Introduction
The discovery of polymorphisms in repetitive DNA has had a tremendous impact on Forensic Science. In the last ten years, highly informative and robust DNA typing systems have been developed which have proved to be very effective in the individualization of biological material of human origin.
DNA typing has advantages over traditional protein analysis. First it is more informative and can be analyzed in minute or degraded material, as DNA is physically much more resistant to degradation than proteins. Secondly, the same DNA genotype can be obtained from any tissue (i.e. blood, saliva, semen, hair, skin, bones), whereas the analysis of protein markers is restricted to cells where these proteins are expressed.
DNA analysis has become standard method in forensic genetics as it is currently applied by most labs for most of the forensic genetic types of expertise, especially in criminal forensic casework (stain analysis and hairs), identification and paternity testing.
PCR-based DNA typing systems have made it possible to analyze DNA obtained from only a few cells as well as from highly degraded human samples or remains, as has been recently demonstrated by the identification of relatively old human remains. An example is the identification of the remains of the Romanov family. The potential of DNA typing has also made possible the resolution of immigration problems or complicated paternity testing cases when the father is not available. Rapid identification of individuals in mass disaster using DNA typing has also been possible. Computerized DNA databases for the identification of criminal offenders have been created in some countries.
Due to all these impressive applications, the media have taken great interest in DNA profiling, the term firmly establishing itself in the everyday language of the man in the street, mainly because of the value of the evidence presented through DNA profiling in certain well-known legal cases.
Initially, the use of DNA profiling was very controversial in some countries, perhaps due to a hasty introduction of this new methodology. However this has ironically contributed to a much more reliable use of DNA profiling.
Two parallel upheavals concerning the introduction of DNA typing technology have accounted for this reliability: the introduction of quality control and accreditation schemes and in particular the spreading use of the statistics in the evaluation of DNA evidence. Although these two aspects began with conventional serological evidence, their importance nowadays is due to the great potential of this technology and the enormous value of the evidence that DNA profiling usually offers.
To understand the significance and value of DNA evidence, it is necessary to analyze the final consequences of the analysis. In some cases one DNA profile will not match another profile or a hypothesis, this event is called an exclusion, or that particular profile may match another or it may match a hypothesis (i.e. a parenthood relationship); in this case there is said to be a match. But how sure can these statements be? How infallible is an exclusion? And, is the DNA in the case of a match as good as a fingerprint? In addition, how can judges, lawyers, juries and the public in general be sure that a particular laboratory properly performs DNA analysis?
The aim in this article is to briefly address these questions.
The Value of an Exclusion Criminal casework
When the DNA profile of a sample found at the scene of the crime does not match with the DNA profile of the suspect it can be concluded that both samples have a different biological origin. In most cases short tandem repeats (STRs) are used for DNA typing and we can expect to obtain the same DNA profile from samples with the same biological origin even if different human tissues are used for comparison.
There are rare cases, particularly some human tumors (especially colorectal cancer) in which micro-satellite instability can arise, so special care needs to be taken when using tissue cancer samples for identification purposes.
In general the ‘a priori’ discrimination power of the STR used in casework is very high. With only a commercially available standard multiplex (Fig. 1) the discrimination power in Caucasians is higher than 0.99999. Most forensic genetic laboratories have a battery of DNA polymorphisms (including SLPs and STRs) of more than 15, so the theoretical discrimination power is enormous.
Generally, in polymerase chain reaction (PCR) based systems, reproducibility and standardization are made easier by using automated fragment separation and detection on automated sequencers and by applying a standardized system-specific sequenced allelic ladder as an internal measurement standard for allele designation. The use of sequenced allelic ladders is crucial for a robust typing.
The STR systems selected by any individual laboratory for casework should preferably be one of those currently used by forensic genetic laboratories and should have been previously validated. Laboratories and companies for the validation of individual STRs and kits have often followed SWGDAM (Scientific Working Group DNA Analysis Methods) validation schemes. In Europe during the last decade, EDNAP (European DNA Profiling Group) has performed an important work of validation of DNA polymorphisms from classical SLPs (single locus probes) analysis to STRs and mitochondrial DNA (mtDNA). For nomenclature the ISFH (International Society for Forensic Haemogenetics) recommendations are complied with by all the laboratories in the field.
Paternity testing
The mutation rate of DNA polymorphisms including both SLPs and STRs are being estimated. For STRs the mutation rates of different loci can differ by several orders of magnitude and even different alleles at one locus can exhibit different mutation rates. Factors affecting the mutation rate include the number of uninterrupted repeats, the size and the complexity of the STRs. Mutation events in the male germ line are five to six times more frequent for STRs than in the female germ line. The average mutation rate in the STRs can be estimated at around 1.2 x 10 ~3.
Figure 1 One of the most commonly used multiplexes (Profiler plus, PE) in forensic genetic laboratories. It contains nine STRs plus amelogenine. The discrimination power of this multiplex in Caucasians is 0.9999.
In general, isolated exclusions by a single marker are obviously not considered as conclusive and the probability of paternity is then calculated excluding the marker having the possible mutational event. In general exclusions are considered in the report as conclusive when the exclusion number is more than three. In these cases the probability of a triple mutation has a chance on average below 10-8. When we are more knowledgeable about the mutation rate of individual markers, it will then be possible to include in a report the theoretical error in case of exclusion.
mtDNA
In the case of mtDNA when the patterns do not match by a single base pair (bp), the possibility of a mutation or heteroplasmy should be considered. The hetero-plasmy and mutation rate depends on the methodology used. With very sensitive methodologies such as SSCP (single strand conformation polymorphism) or SSO (sequence specific oligonucleotides), the rate can be quite high. With sequencing, although the mutation rate is controversial, the possibility of a muta-tional event should be considered. In addition, there are sequence positions that have a higher probability of differing between known maternal relatives. There is also a handful of nucleotide positions that show a propensity for rapid substitution or the presence of heteroplasmy. One example of this is the position 16093, where more than 20 examples of substitution or heteroplasmy have been observed. Thus, if the only difference between two maternal relatives is at position 16093, the results should be at least considered inconclusive, and the data may be strong enough to give a verbal opinion in favor of a common biological origin if the accompanying polymorphisms are uncommon.
For all these reasons, very cautious verbal opinions should be given in the case of differences existing in one base, and the possibility that even in this case the samples can have the same origin should be considered. With differences in more than 3 bp (using sequencing) it seems highly likely that the samples have a different biological origin, but, again, cautious verbal opinions should be given and guidelines should be followed with the appropriate flexibility to account for the unique genetic characteristics of mtDNA.
The Value of a Match Criminal casework
When two DNA profiles match, then the suspect is not excluded from being a contributor. In this case, it is necessary to attach some numerical weight to the evidence of a match, and statistical issues arise. The appropriate weighting for evidence is by means of a likelihood ratio. If E is the evidence of matching DNA profiles, and C (‘the stain was left by the suspect’) and C (‘the stain was left by an unknown man’) are alternative explanations for that evidence, then the relative merits of the two explanations can be compared by the ratio of the probabilities of the evidence under each hypothesis. It is then necessary to consider the probability of the evidence given in each of these hypotheses and the ratio of these two probabilities -the likelihood ratio – can be in the simplest cases to reduce to 1/f, where f is the frequency of the observed genotype among members of the population to which the ‘unknown man’ credibly belongs. This must be estimated using data from a sample of people from the relevant population.
A STR profile is a multilocus genotype and probabilities of these profiles are estimated most simply as the products of the individual allele probabilities. The implied assumption of allelic independence can be tested with exact tests. At single loci, the independence assumption can be avoided by expressions that allow for the effects of population structure.
Nowadays, most experts use likelihood ratios for weighting the value of the evidence and for communicating this value to the courtroom. Frequencist approaches are prone to fallacies especially to the so-called prosecutor fallacy or the transposed conditional. Training of experts and judges on these concepts is necessary to prevent misconceptions when the evidence is presented in the courtroom.
There is a general agreement that the Bayesian approach to inference provides a logical and coherent framework for interpreting forensic transfer evidence. The interpretation of mixed stains is possible only in the context of likelihood ratios. Unlike single-contributor stains, the sample profile may not be certain under either of the two alternative propositions, so the likelihood ratio is the ratio of two probabilities that are less than one. Presenting the probability under only one proposition can be misleading.
Because of the enormous power of the evidence on many occasions provided by DNA analysis, the question often asked about a DNA profile is: Is it as good as a fingerprint? This apparently simple question has been addressed as follows:
For non-DNA evidence (i.e. a fingerprint) individualization depends on a leap of faith that, in Bayesian terms is equivalent to saying: ‘My personal likelihood ratio is so large that, no matter how small the prior odds are, the posterior odds are large enough to individualize with certainty’. For DNA evidence such a state could only be reached by testing more and more loci, but the apparent objectivity of numerical statements then becomes increasingly illusory, and the element of personal belief totally dominates the data. (Evett and Weir 1998).
However, if the likelihood ratio is enormous, expressions such as ‘this profile is unique’ or the mere introduction of the idea of uniqueness should be avoided and a numerical value given, otherwise the expert will be adopting the role of the judge.
Paternity testing
Parentage testing and identification of remains both exploit the genetic laws of the transmission of alleles from parent to child. As with forensic applications, DNA evidence is interpreted using likelihood ratios that compare the probabilities of the evidence under alternative propositions. In a normal paternity case the two propositions are as follows. H1: The alleged father is the father of the child; H2: some other man is the father of the child.
In general, high likelihood ratios are usually attained by DNA analysis in normal and even, but not always, in complicated cases. In complex paternity cases such as those associated with incest or paternity calculations when relatives but not the father are available, the correct consideration of the two alternative hypotheses is crucial.
Unlike the evaluation of the evidence in criminal casework, it is quite common for many experts to present the value of the evidence in terms of Bayesian probability using ‘a priori values’ of 0.5. This is clearly wrong, since, at least in legal cases, the a priori value should be fixed by the judge and not by the expert.
Ychromosome polymorphisms and mtDNA
To evaluate the weight of the evidence in cases of matching using Y chromosome polymorphisms and mtDNA two problems arise. Firstly, paternally connected (using Y chromosome polymorphisms) or maternally connected individuals (in the case of mtDNA) will share identical haplotypes and therefore, cannot be discriminated. Secondly, the evaluation of evidence in terms of likelihood ratio presents some problems given that gene frequencies cannot be estimated using Hardy-Weinberg laws. In addition, population substructuring seems to be more severe in the case of Y chromosome markers than for unlinked autosomal markers. Due to this, large databases including as many populations and individuals as possible are necessary for a more exact assessment. Taking these difficulties into account experts are, in general, giving verbal opinions in cases of matching with mtDNA and Y STRs instead of numerical values.
Communication of the value of the evidence in DNA analysis
Communication is closely linked to, and is as important as, interpretation. In general the most crucial factor in the communication of DNA evidence is not the explanation of the scientific techniques used and the explanation of the DNA results, but the communication of the value of the evidence. There are three main elements of an effective communication.
1. In all except the simplest cases, the interpretation of the scientific results depends on the circumstances, so it is necessary to detail these in the statement.
2. To assess the strength of the evidence it is necessary to consider at least two explanations for its occurrence. The evidence is evaluated by assessing its probability under each alternative hypothesis.
3. The strength of the evidence in relation to one of the explanations is the probability of the evidence given the explanation, divided by the probability of the evidence given the alternative hypothesis.
Some experts are in favor of giving verbal predicates in addition to the numerical values of likelihood ratios. Many laboratories use verbal predicates in paternity testing and verbal equivalents for LR. However, language is imprecise and the verbal predicates are merely a convention. More importantly, the expert is in some way taking the place of the judge when verbal predicates are used. The use of this approach may be understandable when juries are involved, for improving the comprehension of people who are unused to thinking numerically, but it seems an inappropriate approach for legal systems with professional judges and not juries.
Progress in Standards
If DNA analysis is nowadays accepted in all countries all over the world, it is in part due to the progress made in standardization.
Standardization of forensic DNA analysis has made enormous progress in the last few years and this innovation in standardization is comparable to the introduction of DNA technology itself.
Standards are crucial for forensic geneticists. This is due to the fact that only with an agreement about standards is it possible to develop quality control programs and quality assurance programs. In other words, standards are the only way to guarantee to judges, juries and the public that the tests performed and laboratory efficiency are reliable in any specific case. In addition, standards are necessary to allow for second opinions, to interchange data between labs and to create uniform searching procedures in cross border crime.
Two types of standards need to be addressed: technical and procedural. Technical standards include matters such as the genetic systems to be used (including type, nomenclature and methodology), the statistical methods for evaluating the evidence and the communication of the final report. Procedural standards encompass matters of operation, such as laboratory accreditation, laboratory performance, accreditation and licensing of personnel, record keeping and proficiency testing.
In the United States and in some European countries development of procedural standards for forensic genetics laboratories has made considerable progress in the last few years. In some of these countries laboratories have agreed on the requirements necessary for organization and management, personnel, facilities and security, evidence control, validation, analytical procedures, equipment calibration and maintenance, proficiency testing, corrective actions and audits. Proficiency testing programs for DNA analysis are established in some countries, and external and internal controls have been set up by most of the labs in western countries. Progress in accreditation has been effective in many countries in the last few years.
Even more advances have been made in attaining common technical standards. Agreement on genetic systems, types and nomenclature is widespread.
Establishing common standards in forensic DNA analysis is not easy due to the fact that there are very different legal systems and a variety of laboratories performing forensic genetic analysis. The success of the forensic geneticist in achieving common standards (at least compared with other aspects of forensic science and genetics) has been greatly facilitated by the ISFH, which has many national and international working groups, particularly EDNAP, actively involved in establishing common standards. In addition, the existence of commercially available kits for DNA typing, a shared aim of geneticists in the search for common standards and the influence of some leading groups, are other reasons for this success.
However, the efforts in standardization should continue. Progress in common procedural standards and particularly progress in similar requirements between countries for accreditation are necessary. With regard to technical standards other priorities include the harmonization of criminal databases, the coordination and compilation of population databases (especially for mtDNA and Y STRs) and a continuation of progress initiated in the last few years on statistical evaluation and communication of the value of evidence provided by DNA analysis.
Conclusions
In the last ten years, highly informative and robust DNA typing systems have been developed which have proved to be very effective in the individualization of biological material of human origin.
DNA profiling has revolutionized forensic genetics and is widely accepted in legal cases. To make this possible, progress in standardization has been a crucial factor. Advances in establishing common procedural and technical standards have been considerable in the last ten years. Proficiency testing programs for DNA analysis have been established in some countries, and most laboratories have implemented external and internal controls. Progress made in accreditation has been effective in many countries in the last few years. Agreement on genetic systems, types and nomenclature is widespread.
In most cases, but not always, the value of the evidence provided by DNA analysis is enormous. However, uncertainty always exists. As scientists, we must measure this uncertainty and for this we use a standard: the probability. Likelihood ratios are nowadays used for weighting the value of the evidence and for communicating this value to the courtroom and the Bayesian approach to inference provides a coherent framework for interpretation.
