Applying polygenic scores to data

Question

[jared]

I have a polygenic score file from the PGS catalog and I would like to apply this to my dataset (PLINK format) and calculate risk for my samples. I have some experience using PRSice but I believe the process differs since I have a generate PGS where weights are already calculated for each SNP. What would be the easiest way to do this/is there a tutorial or thread out there (I spent a lot of time, but I can't seem to find anything in my case). The PGS file has columns: rsID, chr_name, chr_position, effect_weight, effect_allele, reference_allele, locus_name.

Answer 1

[Sam]

You can directly use the weight file. PRSice does not modify your weight, so the result will still be valid.

If you are concern about the p-thresholding and clumping, you can simply do --no-clump with --bar-levels 1 --fastscore. This will use all the SNPs to calculate the PRS.

[Simon]

I would like to calculate the PRS for a bunch of samples in a bgen file. As is the case in this thread, I just have the files from the PGS Catalog. Could you be more specific about how to use PRSice with these files from the PGS catalog ? The problem right now is that PRSice needs P-Values but these are not provided by the pgs catalog. I just have the already calculated weights. Thanks in advance.

[Sam]

you can ask a new question on this.

In short, from the PGS catalog file, add a column called P-value, fill it with a random number between 0 and 1 (doesn't matter what it is). Then you can use --no-clump --fastscore --bar-levels 1 to run PRSice to get the PRS you want.