This is a test version of Biostars. For the public version, visit https://www.biostars.org.
Why does GWAS Catalog mix OR and beta in the same column?

All associations in GWAS Catalog can be downloaded here: https://www.ebi.ac.uk/gwas/docs/file-downloads

However, the Odds Ratios (ORs) from cases/controls studies and the betas from continuous phenotypes are mixed in a single column called "OR_or_beta". Why would they do that? ORs and betas are naturally different; the OR is multiplicative, and the "no effect" OR is 1, while the beta is additive, and the "no effect" beta is 0. The right values to compare would be betas and log(OR).

In contrast, if you visit a particular association on their web, the OR is clearly not mixed with the beta column. Or if you query their RESTful API for one particular association, you get the betaNum and the orPerCopyNum as different entries. Since I need to work with the full dataset, I find this annoying.

Any clue on why they mix ORs and betas? And how could I tell them appart in the analysis?

gwas gwas_catalog or
• 2,009 views • link
updated by zhoufeng2ye • zhoufeng2ye • written by Juan Manuel Berros Juan Manuel Berros

1 answer

I was converting all the Ors to Betas recently from GWAS-catalog too, and found this really annoying. just wondering if you have found any efficient methods to handle the conversion.

• 0 views • link
written by zhoufeng2ye • zhoufeng2ye

Log in to answer this question.

More posts like this