search

statgen / METAL

Meta-analysis of genomewide association scans
Other
44 stars 12 forks source link

Allele flipping for all SNPs? #19

Open LindsayMMB opened 2 years ago

LindsayMMB commented 2 years ago

Hi,

I've checked both of my input files at least 10 times and for some reason the A1 and A2 alleles get flipped in the METAL output files.

This isn't a big deal to reverse but I was wondering if A2 was originally the affected allele, will the effect estimate and p-value METAL gives me still apply?

In other words, does the allele order given in the output matter in terms of the calculations given?

Many thanks in advance.

welchr commented 2 years ago

The short answer is the allele order in the output does not matter. METAL flips effect sizes towards a consensus effect allele as each study is read in. The consensus or final effect allele is usually the effect allele encountered in the first study file read in, but there are cases where even this is not necessarily true due to an internal allele sorting, see issue https://github.com/statgen/METAL/issues/9.

michaelofrancis commented 2 years ago

I can't imagine there are any new METAL users that do not waste several days trying to figure this out (myself included). Why not fix such a large and confusing problem?

welchr commented 2 years ago

You could try emailing the original authors to see if they would be willing to devote time/funding to it. (I am not the maintainer or author of METAL, just a long time user of the program and I work here in the department.)