SUM vs AVG --score flag

[Emanuele F. Osimo]

Dear Shing et al,

I am looking for clarifications on the --score flag.

In the docs, you say that --score avg - Take the average effect size is the default PRSice behaviour. However, in you seminal Choi et al, A guide to performing Polygenic Risk Score analyses, 2020 paper you describe the methods as

In the standard approach, polygenic risk scores are calculated by computing the sum of risk alleles corresponding to a phenotype

So it seems sum should be the default?

What is the exact difference between avg and sum?

Is one of them just adding up ORs for each SNP? What does the other do please?

BW

Emanuele

[Emanuele F. Osimo]

More specifically, I seem to have found that changing the ORs for each SNP and using --score sum, the resulting PRSs don't change. Is that expected behaviour? I thought the sum of the ORs should change, and therefore the total PRS...

[Sam Choi]

Sorry, that's a wording problem.

If you look at the manual (https://choishingwan.github.io/PRSice/step_by_step/#prs-calculation) , the difference between AVG and SUM is whether we divide the PRS with the number of alleles included, instead of calculating the average effect size. AVG has the benefit of accounting for missingness though in most case, with proper QC, the AVG and SUM score should be more or less the same, as such the result should not change.

Hope this help

Sam