The probability distribution of the number of loci indicating exclusion in a core set of STR markers

The distribution of the number of loci out of the 13 in the CODIS STR set that would show an exclusion (i.e., a genotype set incompatible either with the prosecution hypothesis or with Mendelian transmission) was estimated in different scenarios. The knowledge of this distribution would provide a framework against which casework evidence can be compared. I used allele frequencies in Iberian and in Italian populations to generate individual genotypes at random and to test in 1 million simulation replicates, how many of the 13 loci would give an exclusion in an individual identification case, a paternity case, and a double parenthood case. All three scenarios were tested under an expected overall exclusion, both for unrelated individuals and for cases in which the suspect or the alleged father was the brother of the real culprit or real father. Paternity and double parenthood cases were also tested in the true scenario, with exclusionary loci due to mutation. In individual identification cases, the average number of exclusionary loci was 11.95 with a minimum of 7. This STR set also showed sufficient power to resolve identification cases in which the evidence sample came from a suspect's sib. False paternity cases yielded an average of 7.65 exclusionary loci and exclusions with only one (0.0108%) or two (0.14%) exclusionary loci were obtained only rarely. The cases of exclusion with one locus could lead to likelihood ratios in favour of paternity, while both true and false paternity cases with two exclusionary loci would often lead to non-conclusive likelihood ratios. The average number of exclusionary loci in a paternity case where the alleged father was the real father's brother was 3.82, with a significant number of cases where no exclusions were obtained.