I'm dealing with paired end amplicon sequencing data. I've produced a GVCF file with haplotype calls using:

gatk HaplotypeCaller -R $REF -I "$BAM" -O "$OUT"/results/variants/${SN}_HaplotypeCallerPGT.vcf -ERC GVCF

The vcf file it produces contains the PGT flag, and variants are called in the format 0|1.

I'm then calling the consensus sequence for each haplotype using

samtools faidx "$REF" chr5:80654720-80655181 | bcftools consensus -p "$PREFIX" -H 1pIu "$PREFIX".vcf.gz > "$OUT"/results/consensus/${SN}_consensus.1pIu.fa
samtools faidx "$REF" chr5:80654720-80655181 | bcftools consensus -p "$PREFIX" -H 2pIu "$PREFIX".vcf.gz > "$OUT"/results/consensus/${SN}_consensus.2pIu.fa

But the sequence of both 1plu and 2plu files is exactly the same. Surely variants should segregate as per the PGT flag??