Evaluation of the reliability of DNA typing in the process of identification of war victims in Croatia

Abstract

Aiming to estimate the frequency of various types of errors that can occur in the large-scale process of identification, we identified and compared genotypes of 911 parent-child pairs in the database of 3498 relatives of people that disappeared during the 1991/1992 war in Croatia. Genotypes of 891 pairs (97.8%) were matching, while 20 pairs did not match in one or more loci. Reanalysis of these samples revealed that out of 1822 analyzed genotypes, one genotype was completely wrong, and two genotypes had one wrong allele because of human errors. Five genotypes had a single wrong allele due to either polymerase chain reaction or electrophoresis errors. In five genotypes mutations were the cause of mismatch. Genetic inconsistencies with parentage were found in four "fathers" (4.2%) and three "mothers" (0.36%). As the majority of observed single-locus errors were caused by nonhuman errors, all databases produced with similar technology would probably have comparable level of errors.