choishingwan
commented
4 months ago Use —extract to limit the analysis to specific snps, and use —no-clump to avoid clumping. Might also want to use fastscore and bar-level to avoid thresholding. Though at this point, it might be better to just use plink to calculate the score
Sam
On Mon, May 13, 2024 at 2:49 AM Shawndockercool @.***> wrote:
i want to perform PRS calculation and i want to make sure the same set of SNPs are used every time, so that the result is replicable every time for each individual.
below is the code:
Rscript PRSice.R --dir /usr/local/lib/ --prsice PRSice_linux --base /mnt/BrainHealth/WGS/base/grit.hg38.base --target /mnt/BrainHealth/WGS/plink-209_2023/brain209new.qc --out 209 --print-snp --thread 20 --stat BETA --beta --extract /mnt/BrainHealth/WGS/plink-723_20240402/brain.VQSR.PASS.new.qc.bim --no-regress --no-clump
i use --no-regress since our task is to calculate PRS without phennotype. --extract to limit the SNPs to the QCed ones, --no-clump to minimize differnt SNPs due to target file
Target file is evetything that changes. yet it seems that still, PRSice are using different set of SNPs.
Could anyone give any suggestions, as for how to keep SNPs(thus, the model) the same, to make results replicable on an individual level?
— Reply to this email directly, view it on GitHub https://github.com/choishingwan/PRSice/issues/357, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAJTRYQWJ5U2SORLJOZAQK3ZCBO6TAVCNFSM6AAAAABHTSWRGGVHI2DSMVQWIX3LMV43ASLTON2WKOZSGI4TCOJWHAZTGNA . You are receiving this because you are subscribed to this thread.Message ID: @.***>
i want to perform PRS calculation and i want to make sure the same set of SNPs are used every time, so that the result is replicable every time for each individual.
below is the code:
Rscript PRSice.R --dir /usr/local/lib/ --prsice PRSice_linux --base /mnt/BrainHealth/WGS/base/grit.hg38.base --target /mnt/BrainHealth/WGS/plink-209_2023/brain209new.qc --out 209 --print-snp --thread 20 --stat BETA --beta --extract /mnt/BrainHealth/WGS/plink-723_20240402/brain.VQSR.PASS.new.qc.bim --no-regress --no-clump
i use --no-regress since our task is to calculate PRS without phennotype. --extract to limit the SNPs to the QCed ones, --no-clump to minimize differnt SNPs due to target file
Target file is evetything that changes. yet it seems that still, PRSice are using different set of SNPs.
Could anyone give any suggestions, as for how to keep SNPs(thus, the model) the same, to make results replicable on an individual level?