Non-fatherhood or mutation? A probabilistic approach to parental exclusion in paternity testing

The impact of motherless paternity testing in a South African population

Background

Paternity investigations play an important role in determining biological relatedness, and in South Africa, the outcome of these investigations impacts medical, judicial and home affairs decisions. Short Tandem Repeat (STR) analysis is utilised to perform paternity and kinship analysis, due to the polymorphic nature of STR loci. The cost associated with paternity testing is high, and there is a demand for motherless testing.

Objectives

This study aims to determine what the impact of motherless testing would have been by evaluating 6182 paternity trio cases.

Methods

The AmpFLSTR™ Identifiler™ PCR Amplification kit was used to profile each of the trio cases. A scenario was created where the mother was eliminated from the test results to determine if the paternity outcome would change.

Results

Putative fathers were excluded in 27% of all cases, and in 2.5% of those cases, putative fathers would have been falsely included, had the mother not been tested. These false inclusions are attributed to coincidental STR loci that are shared between the mother and the putative father. The addition of loci to the STR profiling kit may resolve the issue; however, comparable STR data with more loci will have to be evaluated to ensure it overcomes the issue of coincidentally shared loci between unrelated individuals.

Conclusion

We would recommend that within our setup and within similar setups, the mother always be included for testing, except in extreme scenarios such as death. False inclusion of putative fathers could have serious legal implications for testing laboratories.

Mutation Rate Model Used in the DNA VIEW Program

The first problem considered in this paper is the problem of correctness of a mutation model used in the DNA VIEW program. To this end, we theoretically predict population allele frequency changes in time according to this and similar models (we determine the limit frequencies of alleles—they are uniformly distributed). Furthermore, we evaluate the speed of the above changes using computer simulation applied to our DNA database. Comparing uniformly distributed allele frequencies with these existing in the population (for example, using entropy), we conclude that this mutation model is not correct. The evolution does not follow this direction (direction of uniformly distributed frequencies). The second problem relates to the determination of the extent to which an incorrect mutation model can disturb DNA VIEW program results. We show that in typical computations (simple paternity testing without maternal mutation) this influence is negligible, but in the case of maternal mutation, this should be taken into account. Furthermore, we show that this model is inconsistent from a theoretical viewpoint. Equivalent methods result in different error levels.

Big data in forensic genetics

The potential and difficulties of the application of genome wide data in forensics are analyzed. We argue that, besides statistical, computational, ethical, economic and technical validation problems, the state of the art of population genetics theory is insufficient to deal with the forensic use of this type of data. In order to keep the current standards of quantifying and reporting genetic evidence, namely in kinship analyses and identification, substantial improvement in the theoretical framework should be reached, since to obtain genome-wide results is to provide the experts with data that they cannot quantify the corresponding evidentiary value. Therefore, while a satisfactory, generalized theoretical and biostatistical modelling is not achieved, it may well be wiser to improve the already established approaches to a limited, pre-defined number of validated genetic markers, amenable to a consensual handling and reporting. Whole genome population analyses will prove extremely useful in selecting the best suited and most efficient of those markers.

Evaluation of parentage testing accuracy of child trafficking cases: Combining the exclusion probability and likelihood ratio approaches

The Chinese government established a national anti-trafficking DNA database in 2009 to help reunite trafficked children with their families. The database collects DNA information from missing children's parents, trafficked and homeless children, then conducts parentage testing using 18 or more loci to find matched pairs. This article evaluates the matching accuracy of parentage testing in child-trafficking cases, under both Hardy-Weinberg equilibrium and population substructure. Both one-parent and two-parent scenarios are considered, and mutations are taken into account. The number of random matches is first evaluated using exclusion probability (PE). It is found that there are a large number of single parent-child pairs that match at 18 loci, but the PE approach cannot tell which are the true positive ones. The likelihood ratio (LR) approach can help distinguish the true positive matches. So the second step is to obtain the true positive rate and false positive rate of matched pairs of single parent and child according to the LR approach. Based on the results of the two-step procedure, it is concluded that more than 18 loci should be used to ensure a correct match of single parent and child.

Comparison of southern Chinese Han and Brazilian Caucasian mutation rates at autosomal short tandem repeat loci used in human forensic genetics

The short tandem repeat (STR) loci used in human genetic studies are characterized by having relatively high mutation rates. In particular, to ensure an appropriate evaluation of genetic evidence in parentage and forensic analyses, it is essential to have accurate estimates of the mutation rates associated with the commonly used autosomal and sex chromosome STR loci. Differences in STR mutation rates between different ethnic groups should also be determined. Mutation data from two laboratories working with different ethnic groups were extracted from many meiotic transmissions ascertained for 15 autosomal STR loci currently used in forensic routine. Forty-five thousand and eighty-five trios were checked for the biological consistency of maternity and paternity through the analysis of a minimum of 15 loci. Mutations were scored as paternal, maternal, or ambiguous according to the most parsimonious explanation for the inconsistency, using always the least requiring hypothesis in terms of number of repeat differences. The main findings are: (a) the overall mutation rate across the 15 loci was 9.78 × 10(-4) per gamete per generation (95% CI = 9.30 × 10(-4)-1.03 × 10(-3)), and with just 48 (out of 1,587) exceptions, all of the mutations were single-step; (b) repeat gains were more frequent than losses; (c) longer alleles were found to be more mutable; and (d) the mutation rates differ at some loci between the two ethnic groups. Large worldwide meiotic transmission datasets are still needed to measure allele-specific mutation rates at the STR loci consensually used in forensic genetics.

Resolving Sample Traces in Complex Mixtures with Microarray Analyses

This chapter reviews the microarray technology and deal with the majority of aspects regarding microarrays. It focuses on today’s knowledge of separation techniques and methodologies of complex signal, i.e. samples. Overall, the chapter reviews the current knowledge on the topic of microarrays and presents the analyses and techniques used, which facilitate such approaches. It starts with the theoretical framework on microarray technology; second, the chapter gives a brief review on statistical methods used for microarray analyses, and finally, it contains a detailed review of the methods used for discriminating traces of nucleic acids within a complex mixture of samples.

Use of prior odds for missing persons identifications

Identification of missing persons from mass disasters is based on evaluation of a number of variables and observations regarding the combination of features derived from these variables. DNA typing now is playing a more prominent role in the identification of human remains, and particularly so for highly decomposed and fragmented remains. The strength of genetic associations, by either direct or kinship analyses, is often quantified by calculating a likelihood ratio. The likelihood ratio can be multiplied by prior odds based on nongenetic evidence to calculate the posterior odds, that is, by applying Bayes' Theorem, to arrive at a probability of identity. For the identification of human remains, the path creating the set and intersection of variables that contribute to the prior odds needs to be appreciated and well defined. Other than considering the total number of missing persons, the forensic DNA community has been silent on specifying the elements of prior odds computations. The variables include the number of missing individuals, eyewitness accounts, anthropological features, demographics and other identifying characteristics. The assumptions, supporting data and reasoning that are used to establish a prior probability that will be combined with the genetic data need to be considered and justified. Otherwise, data may be unintentionally or intentionally manipulated to achieve a probability of identity that cannot be supported and can thus misrepresent the uncertainty with associations. The forensic DNA community needs to develop guidelines for objectively computing prior odds.

Response to: DNA identification by pedigree likelihood ratio accommodating population substructure and mutations

Mutation models are important in many areas of genetics including forensics. This letter criticizes the model of the paper 'DNA identification by pedigree likelihood ratio accommodating population substructure and mutations' by Ge et al. (2010). Furthermore, we argue that the paper in some cases misrepresents previously published papers.

Please see related letter: http://www.investigativegenetics.com/content/2/1/8.

DNA identification by pedigree likelihood ratio accommodating population substructure and mutations

DNA typing is an important tool in missing-person identification, especially in mass-fatality disasters. Identification methods comparing a DNA profile from unidentified human remains with that of a direct (from the person) or indirect (for example, from a biological relative) reference sample and ranking the pairwise likelihood ratios (LR) is straightforward and well defined. However, for indirect comparison cases in which several members from a family can serve as reference samples, the full power of kinship analysis is not entirely exploited. Because biologically related family members are not genetically independent, more information and thus greater power can be attained by simultaneous use of all pedigree members in most cases, although distant relationships may reduce the power. In this study, an improvement was made on the method for missing-person identification for autosomal and lineage-based markers, by considering jointly the DNA profile data of all available family reference samples. The missing person is evaluated by a pedigree LR of the probability of DNA evidence under alternative hypotheses (for example, the missing person is unrelated or if they belong to this pedigree with a specified biological relationship) and can be ranked for all pedigrees within a database. Pedigree LRs are adjusted for population substructure according to the recommendations of the second National Research Council (NRCII) Report. A realistic mutation model was also incorporated to accommodate the possibility of false exclusion. The results show that the effect of mutation on the pedigree LR is moderate, but LRs can be significantly decreased by the effect of population substructure. Finally, Y chromosome and mitochondrial DNA were integrated into the analysis to increase the power of identification. A program titled MPKin was developed, combining the aforementioned features to facilitate genetic analysis for identifying missing persons. The computational complexity of the algorithms is explained, and several ways to reduce the complexity are introduced.

Validation of software for calculating the likelihood ratio for parentage and kinship

Although the likelihood ratio is a well-known statistical technique, commercial off-the-shelf (COTS) software products for its calculation are not sufficiently validated to suit general requirements for the competence of testing and calibration laboratories (EN/ISO/IEC 17025:2005 norm) per se. The software in question can be considered critical as it directly weighs the forensic evidence allowing judges to decide on guilt or innocence or to identify person or kin (i.e.: in mass fatalities). For these reasons, accredited laboratories shall validate likelihood ratio software in accordance with the above norm. To validate software for calculating the likelihood ratio in parentage/kinship scenarios I assessed available vendors, chose two programs (Paternity Index and familias) for testing, and finally validated them using tests derived from elaboration of the available guidelines for the field of forensics, biomedicine, and software engineering. MS Excel calculation using known likelihood ratio formulas or peer-reviewed results of difficult paternity cases were used as a reference. Using seven testing cases, it was found that both programs satisfied the requirements for basic paternity cases. However, only a combination of two software programs fulfills the criteria needed for our purpose in the whole spectrum of functions under validation with the exceptions of providing algebraic formulas in cases of mutation and/or silent allele.

Estimating mutation rates from paternity casework

We present a statistical methodology for making inferences about mutation rates from paternity casework. This takes account of a number of sources of potential bias, including hidden mutation, incomplete family triplets, uncertain paternity status and differing maternal and paternal mutation rates, while allowing a wide variety of mutation models. An object-oriented Bayesian network is used to facilitate computation of the likelihood function for the mutation parameters. This can process either full or summary genotypic information, both from complete putative father-mother-child triplets and from defective cases where only the child and one of its parents are observed. We use a dataset from paternity casework to illustrate the effects on inferences about mutation parameters of various types of biases and the mutation model assumed. In particular, we show that there can be relevant information in cases of unconfirmed paternity, and that excluding these, as has generally been done, can lead to biased conclusions.

Combining Autosomal and Y-Chromosomal Short Tandem Repeat Data in Paternity Testing with Male Child: Methods and Application

Paternity testing is being increasingly requested with the aim of challenging presumptive fatherhood. The ability to establish the biological father is usually based on the genotyping of autosomal short tandem repeat (STR) in alleged father, mother and child, but the use of Y-chromosomal STR has gained interest in the last few years. In this work, we propose a new probabilistic approach that combines autosomal and Y-chromosomal STR data in paternity testing with father/son pairs taking into account mutation events. We also suggest a new two-stage approach where we first type Y-STRs and possibly autosomal STR for the putative father and son, conditional on Y-STR results. We applied this approach to 22 cases. Our results show that Y-STRs can identify nonpaternity cases with high accuracy but need to be validated with autosomal STR to establish paternity. Moreover, the two-stage approach is less costly than the standard approach and is very useful in motherless cases.

Use of X-linked short tandem repeat loci in routine parentage casework

BACKGROUND

Dealing with genetic inconsistencies in parentage testing, especially in motherless cases, remains a continual difficulty.

STUDY DESIGN AND METHODS

Four difficult cases, comprising two trios and two duos, were selected from routine parentage testing casework. In these, relatively low combined paternity indices were observed as a result of few discrepant loci that were treated as being due to paternal mutations. An additional eight short tandem repeat (STR) loci along the X chromosome were studied in the alleged father and female child to try and help resolve these cases.

RESULTS

In all four cases, the X chromosome haplotypes in the alleged father were different from those in the child, showing decisively that the alleged father could be excluded from being the biologic father of the child.

CONCLUSION

In recent times the study of X chromosome haplotypes has been shown to be useful in parentage testing where the alleged father is absent and where only his close relatives are available for testing. This work demonstrates that such studies can also prove valuable in the testing of standard trios and duos in cases where there only a few genetic inconsistencies amongst the loci tested, making it difficult to distinguish between paternal mutations and a close relative of the alleged father being the biologic father.

Structured Incorporation of Prior Information in Relationship Identification Problems

The objective of this paper is to show how various sources of information can be modelled and integrated to address relationship identification problems. Applications come from areas as diverse as evolution and conservation research, genealogical research in human, plant and animal populations, and forensic problems including paternity cases, identification following disasters, family reunions and immigration issues. We propose assigning a prior probability distribution to the sample space of pedigrees, calculating the likelihood based on DNA data using available software and posterior probabilities using Bayes' Theorem. Our emphasis here is on the modelling of this prior information in a formal and consistent manner. We introduce the distinction between local and global prior information, whereby local information usually applies to particular components of the pedigree and global prior information refers to more general features. When it is difficult to decide on a prior distribution, robustness to various choices should be studied. When suitable prior information is not available, a flat prior can be used which will then correspond to a strict likelihood approach. In practice, prior information is often considered for these problems, but in a generally ad hoc manner. This paper offers a consistent alternative. We emphasise that many practical problems can be addressed using freely available software.

Object-oriented Bayesian networks for complex forensic DNA profiling problems

We describe a flexible computational toolkit, based on object-oriented Bayesian networks, that can be used to model and solve a wide variety of complex problems of relationship testing using DNA profiles. In particular this can account for such complicating features as missing individuals, mutation and null alleles. We illustrate the use of this toolkit with several examples, including disputed paternity with missing or additional measurements, and criminal identification. We investigate the effects on likelihood ratios of introducing mutation and/or null alleles, and show that this can be substantial even when the underlying perturbations are small.

Multistep microsatellite mutation in the maternally transmitted locus D13S317: a case of maternal allele mismatch in the child

Examination of a case of a paternity dispute with 17 autosomal short tandem repeats (STR) loci revealed a mismatch of the maternally transmitted allele at the locus D13S317 in the questioned child. The composition of the alleles of this locus in the mother, questioned child and suspected father was 8/8, 11/11 and 8/11, respectively. The sequence analysis of the regions flanking the locus D13S317 and peak height measurements of the paternal, maternal and child alleles at this locus excluded the possibility of null allele as a cause of the allelic mismatch inherited by the child. The results suggested expansion of the microsatellite repeat motif, TATC by three repeat units as a probable cause for the allelic mismatch in the child. This is a rare case of maternally transmitted multistep microsatellite mutation reported for the first time for this locus in the forensic DNA analysis. The mutation rate at D13S317 locus in maternal and paternal meiosis was 0.04 and 0.14%, respectively, and overall mutation rate was 0.15%. The probability of maternity and paternity were 0.999999 and 0.999999, respectively, for all the 17 autosomal STR loci analyzed. Furthermore, the sequence of two hypervariable regions of mitochondrial DNA, HV1 and HV2 and the maternal alleles of six X chromosome STR loci in the questioned child matched completely with the mother. These results conclusively proved that the mother and suspected father are the biological parents of the questioned child.

The expected performance of single nucleotide polymorphism loci in paternity testing

We discuss the utility of single nucleotide polymorphism loci for full trio and mother-unavailable paternity testing cases, in the presence of population substructure and relatedness of putative and actual fathers. We focus primarily on the expected number of loci required to gain specified probabilities of mismatches, and report the expected proportion of paternity indices greater than three threshold values for these loci.

Calculating the exclusion probability and paternity index for X-chromosomal loci in the presence of substructure

There has been recent interest in the use of X-chromosomal loci for forensic and relatedness testing casework, with many authors developing new X-linked short tandem repeat (STR) loci suitable for forensic use. Here we present formulae for two key quantities in paternity testing, the average probability of exclusion and the paternity index, which are suitable for X-chromosomal loci in the presence of population substructure.

Y-chromosomal microsatellite mutation rates: differences in mutation rate between and within loci

Precise estimates of mutation rates at Y-chromosomal microsatellite STR (short tandem repeat) loci make an important basis for paternity diagnostics and dating of Y chromosome lineage origins. There are indications of considerable locus mutation rate variability between (inter-) and within (intra-) loci. We have studied nine Y-STR loci-DYS19, DYS389I/II, DYS390, DYS391, DYS392, DYS393, DYS385, and DYS388-in 1,766 father-son pairs of confirmed paternity (a total of 15,894 meioses). Five biallelic markers were also analyzed in the fathers-Tat, YAP, 12f2, SRY1532, and 92R7-defining haplogroups 1, 2, 3, 4, 9, and 16, respectively. A total of 36 fragment length mutations were observed: 24 gains (22 single-step, two double-step) and 12 single-step losses. Thus, there was a significant surplus of gains (p=0.045). Overall, the mutation rate was positively correlated to STR repeat length and there was a significant relative excess of losses in long alleles and gains in short alleles (p=0.043). In contrast to the situation in autosomal STR loci and in MSY-1, no noteworthy correlation between mutation rate and the father's age at the child's birth was observed. We observed significant interlocus differences in Y-STR mutation rates (p<0.01). The number of observed mutations ranged from zero in DYS392 to eight in DYS391 and DYS390. We have also demonstrated obvious differences in mutation rates between the haplogroups studied (p=0.024), a phenomenon that is a reflection of the dependence of mutation rate on allele size. Our study has thus demonstrated the necessity of not only locus-specific, but even allele-specific, mutation rate estimates for forensic and population genetic purposes, and provides a considerable basis for such estimates.

FINEX: a Probabilistic Expert System for forensic identification

A series of recent papers have shown how to formulate complex problems of forensic DNA identification inference, such as occur in disputed paternity or criminal identification cases, in terms of Probabilistic Expert Systems (PESs). However, at the present time, general purpose PES software is not particularly well suited to the repetitive tasks of: specifying an appropriate set of marker networks for a specific problem; for editing the many local conditional probability tables; and combining evidence from several genetic markers to evaluate likelihoods. Here, I describe a user-friendly prototype software tool called FINEX developed both to automate such tasks and also to evaluate likelihoods of interest. Ease of use is achieved by a graphical specification language that enables a user to quickly specify a range of forensic DNA problems. I describe the algorithms by which FINEX converts the user input in the graphical specification language and data on observed markers to the Bayesian networks used in PES.

The mutational spectrum of human autosomal tetranucleotide microsatellites

We studied by multiplex amplification and single-run electrophoretic analysis 10 microsatellite loci, composed of nine tetranucleotide-repeats (D1S1612, D3S2387, D4S2431, D5S2501, D10S1237, D15S657, D16S2622, D18S1270, and IFNAR-ALU) and one trinucleotide repeat (D2S1353). After elimination of proven null allele events involving D1S1612 and D5S2501 and of all data of D3S2387, in which we suspected but could not prove the occurrence of null alleles, we were left with nine loci, encompassing 24,224 meioses and 23 mutations. Twenty-two of the mutations (96%) were single-step events. Moreover, 18 of the mutations were paternal, four were maternal, and one was indeterminate. There was no significant difference between the number of additions and deletions in the mutants. Our findings are compatible with a simple model in which tetranucleotide microsatellites mutate primarily in paternal germinative cells by DNA slippage, such that the vast majority of mutations are equiprobable additions or deletions of a single-repeat unit. By combining the data from our tetranucleotide loci with literature information of highly and lowly mutable microsatellites, we observed a very highly significant correlation between mutation rate and the geometric mean of the length of the longest perfect repeat region (LRPR), compatible with a power or exponential relationship. The variation of the length of the LRPR explained as much as 80% of the variance of the mutation rate of autosomal tetranucleotide microsatellites.