bcftools call transforms the mpileup -3aag variant into a TAAGAAGA>TAAGA variant

Question

[Michi]

Hi, from a previous run of my sample, I know that at the position in question there is a TAAG>T variant (deletion of AAG), which is also clearly visible in IGV.

A samtools mpileup confirms this:

1    225685621    T    10    ,,,-3aag,-3aag,-3aag,-3aag,,-3aag,,-3aag    CFFG7GGWGG    34    <<<<<<<>>><<<<<><>><>>>><<><<<>>><    CFHHJJJHFIIEJJGJJJJJJIJJJFCFFEJJHD

But, and here is my issue, the bcftoolsc call, or bcftools view command turns this variant into the following, which disagrees with the previous outputs

1    225685621    .    TAAGAAGA    TAAGA    174    .    INDEL;IDV=7;IMF=0.636364;DP=11;VDB=0.0633369;SGB=-0.616816;MQ0F=0;ICB=1;HOB=0.5;AC=1;AN=2;DP4=0,2,0,6;MQ=60    GT:PL:DP:DV:DP4    0/1:208,0,87:8:6:0,2,0,6

Thus, I'd like to understand why this is happening, and how I could prevent that behaviour. Is it possible that this is a bug?

Thanks for any help,
Michael

EDIT:

I could solve the problem by normalizing the call with piping the bcftools call output to:

bcftools norm -f genome.fa -

But still, I do not understand the representation given above.

Answer 1

[Devon Ryan]

The outputs are in complete agreement, they both say that AAG is deleted, they just represent that in different ways.

[Michi]

Ha, I was really hoping that was the answer, but I cannot see it.

If you remove TAAGA from TAAGAAGA, the result is AGA as a difference, ergo the deletion.. Now I assume I am wrong with my argument, so could you please elaborate? Thank you

[Devon Ryan]

TAAGA isn't what's removed, it's the result of the change (i.e., we start with TAAGAAGA, something happens, and we then see TAAGA). VCF files will represent the states observed, not a reference and then change.