Power for mapping by admixture linkgage disequilibrium: a case/control simulation study. C.L Pfaff, M.D Shriver. Penn State University, University Park, PA.
Admixture between genetically distinct populations creates association between loci. After a few generations, the linkage disequilibrium between unlinked loci decays, and the remaining disequilibrium can be used to map genes (Mapping by Admixture Linkage Disequilibrium). In order to characterize the behavior of MALD, we designed a simulation program that determines the power for MALD depending on a set of parameter values (e.g. number of generations since admixture, proportion of admixture, recombination fraction (q), sample size, and marker and disease allele frequencies). This program examines the power of a case/control study to detect linkage disequilibrium between a marker and a disease allele. Our results show that microsatellite markers are, on average, more powerful than SNPs for MALD. The average power for a SNP marker is 56% (q=0.01, N=300 case, 300 control), whereas an average microsatellite marker achieves >90% power under the same conditions. This difference is mainly due to higher dC levels (the sum of all positive allele frequency differentials at a locus) at microsatellite loci than at SNP loci. We have found that a high dC value is the most important characteristic of a good marker for MALD. For example, a marker with a dC value of 0.71 achieves a power of almost 90% when q=0.06, whereas a marker with a dC value of 0.38 attains a power of only 57% at the same q. Similarly, markers with high dC levels demonstrate powers of 89% with only 5% admixture (q=0.001) whereas markers with lower dC values have powers of 48% under the same conditions. We estimate the average dC value for microsatellites to be 0.42. Values of dC above 0.20 generate power estimations >85% at q<0.01. Thus, for candidate gene mapping, most microsatellites and many SNPs will provide suitable power. However, for genome-scan mapping a panel of high d markers must be chosen to achieve sufficient power. The results of these simulations indicate that MALD can and should be an important technique for mapping complex disease genes, and has more statistical power than other mapping methods. The results of current efforts comparing the power of MALD to the power of affected sib-pair mapping will also be discussed.
