Reverse quantile plot

[giuliapontali]

Hi, I'm calculating the PRS for a binary trait. I'm using a GWAS composed of ~8000 samples and as target data ~2000 samples. In the target data, we have the 323 participants affected (1) and 2187 that are not affected(0). Looking at the quantile plot it seems that is reverse. What do you suggest doing?

The code that I using is the following:

/data/shared/bioinf/R/bin/Rscript-4.0-BioC3.12 PRSice.R --prsice ./PRSice_linux --A1 Allele1 --A2 Allele2 --base /GWAS/SAIGE_binary_association.txt --binary-target T --ignore-fid --beta --stat BETA --pheno /pheno/muo.txt --pheno-col metNotHlty --pvalue p.value --snp SNPID --target /genotypingData/10k_imputedData --keep /pheno/subset.txt --nonfounders --quantile 10 --out /prs_muo

[choishingwan]

It is possible, especially if your base and target are for different phenotype or if there is a negative correlation between the phenotype measurement in base and in target.

On Mon, May 17, 2021 at 3:13 PM jPontix @.***> wrote:

Hi, I'm calculating the PRS for a binary trait. I'm using a GWAS composed of ~8000 samples and as target data ~2000 samples. In the target data, we have the 323 participants affected (1) and 2187 that are not affected(0). Looking at the quantile plot it seems that is reverse. What do you suggest doing?

The code that I using is the following:

/data/shared/bioinf/R/bin/Rscript-4.0-BioC3.12 PRSice.R --prsice ./PRSice_linux --A1 Allele1 --A2 Allele2 --base /GWAS/SAIGE_binary_association.txt --binary-target T --ignore-fid --beta --stat BETA --pheno /pheno/muo.txt --pheno-col metNotHlty --pvalue p.value --snp SNPID --target /genotypingData/10k_imputedData --keep /pheno/subset.txt --nonfounders --quantile 10 --out /prs_muo

[image: prs_sr1_QUANTILES_PLOT_2021-05-13] https://user-images.githubusercontent.com/45068286/118543363-7d2a6b80-b754-11eb-99f3-b2937269bc89.png

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/choishingwan/PRSice/issues/267, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAJTRYQ4DKXRU36WLRH2VQLTOFTELANCNFSM45BBMQHQ .

[giuliapontali]

The base and target data are based on the same binary phenotype

[choishingwan]

The encoding of case control are identical in both?

What's the association R2 and P-value?

On Wed, May 19, 2021 at 9:31 AM jPontix @.***> wrote:

The base and target data are based on the same binary phenotype

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/choishingwan/PRSice/issues/267#issuecomment-844109313, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAJTRYX2NZ6HC7WI2RNS2OTTOO4ULANCNFSM45BBMQHQ .

[giuliapontali]

Yes, I built my gwas with the same encoding that I have in the target data.

PRS.R2= 0.0352465 P=1.65997e-12

[choishingwan]

That is strange.

Which version of PRSice are you using? Also, can you check the genetic correlation between your base and target samples (if you have enough sample size).

[giuliapontali]

I'm using the PRSice 2.3.3 (2020-08-05). Ok, I'm going to check the correlation between the base (7378 samples) and the target (2509 samples) samples. Thanks

[stale[bot]]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[anastazie-dnanexus]

I have the same issue and solved it by reversing --a1 and --a2 as following: --a1 A2 \ --a2 A1 \ I have GWAS results from regenie, soA1 is Ref and A2 is Alt.