PRSice2 p-value thresholds with --no-regress option

[Allie Lake]

Hello,

I am using PRSice2 to simply generate PRS for each individual, rather than perform a regression for a particular phenotype. I am unclear on how to interpret the output in the .all.score file after running the program with the --no-regress option. It seems that although the default p-value thresholding parameters were used, the output only contains a single PRS column (yet the first line of the file displays each p-value from 0.0001 to 1), so I'm unclear as to which threshold was used to generate this column, and why PRSs are not being generated for all the thresholds.

Here are the parameters I used:

Rscript $PRSice --dir . \

--prsice ${shome}/downloads/PRSice_linux \

--base ${home}/data/PGC_SCZ2/ckqny.scz2snpres.noChr \

--target ${mega_data} \

--no-regress \

--chr hg19chrc \

--bp bp \

--snp snpid \

--A1 a1 \

--A2 a2 \

--stat or \

--pvalue p \

--out output/PGC_SCZ2 \

--clump-kb 250 \

--clump-r2 0.1 \

--clump-p 1 \

--maf 0.1 \

--perm 10000 \

--thread 1 \

--print-snp

Thanks!

Allie

[Sam Choi]

It sounds like a bug or the program was terminated early. Mind sending me the full log?

[Allie Lake]

I'm attaching the log file. It appears that it ran successfully.

Thanks,

Allie

[Sam Choi]

Strange, I cannot reproduce the problem with my toy data. Mind sending me the head of the .all.score output?

[Sam Choi]

btw, forgot to mention that you don't need to use --perm when you are using --no-regress (i should put a warning) as no association was performed

[Allie Lake]

My subsequent runs have not reproduced this error. I'm not sure what happened, but thank you for looking into this.