Having PLINK 1.9 to assume that non-called variants are equal to REF

[Elielson Veloso]

Hello!!

I was wondering if I could use PLINK 1.9 to call missing genotypes using a 200 SNP list (hotspot list). So , I want all the 200SNP hotspot list that were not called (and thus, are not present in the VCF files I used to merge and to do the association analysis) to be considered as =REF.  Is it possible?

I appreciate your help!

Elielson Veloso

[Zuxi Cui]

I don't get your motivation for doing so. However, you can do it by "--recode 12" and replacing missing with 2 (usually major allele) in the output file.

Terry

--
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 on the web visit https://groups.google.com/d/msgid/plink2-users/c3e1574f-8e98-4264-8791-5c840d50bba9n%40googlegroups.com.

[Zuxi Cui]

I suggest you check the positions of those SNPs and see if they are close to each other or spread all over the genome.

If they are close enough to each other, you may try a regional imputation. Or, a gene-based analysis like SKAT-O.

Otherwise, if they spread all over the genome, I don't have a good idea to increase the power of your analysis.

Terry

On Wed, May 17, 2023 at 6:58 PM Elielson Veloso <eliel...@gmail.com> wrote:

Hello, Zuxi!! Thanks for you answer!!

 Let me try to explain my question better... I am trying to do an association analysis from data coming from a panel of 200 SNPs (I know it is very unlikely that I will find any strong association when working with such an small number of SNPs).. however, since these SNPs were chosen because of their polimorfic nature, I was wondering if it would be reasonable to include reference alelles in the association analysis. Would you have any suggestion on what is the right way to follow in order to get the best from this analysis? Would a polygenic risk approach be the best option instead of association analysis by logistic regression?

Thanks in advance!