Akaike's information criterion for a measure of linkage disequilibrium

 The extent and distribution of linkage disequilibrium (LD) in humans is a current topic especially for gene mapping of complex diseases. Akaike's information criterion (AIC) was applied to estimate LD and compared with other standard LD measures, D′ and r ². By comparison of an independent model (IM; linkage equilibrium) and a dependent model (DM; linkage disequilibrium), the parsimonious model is the one with the smaller AIC score. Therefore, the extent of LD by AIC is expressed as AIC(IM) — AIC(DM)(AIC(LD)). A total of 39 single-nucleotide polymorphisms on a 1.6-Mb region of chromosome 21q22 were identified, and genotyped in 192 Japanese individuals. All possible pairs were analyzed to estimate LD and the analyses were compared. AIC(LD) became highly positive as the D′ value increased and was negative at D′ values of around 0.2. Because a negative value of AIC(LD) implies linkage equilibrium, D′ values below 0.2 should be regarded as linkage equilibrium. The LD estimate by AIC yielded results similar to those obtained by r ², indicating that AIC(LD) would be useful for fine gene mapping. Received: July 3, 2002 / Accepted: October 2, 2002