PRSice: how to obtain exact list of SNPs used for PRSice.best

Question

[m98]

I have run a standard PRSice analysis as follows:

PRSice.R \\
--prsice /usr/local/bin/PRSice_linux \\
--base !{base} \\
--snp SNPID \\
--chr CHR \\
--bp POS \\
--A1 Allele1 \\
--A2 Allele2 \\
--stat BETA \\
--pvalue p.value \\
--beta \\
--target !{name}_chr#_filtered \\
--binary-target !{params.binary_trait} \\
--pheno !{pheno} \\
--cov !{cov} \\
--thread !{task.cpus} \\
--clump-kb !{params.clump_kb} \\
--clump-r2 !{params.clump_r2} \\
--clump-p !{params.clump_p} \\
--no-clump !{no_clump} \\
--missing !{params.missing} \\
--ld-hard-thres !{params.ld_hard_thres} \\
--model !{params.model} \\
--score !{params.score} \\
--quantile !{params.quantile} \\

My question is: how can I find out which exact SNPs were included in PRSice.best? When I look in PRSice.summary, I can the number of SNPs (Num_SNP) that were included but not the list of the SNPs. Is there any way to obtain this information?

Thanks

Answer 1

[Sam]

Use --print-snp which will generate a output prefix.snp file. This file contain all the post-clump SNPs and their p-value. You can then filter out the SNPs included in the best score by filtering according to the p-value threshold

[paulo.saquilayan]

Are you able to take the lowest P value SNPs from the .SNP file to draw conclusions about the model or will I have to use PRSet in conjunction to be able to interpret it.

[Sam]

PRSet is a completely different module which is for set based PRS. To extract the SNPs in the best score, take the p-value threshold from the summary file and filter the .snp file accordingly (that'd be similar for PRSet, though we don't generally recommend running PRSet with thresholding)

[paulo.saquilayan]

Thank you again! Appreciate all the work you do and your continued maintenance of PRSice.

[mano]

Hi, I have run PRSice2 for my own data and got output of around 288424 SNPs after clumping. From the summary file and print-snp file, I tried to extract significant SNPs based on the P value mentioned in it but, I didn't get the exact same number of SNPs mentioned in the summary file. That is the number of SNPs mentioned in the summary file was around 7954 but when I filtered SNPs with P value it was around, 450087, Am I doing any mistake? Please help me extract only significant SNPs for my trait PRS. Thanks in advance.

[Sam]

Will need the log and more information on how you get those SNPs count for me to better answer this.