plink vcf recodeA error ALT allele duplicates REF allele

Question

[cynthiamonster]

I am using plink to to export sequence data in vcf format to a raw file but keep getting the following error at a specific line that halts the entire process:

"ALT allele duplicates REF allele on line 166993 of .vcf file"

using the following command

plink --vcf "infile.vcf.gz" --extract exomeids.txt --out "outfile" --recodeA

When I examine this line I see that all subjects appear to be homozygous (listed as either 0/0 or 1/1). I can remove this exact chromosome or position but then get the same error on a different line. Is there a more succinct way to remove this error than listing all the positions that result in this error?

Note that this position is not one that I want to extract for my results but it is still processed in the initial part of the --out command. I have also used similar commands for other vcf files with no errors so I am curious if something is wrong with the coding on this file since I cannot find documentation of these error elsewhere.

Thanks for your help in advance.

Answer 1

[chrchang523]

How was this VCF file generated? The REF and ALT alleles should never be identical; monomorphic variants should have either "." or some unobserved alternate allele in the ALT column.

With that said, you can recover from this situation with a script that detects lines where the REF and ALT columns are identical, and (i) sets ALT to "." and (ii) replaces all instances of "1/1" with "0/0" on those lines.

If that's too much of a hassle, you can also just delete those lines with cat infile.vcf | grep '^#' > infile_filtered.vcf followed by cat infile.vcf | grep -v '^#' | awk '{if ($4 != $5) print $0}' >> infile_filtered.vcf

[cynthiamonster]

Yes, thanks for the code. I was about to go down that path but was hoping there was a more simple way to exclude them from the --out command

Note: --maf setting restricting minor allele frequencies to larger values hasn't worked to exclude these either which doesn't fully make sense to me either.

[chrchang523]

This is because VCF import happens before everything else plink does: the entire file is imported, and only then are flags like --maf applied. Otherwise, it would be necessary to add --maf logic to every single import routine, etc.

(The full order of operations is at https://www.cog-genomics.org/plink/1.9/order .)

[cynthiamonster]

Thanks. Now I understand why --maf is not working but the --not-chr is effective since it is processed before the --vcf turns the file into binary format. Based on this order of operations it appears that I would not be able to exclude an exact position within a chromosome using --from --to commands since that happens after --vcf processing. I guess I'm back to deleting the lines of the file as was originally suggested.

Appreciate all the help!

Answer 2

[Pierre Lindenbaum]

awk '($0 ~/^#/ || $4!=$5)' input.vcf > out.vcf