Reporting of highly individual genetic typing results: a practical approach

This paper considers the interpretation of serological typing data as a problem in forensic science, as opposed to a problem in population genetics or statistics. Controversies arising in this area are partly due to an overly narrow perspective that ignores basic forensic science principles. After an initial discussion of the special problem that deoxyribonucleic acid (DNA) blood typing poses to forensic science, the three difficulties common to all the proposed interpretive methods are discussed. These are: predicting genotype incidence from allele frequencies, predicting frequencies for the joint occurrence of genotypes in a number of different genetic marker systems, and determining the appropriate population to use to measure the frequencies. The inability to test assumptions that are inherent in our routine methods is noted. This is a procedural weakness that unnecessarily limits the admissibility of DNA typing evidence in court. A practical solution to this problem is offered that begins with minimal assumptions. Initially a statement is made based on (1) how many reference samples the laboratory has typed and (2) how many of these samples show genotypes corresponding to the case samples. The second stage of the presentation begins with a statement that additional assumptions are necessary to fully interpret the evidence and that although these assumptions are scientifically very reasonable, they cannot be absolutely proven. The presentation can then proceed, if desired, to consideration of the specific assumptions and frequency estimates of any of the methods that have been proposed to date. To follow this approach population data must be kept in a form that allows the simple first-stage statement to be made. This means that each individual's record would include typing results in each genetic marker system. Although this method of data storage differs from that used in most forensic science laboratories, it is exceptionally versatile, and allows great flexibility in data analysis.