Extracting genotype count for case-control study

[N]

I'm using PLINK to find SNP that are associated with case-control study.

I used the --assoc flag, and I found some interesting SNPs.

But, it gives allele frequencies, and want to get genotype count (#homozygous for a1, #hom. for a2 and #hetrozygous) for case and control separately.

Is it possible?

--freqx does print genotype count, but this is for case an control together and not separately.

[Christopher Chang]

Try --freqx + --filter-cases/--filter-controls.

[N]

Thanks

[N]

Another issue:

Can I make it works with case control from an alternate phenotype file?

Because now, I see it detect cases and control only from the dummy phenotype it has in binaries.

I tried  --pheno, but with no success

[Christopher Chang]

Strange, --pheno should work with --freqx + --filter-{cases,controls}.  Can you post the .log file from your failing run?

[N]

Here is the log file

[Christopher Chang]

Can you clarify what the problem is?  The .log file implies that the case/control phenotypes from the --pheno file are being used; is this not actually happening?

[N]

What I see is that --pheno works for a given phenotype (e.g, with the flag --pheno), but not with the flag , --all-pheno which is expected to generate a feqx file for each phenotype. 

[Christopher Chang]

Ah.  That's because --all-pheno currently only affects association analysis commands like --linear; you'll need to write a short shell script to emulate it for other commands (the --mpheno flag will be helpful here).

[N]

OK, I'll do that

[Josef Frank]

How about using the --hardy option?

This provides genotype counts separately for cases and controls as well as for the pooled sample