High estimated likelihood ratio might be insufficient in a DNA-lead process of identification of war victims

With the advance in typing tools and extraction procedures in recent years, DNA analysis has developed into an amazingly powerful method for forensic analysis. For a number of years, autosomal STR (Short Tandem Repeat) typing has been used as a tool in the process of identification of war victims in Croatia. Although DNA typing is very effective in detecting possible identities of exhumed skeletal remains, this approach bears some risk of false identification. The paper presents the case of a match between skeletal remains and the son and wife of a missing person in 13 STR loci. Even though these skeletal remains also matched in 13 loci the mother of the same missing person, additional genetic testing (Y-chromosome and mitochondrial DNA) unequivocally excluded the proposed identity. Although likelihood ratio is the best measure of the significance of a genetic match between exhumed skeletal remains and relatives of the missing person, the meaning of likelihood ratio is not as clear in database matching as in simple paternity cases and great care is needed to avoid wrong interpretation. To reduce the risk of possible false identifications, in addition to DNA evidence, other types of evidence (such as information about the time, place and other conditions of disappearance), as well as on anthropological and other »classical« forensic data are being used as a »control mechanism« in the DNA-lead process.