Missing data in covariate file in PLINK

Question

[landscape95]

Hi everyone, I am doing the summary statistics for SNP association in Plink. However, the program says I have missing values even though I have checked the phenotype information, it doesn't have -9/0/NA values. When I checked the covariate file, some values had -9/0 values, these values are the true values, they are not annotated as missing, may I ask how can I flag another value for the covariate file only? I don't want it effect my phenotype missing values which can be -9/0/NA in my future work. Your help is really appreciated!

Answer 1

[karl.stamm]

PLINK v1.9 has a parameter --missing-phenotype which lets you assign a number other than -9 as the key for missing value. see their docs for encoding parameters https://www.cog-genomics.org/plink/1.9/input#pheno

[landscape95]

Thank you very much! I have read this document but according to it, the --missing-phenotype will be applied across all phenotype-related data, the .fam file will also be considered the same. I wish that I could assign the missing phenotype differently between the .bim file and the covariate file. Is there any way that I can do it? Thank you!

[karl.stamm]

I don't know what your .bim is, but you should be able to recode any file, or just run the software separately for each condition. I gather PLINK is just running one column of the phenotype file anyway, you'll be doing separate analyses. You're trying to do several analyses right now with incompatible annotation files. Re-generate the BIM to match the phenotypes, or vice versa.