I noticed that the variant 5_40669906_G_A has two effect sizes for Liu et al. (GCST003044):
Based on de Lange et al., it seems that the odds ratio should be greater than one. I’m not sure if other variants have the same problem for Liu et al.
Based on de Lange et al., it seems that the odds ratio should be greater than one. I’m not sure if other variants have the same problem for Liu et al.
Hi @cdeboever, welcome to the Open Targets Community! 
Within the GWAS Catalog, associations may appear multiple times for the same study but different cohort experiments (as described in the GWAS Catalog FAQ).
However, in the example you gave, it looks like the number of case and controls for both variants is the same, so we are investigating it further. Thanks for letting us know!
Thanks @hcornu, yes the case/control numbers are the same, and it’s also odd that the beta values are the same except one is negative. I appreciate the you and the team looking into it!
The variant also has the same issue for GCST003043:
