Description
Hi.
When upgrading from bcftools 1.11 to bcftools 1.16 I noticed a completely erroneous behaviour.
Running bcftools 1.11
bcftools annotate --collapse all -c +ID -a ./dbsnp-155-split.vcf.gz -r chrM ./sample.gatk_annotatedM.vcf.gz --output-type z -o sample.annotatedM_1.11.vcf.gz --threads 16
behaves as expected populating ID field in the VCF entries with unknown (.) ALT allele and uncalled or HOM-REF genotypes.
Running the same command using bcftools 1.16 results in zero HOM-REF entries annotated with ID from the external annotation resource.
For instance, the source entry
chrM 1943 . A . 24.76 . AN=50;DP=3039 GT:AD:DP:RGQ 0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0:3038,1:3039:4
is annotated by bcftools 1.11 as
chrM 1943 rs1556422584 A . 24.76 . AN=50;DP=3039 GT:AD:DP:RGQ 0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0:3038,1:3039:4
while annotation by bcftools 1.16 results in no ID field annotation
chrM 1943 . A . 24.76 . AN=50;DP=3039 GT:AD:DP:RGQ 0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0/0:3038,1:3039:4
This behaviour is reproducible in all canonical chromosomes, haploid and diploid genotypes (as well as other genotype ploidy values). I also noticed the same behavior in bcftools v1.15.1.
The test files are supplied in an archive with file names stored as used in the provided command.