multiallelic variants -glm and --adjust

[sy06...@gmail.com]

Hi Chris, 

I was wondering how does --glm deals with the multialleleic variants with the format below ?

what format does the multialleleic variant should be in order to be accepted?  

Is there a better option than using  --max-alleles [ct] (Exclude variants with more than the given # of alleles. )    to get ride of the   *  ( =  all other alleles)  ? 

If i simply split this into 2 variants  ACC> AC and ACC>A  ,  with this VCF file,  i will now have  for subjects in this rows with the same genotype  (e.g 0/1)  for the 2 variants  ( e.g subject 1  who is 0/1 for ACC > AC,A,*     will not be 0/1 for ACC> AC and ACC>A  . Which is incorrect) .

And related to this, how does --adjust will deal with these positions ?  will it consider them as separate or unique "test" ?  

I would very much appreciate any advice  and  guidance, 

cheers, 

Saliha

An example fromVCF file 

CHROM	POS	ID	REF	ALT	QUAL	FILTER	INFO
chr1	10146	.	ACC	AC,A,*	26103.67	.	AC=340,2,5;AF=0.199,0.001,0.003;AN=1710…..

[Christopher Chang]

You shouldn't need to change the format of this multiallelic variant for plink2 to accept it.

If you want to exclude '*', --max-alleles doesn't actually do what you want, since the '*' allele code is treated just like other allele codes (i.e. this is considered a variant with 3 ALT alleles).  You'd need to use a script to modify your VCF (perhaps by replacing all genotypes involving the '*' allele with missing calls) before feeding it to plink2.

--adjust's multiple-testing corrections count the number of direct allele tests, since those are what appear in the --adjust output file.  So a variant with 3 ALT alleles would increase the number of tests by 3.

[sy06...@gmail.com]

great thanks a lot for your time, 

Cheers, 

Saliha