Erroneous calculations of weight of DNA evidence may lead to miscarriage of justice in Pakistan

Dear Editor,

The correct weightage of DNA evidence can be calculated by knowing the behavior of DNA marker in the relevant population. Allele frequencies, the prevalence of micro variants in the population, the pattern of inheritance, and the difference of allele frequencies among various subpopulations residing at a particular geographical area are the important factors for evaluating the weight of evidence. Inbreeding, population substructures and migration increase genetic relatedness between individuals (Steele and Balding 2014). In the presence of the abovementioned scenario, a genetically related individual already sharing more alleles compared to the unrelated individuals becomes genotypically more related and chances of inaccurate calculation of random match probability increase many folds. In Pakistan, there are two functional forensic laboratories, which are performing forensic DNA analysis in civil and criminal cases. One is the National Forensic Science Agency (NFSA), Islamabad, while the other is the Punjab Forensic Science Agency (PFSA), Lahore. It has passed nearly 9 years since their establishment, but no population database of allele frequencies of STR markers has been developed by either of the laboratories. NFSA is not currently doing any statistical calculations for forensic casework (Mateen et al. 2018). The forensic report generated by DNA department of NFSA contains allelic match between the profile obtained from evidence and profile(s) obtained from the victim and/or suspect without any statistical calculation of estimating the probability of a match. PFSA calculates “Random Match Probability” for forensic casework and relationship testing. In the absence of local population database of STR loci, PFSA is using Caucasian STR allele frequencies of US-FBI database, USA, to calculate match probabilities. Diverse subpopulations dwelling across the country are genotypically and phenotypically different from each other. Allele frequencies of few populations residing in Pakistan are reported by researchers, but allele frequencies of all population substructures are not yet elucidated. Allele frequencies of Balochi, Bruhi, Burusho, Kalash, Makrani, Pathan, Sindhi, Saraiki, and Gujjar have already been reported and are available at (http://spsmart.cesga.es/search.php?dataSet=strs_local) (Jilani et al. 2016; Shan et al. 2016), while some other subpopulations, e.g., Kalash, Hindkowans, Shina, Kalyu, Balti, and Kashmiri, are not explored yet for determining allele frequencies in Pakistan. A strict caste/clan system prevails in the country where marriages occur only within the family or within the same caste. It seldom happens that one marries to a person who does not belong to one’s caste. In such cases, few alleles become more prevalent in a particular sub-population and few become rare. When match probabilities are calculated using Caucasian allele frequencies from the US-FBI database, chances of error increase in finding accurate statistical results. This could be worse when two related individuals such as full siblings or half-siblings or even cousins are involved in a crime or when a mixed DNA profile of three or more than three contributors is obtained from an evidence (Lohmueller and Rudin 2013; Toscanini et al. 2012). In estimating relatedness in the case of a missing person, legitimacy of a child, or property disputes, application of an unrelated dataset of allele frequencies would result in erroneous estimations.

No consolidated population DNA database containing allele frequencies of STR loci for Pakistani population is available for forensic casework and relationship testing. As such, the government has not provided funding to any institution for the establishment of a national population DNA database. There are various international databases available online for calculating allele frequencies for STR loci. STRbASE (http://strbase.org) is developed by the European Network of Forensic Science Institutes (ENFSI) for calculating random match probabilities in European populations. These calculations are made using 16 STR loci (Gill et al. 2006; Parson and Roewer 2010). STRBase (http://www.cstl.nist.govstrbase) developed by the National Institute of Standards and Technology (NIST) is a comprehensive database, which provides information frequencies of common and rare alleles. This also uses up to 16 loci and calculations are made using Omnipop software which is linked with this database (Ruitberg et al. 2001). ALFRED, the ALlele FREquency Database (https://alfred.med. yale.edu/), is a large allele frequency database with a sample size of N > 660,000 individuals. This fairly large database is used to calculate allele frequencies of populations residing in Europe and outside Europe (Rajeevan et al. 2011). Similarly, pop. STR (http://spsmart.cesga.es/popstr.php), PopAffiliator (http://cracs.fc.up.pt/_nf/popaffiliator2), and ALLST*R (http://allstr.de) are online available databases of allele frequencies of STR loci for statistical calculations (Amigo et al. 2009; Bodner et al. 2016; Pereira et al. 2011). But these databases are not a true depiction of allelic frequencies of all subpopulations residing in Pakistan.

The allele frequencies calculated for European and American populations are different from those for population substructures living in Asia due to evolutionary and cultural history (Venables et al. 2016). The available STR databases developed in other jurisdictions have different genetic characteristics compared to the genetic makeup of Pakistani populations. Although, it is recommended that such calculations can be used by applying a correction to avoid variation in sub-population (Holsinger and Weir 2009), if these frequencies do not represent the true allele frequencies of other population, such corrections would be all in vain. Therefore, STR database containing all population substructures need to be established for calculating the correct weight of DNA evidence. This database should have an appropriate size and representation of each sub-population having F_ST < 0.05. The subpopulations having F_ST < 0.05 can be grouped into one population (Council 1996). This database can be used by all national forensic labs to estimate match probabilities in forensic casework and relationship testing.

Moreover, Thermo Fisher Scientific and Promega have marketed STR kits having more than 21 STR loci. Allele frequencies for Asians are available with GlobalFiler™ PCR Amplification Kit (Thermo Fisher Scientific, South San Francisco, CA, USA) and PowerPlex® 21 System (Promega, Madison, WI, USA). There are two issues in these allele frequencies, i.e., first, the sample size is not large enough, and, second, frequencies of few alleles in many loci are not given. In Pakistan, forensic laboratories are considering to use these kits for forensic case work, paternity, and kinship analysis. The old databases do not have allele frequencies for newly introduced STR loci, so there is a need for calculating allele frequencies for new STR loci added in these systems.

It is already reported that while establishing paternity, chances of error is about 3% even when proper allele frequencies are applied for calculating the paternity index (Green and Mortera 2017; Karlsson et al. 2007). Forensic DNA analysis requires flawless statistical calculation for random match probability, likelihood ratio, and relationship testing excluding close relatives among donors. If allele frequencies do not truly represent a population sub-structure, the chance of erroneous inclusion increases to a great degree.

Lawyers and judiciary have little or no knowledge about the importance of calculating the probabilities and determining the weightage of forensic DNA evidence in a case. What they understand is the statement in a report that is “Match” or “Not Match”, and they are unable to understand the statements related to the probability of matching any other individuals having the same profile in the relevant population. For a better understanding of forensic DNA analysis reports, lawyers and judges must be educated on why and how match probabilities are used in these reports. In Pakistan, only few institutions are offering degree programs in Forensic Biology or Forensic Genetics, in which such forensic calculations are part of the curriculum. Few institutions/universities, i.e., CAMB, University of Punjab, Lahore; DoC, GC College University, Lahore; IBBT, University of Animal and Veterinary Science, Lahore; DLS, University of Management and Technology, Lahore; and University of Health Sciences, Lahore, are offering degree programs in Forensic Sciences. All degree programs are approved and accredited by the Higher Education Commission, Islamabad. All institutions, except GC College University, offer programs of Forensic Biology. GC College University offers a graduate program in Forensic Chemistry. It also offers a 1-year MS program for lawyers, judges, and personnel from law enforcing agencies and general public. Without understanding the exact weightage of DNA evidence, it is difficult to conduct a fair trial which may result in a miscarriage of justice. In the future, if these cases would be reanalyzed for statistical calculations using true allele frequencies, there are chances of finding a fair number of wrongful convictions across the country, based on forensic DNA reports. There is a great need to develop a national population DNA database for all subpopulations residing within the country.