Ancestral Reconstruction for Association Mapping

One of the greatest promises of modern genetics is the ability to identify loci responsible for, or at least highly correlated with, important physical traits and disease susceptibility. This will enable benefits ranging from improved disease risk assessment and crop design to ultimately a better understanding of how genotype variation effects variations in phenotypes. A key component in this task is association mapping where correlations between the clustering in the trait of interest and the clustering observed in the genealogy of a set of sampled sequences has the potential to localise causative genomic locations with relatively high precission. Association mapping is commonly based on single nucleotide polymorphism (SNP) data, where individuals have been typed on a number of marker loci where two different nucleotides are present with reasonably high frequency in a population. However, more complex data including multi-allelic loci and insertiondeletion variation, in particular microsatellite repeat count variation can also be utilised. The human HapMap project International HapMap Consortium [2005] identified and continues to refine a dense set of SNP marker locations, and similar data is becoming available for other organisms. The availability of high density chips harnessing this information Barrett and Cardon [2006] allows genome-wide scans to be performed in large sets of individuals, and with the constant decrease in cost and increase in fidelity of sequencing and chip technologies it is to be expected that high density association mapping on large data sets becomes standard for the study of within-population phenotypic variation. The two fundamental features of SNP data, and similar types of population variation data, that makes association mapping feasible is a relatively low mutation rate and the presence of recombinations in the evolutionary history of the data across the region covered. The low mutation rate causes both variation in the trait of interest and the type of a marker locus to be clustered into one or at most a few clades in the local genealogy of a sample. Recombination causes the genealogy to