PRSice polygenic score with only controls

Hi,

I'm trying to calculate polygenic scores using PRSice; my dataset contains only controls (PLINK-coded as "-9"). When running PRSice, I get the following message:

Error in if (levels(as.factor(prof$PHENO))[1] == "1" & levels(as.factor(prof$PHENO))[2] ==  :
  missing value where TRUE/FALSE needed
Execution halted

In principle, this happens because PRSice is trying to get different levels for the phenotype in order to do a regression and compare cases vs controls. But even when specifying "no.regression T" or "no.regression T report.individual.scores T", the program gives the same output. Has anyone experienced/solved this before?

A brute-force alternative could be to insert "random" phenotypes (namely, to use both "1"s and "2"s for every pheno), and just neglect the regression outputs. But this is not quite orthodox...

Any help is appreciated. Thanks,

snp polygenic prsice

• 5,162 views • link

updated 10.0 years ago by coleman_jonathan

• written 10.0 years ago by aldoc •

1 answer

This a known bug - no regression still requires a non-missing phenotype. I recommend you run PRSice with a dummy phenotype file as you describe and the no regression option on

• 1 views • link

written 10.0 years ago by coleman_jonathan

Thanks! I tried it with two different "random phenotype" files, and gives the same results. Just for future reference, the whole command was:

R --file=PRSice_v1.25.R -q --args plink PRSice_v1.25/plink_1.9_linux_160914 base referencefile.txt target myplinkfile slower 0 sinc 0.01 supper 0.5 report.best.score.only F no.regression T report.individual.scores T covary F pheno.file randomphenotypes.txt

The final output (after automatic cleanup) was two files: PROFILES.nosex and PRSice_SCORES_AT_ALL_THRESHOLDS.txt.

Best,

• 1 views • link

written 10.0 years ago by aldoc •

Hi,

Even I am also getting same error. But, after giving phenotype file, this is the error now, Error in family$linkfun(mustart) : Argument mu must be a nonempty numeric vector Calls: summary ... glm -> eval -> eval -> glm.fit -> <anonymous> -> .Call Execution halted

• 0 views • link

written 7.3 years ago by sofie_carolina •