ from both of the stated pedigrees. It is well known that this may introduce bias in the test results. Previous work has established formulas for the expected value and variance of the LR, given the two competing hypotheses and the true relationship. However, the proposed method only handles cases without inbreeding. In this paper we extend these results to all possible pairwise relationships. The key ingredient is formulating the hypotheses in terms of Jacquard coefficients instead of the more restricted Cotterman coefficients. While the latter describe the relatedness between outbred individuals, the more general Jacquard coefficients allow any level of inbreeding. Our approach also enables scrutiny of another frequently overlooked source of LR bias, namely background inbreeding. This ubiquitous phenomenon is usually ignored in forensic kinship computations, due to lack of adequate methods and software. By leveraging recent work on pedigrees with inbred founders, we show how background inbreeding can be modeled as a continuous variable, providing easy-to-interpret results in specific cases. For example, we show that if true siblings are subjected to a test for parent-offspring, moderate levels of background inbreeding are expected to inflate the LR by more than 50%.

Prioritising family members for genotyping in missing person cases: A general approach combining the statistical power of exclusion and inclusion.

Missing person identification typically involves genetic matching of a person of interest against relatives of the missing person. In cases with few available relatives, exhumations or other substantial efforts may be necessary in order to secure adequate statistical power. We propose a simulation approach for solving prioritisation problems arising in such cases. Conditioning on the already typed individuals we estimate the power of each alternative, both to detect the true person, and to exclude false candidates. Graphical summaries of the simulations are given in complementary power plots, facilitating interpretation and decision making. Through a series of examples originating from the well-known Missing grandchildren of Argentina we demonstrate that our method may untangle complex prioritisation problems and other power-related questions. In particular we offer novel insights in recent cases where only children of the potential match are available for testing. We also show that X-chromosomal markers may give high statistical power in missing person identification, but that this requires careful selection of relatives for genotyping. All simulations, power calculations and plots are done with the R package forrel.

Advancing estimation of chronological age by utilizing available evidence based on two radiographical methods.

This paper describes a strategy for estimating chronological age of individuals based on age indicators of X-ray of the hand and the third molar tooth. The great majority of studies in the field provide group-wise data of different formats, which makes them difficult to compare and utilize in a model. In this paper, we have provided a framework to utilize different types of data formats to build a common model for estimating chronological age. We used transition analysis to describe the relationship between the age indicators and chronological age. Further, likelihood ratio weight of evidence and posterior distribution of chronological age were used to model the distribution of chronological age given the observed age indicators. Being able to utilize such a large amount of data, with different data formats, from different studies, as presented in this paper improves previous age estimation methods.

The failing measurement of attitudes: How semantic determinants of individual survey responses come to replace measures of attitude strength.

The traditional understanding of data from Likert scales is that the quantifications involved result from measures of attitude strength. Applying a recently proposed semantic theory of survey response, we claim that survey responses tap two different sources: a mixture of attitudes plus the semantic structure of the survey. Exploring the degree to which individual responses are influenced by semantics, we hypothesized that in many cases, information about attitude strength is actually filtered out as noise in the commonly used correlation matrix. We developed a procedure to separate the semantic influence from attitude strength in individual response patterns, and compared these results to, respectively, the observed sample correlation matrices and the semantic similarity structures arising from text analysis algorithms. This was done with four datasets, comprising a total of 7,787 subjects and 27,461,502 observed item pair responses. As we argued, attitude strength seemed to account for much information about the individual respondents. However, this information did not seem to carry over into the observed sample correlation matrices, which instead converged around the semantic structures offered by the survey items. This is potentially disturbing for the traditional understanding of what survey data represent. We argue that this approach contributes to a better understanding of the cognitive processes involved in survey responses. In turn, this could help us make better use of the data that such methods provide.

Evaluating the statistical power of DNA-based identification, exemplified by 'The missing grandchildren of Argentina'.

Methods and implementations of DNA-based identification are well established in several forensic contexts. However, assessing the statistical power of these methods has been largely overlooked, except in the simplest cases. In this paper we outline general methods for such power evaluation, and apply them to a large set of family reunification cases, where the objective is to decide whether a person of interest (POI) is identical to the missing person (MP) in a family, based on the DNA profile of the POI and available family members. As such, this application closely resembles database searching and disaster victim identification (DVI). If parents or children of the MP are available, they will typically provide sufficient statistical evidence to settle the case. However, if one must resort to more distant relatives, it is not a priori obvious that a reliable conclusion is likely to be reached. In these cases power evaluation can be highly valuable, for instance in the recruitment of additional family members. To assess the power in an identification case, we advocate the combined use of two statistics: the Probability of Exclusion, and the Probability of Exceedance. The former is the probability that the genotypes of a random, unrelated person are incompatible with the available family data. If this is close to 1, it is likely that a conclusion will be achieved regarding general relatedness, but not necessarily the specific relationship. To evaluate the ability to recognize a true match, we use simulations to estimate exceedance probabilities, i.e. the probability that the likelihood ratio will exceed a given threshold, assuming that the POI is indeed the MP. All simulations are done conditionally on available family data. Such conditional simulations have a long history in medical linkage analysis, but to our knowledge this is the first systematic forensic genetics application. Also, for forensic markers mutations cannot be ignored and therefore current models and implementations must be extended. All the tools are freely available in Familias (http://www.familias.no) empowered by the R library paramlink. The above approach is applied to a large and important data set: 'The missing grandchildren of Argentina'. We evaluate the power of 196 families from the DNA reference databank (Banco Nacional de Datos Genéticos, http://www.bndg.gob.ar. As a result we show that 58 of the families have poor statistical power and require additional genetic data to enable a positive identification.

DNA Commission of the International Society for Forensic Genetics (ISFG): Guidelines on the use of X-STRs in kinship analysis.

Forensic genetic laboratories perform an increasing amount of genetic analyses of the X chromosome, in particular to solve complex cases of kinship analysis. For some biological relationships X-chromosomal markers can be more informative than autosomal markers, and there are a large number of markers, methods and databases that have been described for forensic use. Due to their particular mode of inheritance, and their physical location on a single chromosome, some specific considerations are required when estimating the weight of evidence for X-chromosomal marker DNA data. The DNA Commission of the International Society for Forensic Genetics (ISFG) hereby presents guidelines and recommendations for the use of X-chromosomal markers in kinship analysis with a special focus on the biostatistical evaluation. Linkage and linkage disequilibrium (association of alleles) are of special importance for such evaluations and these concepts and the implications for likelihood calculations are described in more detail. Furthermore it is important to use appropriate computer software that accounts for linkage and linkage disequilibrium among loci, as well as for mutations. Even though some software exist, there is still a need for further improvement of dedicated software.

Exact likelihood ratio calculations for pairwise cases.

Some practical and theoretical aspects of evaluation of evidence based on the likelihood ratio (LR) in kinship cases are discussed. If relationships are complex or if complicating factors like mutation, correction for population structure or silent alleles need to be accounted for, available software may fail. We present an explicit general formula for non-inbred pairwise cases. Equipped with this formula it is possible to evaluate, say, how strongly a shared rare allele, points towards a specific relationship. Moreover, a general expression as the one presented, adds to the understanding of models and the underlying biological mechanisms. It is also useful for checking software and defining the limitations of programs. Some ideas for improving software may also be generated by the derivation of exact expressions. We argue that a proportional mutation model is well suited from a pragmatic point of view and derive some theoretical properties of this model. Several examples based on the general pairwise formula and its implementation in the freely available R package mut are presented.

Key individuals for discerning pedigrees belonging to the same autosomal kinship class.

The existence of pedigrees belonging to the same kinship class (i.e. indistinguishable through independent markers) is well known in the forensic community, and theoretical frameworks for autosomal and X-chromosomal markers were already developed for a pair of individuals. Nevertheless, studies for the cases where a greater number of individuals is available for testing are still lacking. With this work, we intend to pave the way for a theoretical and general framework, identifying the individuals/relatives that should be chosen to distinguish autosomal analyses between such pedigrees. In this work we identify the individuals/relatives that are non-informative for calculations (the pedigrees under discussion remaining indistinguishable independently of their genetic profile), as well as those that will likely be very informative, influencing the statistical outcome. For example, given the respective genotypes, to compare the likelihoods of the father of the individual B to be: (a.) the father, or (b.) a full-brother, of the individual A, the hypotheses H1: "The individual A is paternal half-sibling of the individual B", and H0: "The individual A is paternal uncle/aunt of the individual B" are considered. It is proved that considering just individuals A and B the hypotheses are equally likely. In this work we show that the same is also true for the case where the mother of B is available for testing, notwithstanding the hypotheses being differently weighted if the mother of A is considered. Similar considerations are done for other kinship hypotheses and/or individuals/relatives. Irrelevance of the genetic profile of some specific relatives are mathematically demonstrated, and data obtained from 20,000 simulated families are presented for the other cases.

The implications of shedder status and background DNA on direct and secondary transfer in an attack scenario.

In court questions are often raised related to how trace DNA was deposited, directly during the crime or innocently for instance by secondary transfer. It is therefore of interest to have knowledge of the probability of transfer or secondary transfer in different situations. Factors that could influence transfer probabilities are background DNA and the shedder status of the involved persons. In this study, we have classified participants as high or low DNA shedders. We observed DNA transfer in a simulated attack scenario, and demonstrated that shedder status has a significant influence of transfer rates. We have examined the background DNA in samples from T-shirts worn in an area with frequent human traffic and detected multiple contributors. We further demonstrated that DNA from co-workers of a T-shirt wearer can be secondarily transferred from the environment and detected in samples, and that the composition of background DNA is correlated with the shedder status of the wearer. Finally, we have illustrated the inference with the results of transfer probabilities and a fictive case with the use of a Bayesian network.

Degradation in forensic trace DNA samples explored by massively parallel sequencing.

Routine forensic analysis using STRs will fail if the DNA is too degraded. The DNA degradation process in biological stain material is not well understood. In this study we sequenced old semen and blood stains by massively parallel sequencing. The sequence data coverage was used to measure degradation across the genome. The results supported the contention that degradation is uniform across the genome, showing no evidence of regions with increased or decreased resistance towards degradation. Thus the lack of genetic regions robust to degradation removes the possibility of using such regions to further optimize analysis performance for degraded DNA.

Characterization of degradation and heterozygote balance by simulation of the forensic DNA analysis process.

Simulation experiments were used to show the impact of varying extraction efficiency, aliquot proportion, and PCR efficiency on the heterozygote balance of a range of diploid and haploid cells. Reducing either parameters introduces variance. It is well-known that the variance in heterozygote balance increases as the amount of DNA is reduced. Surprisingly the distribution is in fact diamond shaped - the variance start to decrease at very low amounts of DNA. Simulations suggest that pristine diluted DNA is an acceptable approximation in validations to infer heterozygote balance. However, the difference in distribution of the variance between diploid and haploid cell types may, under some circumstances, need to be considered in statistical models. Finally, we exemplify how simulations can be used to predict the outcome of PCR for degraded samples. Visualizing the predicted DNA profile as an electropherogram can help to identify the best approach for sample processing.

The likelihood ratio as a random variable for linked markers in kinship analysis.

The likelihood ratio is the fundamental quantity that summarizes the evidence in forensic cases. Therefore, it is important to understand the theoretical properties of this statistic. This paper is the last in a series of three, and the first to study linked markers. We show that for all non-inbred pairwise kinship comparisons, the expected likelihood ratio in favor of a type of relatedness depends on the allele frequencies only via the number of alleles, also for linked markers, and also if the true relationship is another one than is tested for by the likelihood ratio. Exact expressions for the expectation and variance are derived for all these cases. Furthermore, we show that the expected likelihood ratio is a non-increasing function if the recombination rate increases between 0 and 0.5 when the actual relationship is the one investigated by the LR. Besides being of theoretical interest, exact expressions such as obtained here can be used for software validation as they allow to verify the correctness up to arbitrary precision. The paper also presents results and advice of practical importance. For example, we argue that the logarithm of the likelihood ratio behaves in a fundamentally different way than the likelihood ratio itself in terms of expectation and variance, in agreement with its interpretation as weight of evidence. Equipped with the results presented and freely available software, one may check calculations and software and also do power calculations.

Contamination during criminal investigation: Detecting police contamination and secondary DNA transfer from evidence bags.

As the profiling systems used in forensic analyses have become more sensitive in recent years, the risk of detecting a contamination in a DNA sample has increased proportionally. This requires more stringent work protocols and awareness to minimize the chance of contamination. Although there is high consciousness on contamination and best practice procedures in forensic labs, the same requirements are not always applied by the police. In this study we have investigated the risk of contamination from police staff. Environmental DNA was monitored by performing wipe tests (sampling of hot spots) at two large police units (scenes of crime departments). Additionally, the DNA profiles of the scenes of crime officers were compared to casework samples that their own unit had investigated in the period of 2009-2015. Furthermore, a pilot study to assess whether DNA from the outside package of an exhibit could be transferred to a DNA sample was carried out. Environmental DNA was detected in various samples from hot spots. Furthermore, 16 incidences of previously undetected police-staff contamination were found in casework that had been submitted between 2009 and 2015. In 6 cases the police officers with a matching DNA profile reported that they had not been involved with the case. We have demonstrated that DNA from the outside package can be transferred to an exhibit during examination. This experience demonstrates that when implementing the new multiplex systems, it is important to ensure that 'best practice' procedures are upgraded, and appropriate training is provided in order to ensure that police are aware of the increased contamination risks.

Lipoprotein subfractions by nuclear magnetic resonance are associated with tumor characteristics in breast cancer.

BACKGROUND: High-Density Lipoprotein (HDL)-cholesterol, has been associated with breast cancer development, but the association is under debate, and whether lipoprotein subfractions is associated with breast tumor characteristics remains unclear. METHODS: Among 56 women with newly diagnosed invasive breast cancer stage I/II, aged 35-75 years, pre-surgery overnight fasting serum concentrations of lipids were assessed, and body mass index (BMI) was measured. All breast tumors were immunohistochemically examined in the surgical specimen. Serum metabolomics of lipoprotein subfractions and their contents of cholesterol, free cholesterol, phospholipids, apolipoprotein-A1 and apolipoprotein-A2, were assessed using nuclear magnetic resonance. Principal component analysis, partial least square analysis, and uni- and multivariable linear regression models were used to study whether lipoprotein subfractions were associated with breast cancer tumor characteristics. RESULTS: The breast cancer patients had following means: age at diagnosis: 55.1 years; BMI: 25.1 kg/m(2); total-Cholesterol: 5.74 mmol/L; HDL-Cholesterol: 1.78 mmol/L; Low-Density Lipoprotein (LDL)-Cholesterol: 3.45 mmol/L; triglycerides: 1.18 mmol/L. The mean tumor size was 16.4 mm, and the mean Ki67 hotspot index was 26.5%. Most (93%) of the patients had estrogen receptor (ER) positive tumors (≥ 1% ER+), and 82% had progesterone receptor (PgR) positive tumors (≥ 10% PgR+). Several HDL subfraction contents were strongly associated with PgR expression: Apolipoprotein-A1 (ß 0.46, CI 0.22-0.69, p < 0.001), HDL cholesterol (ß 0.95, CI 0.51-1.39, p < 0.001), HDL free cholesterol (ß 2.88, CI 1.28-4.48, p = 0.001), HDL phospholipids (ß 0.70, CI 0.36-1.04, p < 0.001). Similar results were observed for the subfractions of HDL1-3. We observed inverse associations between HDL phospholipids and Ki67 (ß -0.25, p = 0.008), and in particular between HDL1's contents of cholesterol, phospholipids, apolipoprotein-A1, apolipoprotein-A2 and Ki67. No association was observed between lipoproteins and ER expression. CONCLUSION: Our findings hypothesize associations between different lipoprotein subfractions, and PgR expression, and Ki 67 % in breast tumors. These findings may have clinical implications, but require confirmation in larger studies.

Exclusion probabilities and likelihood ratios with applications to mixtures.

The statistical evidence obtained from mixed DNA profiles can be summarised in several ways in forensic casework including the likelihood ratio (LR) and the Random Man Not Excluded (RMNE) probability. The literature has seen a discussion of the advantages and disadvantages of likelihood ratios and exclusion probabilities, and part of our aim is to bring some clarification to this debate. In a previous paper, we proved that there is a general mathematical relationship between these statistics: RMNE can be expressed as a certain average of the LR, implying that the expected value of the LR, when applied to an actual contributor to the mixture, is at least equal to the inverse of the RMNE. While the mentioned paper presented applications for kinship problems, the current paper demonstrates the relevance for mixture cases, and for this purpose, we prove some new general properties. We also demonstrate how to use the distribution of the likelihood ratio for donors of a mixture, to obtain estimates for exceedance probabilities of the LR for non-donors, of which the RMNE is a special case corresponding to L R>0. In order to derive these results, we need to view the likelihood ratio as a random variable. In this paper, we describe how such a randomization can be achieved. The RMNE is usually invoked only for mixtures without dropout. In mixtures, artefacts like dropout and drop-in are commonly encountered and we address this situation too, illustrating our results with a basic but widely implemented model, a so-called binary model. The precise definitions, modelling and interpretation of the required concepts of dropout and drop-in are not entirely obvious, and we attempt to clarify them here in a general likelihood framework for a binary model.