Hi,
I have used plink (--assoc option) to detect the level of association between a list of snps and two quantitative traits ( list_snps~QT1, list_snps~QT2). For quality control, I have included two snps (rs123_star, rs456_star) that have been previously genotyped with TaqMan. (So if everything went well in the analysis, rs123 from the list should have around the same significance levels as rs123_star; and similarly for rs456 and rs456_star)
At a preliminary look at the top significant SNPs (sorted by F value) for each association I could observe :
Table 1 : snps~QT1 : rs123, rs456 missing from top 40
Rank id QT1_F_value QT2_F_value
1. rs123_star 3285,58 1535,86
6. rs456_star 802,02 19,27
Table 2 : snps~QT2 : rs456, rs456_star are missing from top 40; additionally the differences between the F values are extreme for rs123 and rs123_star.
Rank id QT2_F_value QT1_F_value
1. rs123_star 1535,856 3285,582
12 rs123 470,0224 12,36226
What could be the explanation of such a huge difference in significance levels between the same SNPs ?
(the genotypes for rs123 and rs456 resulted after a GWAS (chip data); the genotypes for rs123_star and rs456_star were obtained using Taqman genotyping on the same population)