SNP in that region. The use of selected SNPs to cover an arbitrarily sized genomic region does not produce a robust genetic map [9], and some of the problems with genetic association studies result from the incorrect representation of the pattern of genetic variation [8]. We have found that some regions of the mouse genome have very low levels of linkage disequilibrium between SNPs [10], while other regions can have very high rates of polymorphism among the inbred strains [11] with frequent changes in the pattern of genetic variation, which reduces the utility of representative SNPs in these regions. Since the genetic variation within a region must be fully analyzed to know where there is a change in the genetic pattern, the poor performance of GWAS using representative SNPs is not surprising [2, 3].
