Evaluation of the allele-sharing approach, known as the IBS method, in kinship analysis

To infer relatedness from genetic data based on short tandem repeats, the exact method, in which shared allele frequencies are applied to relevant equations, has been conventionally used. An alternative approach is the IBS method that is based on the number of shared alleles between individuals. In the present study, the performance of the IBS method in pairwise kinship analysis was compared with the exact method using simulated data of 10,000 genotype pairs for 15 loci in the ABI Identifiler system. The likelihood ratio in allele-sharing of zero, one and two was calculated from joint probabilities based on allele frequencies of the Japanese population. Whereas the IBS method generally produced lower values of combined indices, smaller deviations of the distributions were evident. The threshold for identification of full siblings relative to non-relatives was comparable with that of the exact method, indicating that both inference powers were almost identical. The likelihood ratio in the IBS method depends on the heterozygosity at a locus, and heterozygosities of the 15 loci were consistent across various population groups, particularly in East Asians. The convenience of fixed LR values in the IBS method is beneficial for cases with uncertain allele frequencies and rare alleles.