Statistical and genetic aspects of quality control for DNA identification

Quality control for likelihood ratio (LR) tests on restriction fragment length polymorphisms (RFLPs) and polymerase chain reaction (PCR) markers is provided by the 4N6 program. Examples are given for test of exclusion, coincidence, and kinship that should be used routinely in validation of DNA profiling. They confirm that "in a knowledgeable court DNA profiling is no longer exposed to risk of illogical presentation, blind acceptance or arbitrary rejection" (N. E. Morton, Eur. J. Hum. Genet. 1993, 1, 172-179). However, they remain to be implemented in national systems of quality control. There can be no effective quality control until attempts to fudge gene frequencies in order to retain the Hardy-Weinberg assumption (recommended by the National Research Council) ar abandoned in favour of the population genetic approach that takes gene frequencies from a real population and uses kinship to express genotype probabilities. Since an expert witness cannot know the identity of the culprit, he is obliged to consider multiple hypothesis without usurping the prerogative of the court to favour one of them on the basis of other evidence. It remains to be seen whether the new NRC Committee (and more importantly the international community of forensic scientists) will adopt this fundamental principle, which is especially important in populations like India with preferential endogamy and consanguineous marriage. Alternative procedures and the causes, effects, and remedies of four types of error are also considered.