group of 66.000 (17%) markers showing a significant difference in the magnitude of the association sign between empiric and imputed measures, these elements could potentially influence subsequent GWAS analysis. The same comparison was carried using the QQ plots graphical representation and, as expected, the use of this common filtering criterion was successful in reducing part of the spurious results (Additional file 1, Figure S1A). Both graphical representations highlight the effectiveness of quality and MAF filters for purging markers with low concordance between association statistics based on imputed and directly genotyped allelic frequencies. The term concordance used in this report refers to the discovery of markers showing evidence of association using a pre defined significance threshold, a concordant marker would be considered associated by their directly genotyped and imputed allelic frequencies. Nevertheless, several markers surpassing this quality control showed considerable bias between association statistics determined by the allelic frequencies.
