Informativity assessment for biallelic single nucleotide polymorphisms

Common single nucleotide polymorphisms (SNPs) have the potential to provide a widely used means of simple and robust kinship testing. Suitable measures of polymorphism informativity are therefore required in order to guide the search for the most efficient combinations of SNPs. In the context of kinship testing, such measures should preferably be related to Z, the power of excluding false paternity in trios comprising mother, child and alleged father. Since the bulk of SNPs is expected to be biallelic, a Z-related measure of informativity can be defined for SNPs in a particularly elegant manner: allele frequency vectors of sets of n biallelic SNPs that give rise to the same Z value approximate to an n-dimensional sphere around (1/2,...,1/2). Owing to this relationship, it can be shown that the number N of maximally informative SNPs (i.e., of SNPs with allele frequencies 1/2), providing the same Z value as a given set of n SNPs, approximates to 2n times the average gene diversity of the latter. Linear regression analysis of a large number of simulated SNP sets reveals that only a minor linear correction of Nis required for large n. Since Z= 1-(13/16)N, Ncan also be calculated easily for multiallelic markers with known Z. The "equivalent number of maximally informative SNPs", N, is therefore suggested as a measure of marker informativity in the context of kinship testing.