dosage=DS or dosage=HDS?

[风筝]

Hello,

I am running GWAS using plink2 for HRC imputed data (Minimac4). I notice in PLINK2 website, it says “to import Minimac4-style DS+HDS phased dosage, add 'dosage=HDS'. 'dosage=DS' (or anything else for now) causes the named field to be interpreted as a Minimac3-style dosage". I do find the glm results are slightly different between using "dosage=DS" and using "dosage=HDS".

So what is the difference between DS and HDS? Do I need to use dosage=HDS instead of dosage=DS in PLINK2? 

Thank you very much!

The code I use:

plink2 \

--vcf chr22.dose.vcf.gz dosage=DS \

--id-delim \

--keep pheno_pooled_sample.txt \

--missing variant-only \

--freq \

--hardy \

--pheno pheno_pooled_invBP.txt \

--pheno-name inv_SBP inv_DBP inv_PP inv_MAP \

--glm omit-ref \

--out GWAS_results/BP_pooled_chr22

[Christopher Chang]

The difference is that DS is unphased and HDS is phased.  However, --glm does not care about genotype/dosage phase; the only reason I'd expect dosage=HDS results to actually differ from dosage=DS results is rounding error.

[Khoa Ho]

Hi,

I'm Khoa who also facces the same confusion as 风筝 here. I wonder if setting "dosage=HDS"  actually includes DS in or just HDS values alone? If no DS is imported then I guess one should set "dosage=DS" when planning to use it for glm right? 

[Chris Chang]

The documentation explicitly says “DS+HDS”, there isn’t really anything else to say.

You can always look at the top of a re-exported VCF (“—export vcf vcf-dosage=HDS”) if you’re unsure what was imported.

--
You received this message because you are subscribed to the Google Groups "plink2-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to plink2-users...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/plink2-users/23968e28-9196-4ac2-810a-358d45e4c7a2n%40googlegroups.com.