This is a test version of Biostars. For the public version, visit https://www.biostars.org.
PLINK, Unphased heterozygous hardcalls in partially-phased variants are poorly represented with bits=8

Hi, I'm currently trying to convert dosage data from the vcg.gz format to bgen 1.2 format (8 bits), using plink2, in order to use it later with LDpred-2. However, during the conversion process, I encountered a warning message:

"Warning: Unphased heterozygous hardcalls in partially-phased variants are poorly represented with bits=8.
It is necessary to use e.g. --dosage-erase-threshold 0.006 to re-import them cleanly."

How could I solve this?

The command used to convert from vcf.gz to bgen-1.2 is :

plink2 --vcf file_chr${CHR}.dose.vcf.gz 'dosage=HDS'  --export bgen-1.2 ref-first bits=8 --out ${out}_chr${CHR}

I've tried to add --dosage-erase-threshold 0.006 but I get the same warning.

Many thanks!

plink bgen vcf gwas ldpred-2

1 answer

The warning advises you to use --dosage-erase-threshold 0.006 if/when you use plink2 to re-import the data from the .bgen.

You can't avoid the warning without phasing the rest of your data. This is an intrinsic limitation of the 8-bit .bgen format.

Log in to answer this question.