Hi all,

I have a question about the direction of effect in polygenic risk scores. An illustrative example:

GWAS data:

snp                 maj/min                OR
rs11884770        C/T             0.98

In my VCF file (same genome build as the above):

snp                   ref/alt               
rs11884770        T/C

So in this case, the reference allele is the "effect" allele (T), and that effect is negative. My naive understanding is that I should simple count the number of Ts in this case, and then weight this in the negative direction by multiplying by log(0.98) = -0.0202.

But I read somewhere else that it's correct to count the Cs (as this is the "risk" allele in this case) and invert the direction of effect. Which is correct? I do note that many scores in prscatalog report plenty of negative weightings, which suggests that my original idea is right?

Many apologies if this is confusing - I have confused myself thoroughly also.