A New Powerful Method For Genome-wide Association Mapping Using Local Genealogies In A Mixed Model

Local genealogies are genealogies for the longest chromosome region around a marker that do not require recurrent mutation or recombination (Mailund et al. 2006). A local genealogybased genome-wide association mapping approach, Blossoc, was introduced by Mailund et al. (2006) for a case-control study and extended by Besenbacher et al. (2009) for complex traits. Ledur et al. (2009) used Blossoc for QTL mapping using the dataset simulated for QTLMAS-XII workshop by Lund et al. (2009). Blossoc was found to perform best out of several approaches used to analyze the common dataset (Crooks et al. 2009). However, the method developed can be applied only for unrelated samples which are not always available in human genetics and generally unavailable in livestock populations. Ledur et al. (2009) corrected the data for the pedigree and then analyzed as if it was a population sample of unrelated individuals. Precorrecting the phenotypes for pedigree, however, is not an optimal solution because it reduces power to identify QTL (Aulchenko et al. 2007). Therefore, a method which can utilize the genealogy information and simultaneously account for multiple levels of relatedness among individuals is expected to have more power when the samples actually come from a complex pedigree. Yu et al. (2006) present a unified mixed model method for association mapping that accounts for multiple levels of relatedness. The mixed model approach has the added advantage of being flexible, as it can be applied to both family-based and population samples. We present here a new method, GENMIX (genealogy based mixed model), which uses local genealogies in a variance component based model for association mapping for complex traits, thereby combining a genealogy-based approach with a mixed model.