Plink altering VCF IDs

Question

[dec986]

I'm running a VCF command thus:

plink --vcf chrX.M.vcf.gz --set-hh-missing --output-chr M --recode vcf --out chrX.M.set_hh_missing

but the output is distorted in that the sample IDs are doubled, for example

ABCDEF becomes ABCDEF_ABCDEF

None of the options seem like they would do this, after reading the instructions. Why is this happening? How can I prevent the sample IDs from doubling like this?

Answer 1

[chrchang523]

plink 1.x requires two-part sample IDs, so there are some rough edges re: import/export of single-part sample IDs. See https://www.cog-genomics.org/plink/1.9/input#double_id and https://www.cog-genomics.org/plink/1.9/data#recode for details.

Two workarounds:

• Add "--const-fid 0 --keep-allele-order", and replace "--recode vcf" with "--recode vcf-iid" in your command line.
• Use plink 2.0 when working with VCFs. In addition to natively supporting single-part sample IDs, it preserves VCF header lines and QUAL/FILTER/INFO columns, does not automatically swap REF/ALT alleles on you (this is what --keep-allele-order in the first workaround counteracts), and can handle multiallelic variants, phase, and dosage data.

Also note that you can add 'bgz' to --recode ("--recode vcf-iid bgz") to request bgzipping of the VCF.