search

qlu-lab / PUMAS

Fine-tuning polygenic risk score models using GWAS summary statistics
MIT License
41 stars 11 forks source link

PRS by selected variants #12

Open kys21207 opened 1 year ago

kys21207 commented 1 year ago

Hi,

I am wondering whether I can find an optimal PRS model based on pre-selected variants of interest using PUMAS.
If so, could tell me what steps I need? Thank you so much,

KJ

qlu-lab commented 1 year ago

Hi KJ,

Yes, you can do that in two ways: (1) Only keep those variants of interest in the GWAS summary statistics and use this reduced sumstats for subsampling and PRS evaluation. Doing this is computationally efficient if you are only interested in a small subset of variants (e.g., LD-pruned SNPs or SNPs from a chromosome). (2) Use the full summary statistics for subsampling but only keep those variants of interest in the SNP weights file (i.e., --weight_path) when evaluating PRS accuracy. This approach is more generic and should be applicable to any subset of variants.

I hope this is helpful and feel free to let me know if you have other questions/comments.

Zijie