Item

Abstract

Although considerable progress has been made in developing statistical methods for modeling population structure confounding in GWAS, these methods are not perfect, and power will inevitably be reduced if genome-wide correlations between loci are too strong. The traditional alternative is using a controlled cross between two genotypes in which unlinked loci are effectively randomized with respect to each other. This has the disadvantage that resolution is much poorer than GWAS, and allele frequencies are distorted. To get around this problem, we generated a mapping population that kept the natural allele frequencies, but decreased the amount of linkage disequilibrium. We did this by carrying out well over 1000 crosses between parents chosen at random from natural lines of Arabidopsis thaliana, and propagated each cross to the F9 generation. We are using genotype data to compare the genetic structure of the parents and F9s. We compare the performance of GWAS in these two cohorts to assess the effect of the population structure correction on the identification of genotype- phenotype associations for three traits related to growth and reproduction.