DINGO: increasing the power of locus discovery in maternal and fetal genome-wide association studies of perinatal traits.

Four of these SNPs exhibited significant ( p -value < 0.05) discordant maternal and fetal effects in the conditional analyses– most notably rs7034200 at GLIS3 , a known type 2 diabetes locus, that was highly significant in the two-degree-of-freedom test ( p -value 2df = 1.15 × 10 −12 ), but only exhibited relatively modest non-significant p -values in the maternal and fetal meta-analyses (minimum p -value meta = 7.39 × 10 −5 ), and likewise rs560887 at G6PC2 , a known locus for fasting glucose which also exhibited radically different p -values across the two tests ( p -value 2df = 4.37 × 10 −20 , minimum p -value meta = 6.25 × 10 −7 ).

also in 82,756 other papers

However, whilst all loci that met genome-wide significance using the simple meta-analytic strategy were also significant or almost significant using DINGO (i.e. all loci would be flagged by researchers as significant or suggestive), many genome-wide significant loci that displayed directionally discordant effects in the two-degree-of-freedom DINGO test exhibited p -values that were far less extreme using the simple meta-analytic strategy, including some loci that may have been missed entirely (e.g. the GLIS3 locus discussed below).

also in 1,004 other papers