F1R2 and F2R2 annotations not updated by GenotypeGvcfs

[nh13]

The gVCF output of HaplotypeCaller:

... ATTTTTT A,ATT,ATTT,ATTTT,ATTTTT,<NON_REF> ... GT:AD:DP:F1R2:F2R1:GQ:PL:SB	4/5:8,2,5,8,19,17,0:59:5,2,5,8,17,17,0:3,0,0,0,2,0,0:99:841,879,2162,594,1484,1406,395,983,921,881,140,419,309,264,313,382,470,281,174,0,402,828,1443,1223,962,474,507,1363:5,3,49,2

The VCF output of GenotypeGvcfs:

... ATT A,AT ... GT:AD:DP:F1R2:F2R1:GQ:PL ... 1/2:8,19,17:59:5,2,5,8,17,17,0:3,0,0,0,2,0,0:99:841,140,313,382,0,402

In both cases, F1R2 is 5,2,5,8,17,17,0. There are seven alleles in the gVCF but only three in the VCF, so the F1R2 is wrong for the latter.

[ldgauthier]

Traditionally we've only used the OxoGReadCounts annotation for somatic data, which was single-sample until very recently. I'll put a note in the docs and check the GGVCFs FORMAT combine operations.

Implementing an actual combine shouldn't be too hard. Is this a blocking issue for you @nh13 ?

[nh13]

@ldgauthier not a blocking issue at all, I just wanted to report it, in case others stumble upon it, and if you had free time to take a look.

[nh13]

@ldgauthier this does block downstream tools, like bcftools from using VCFs with this annotation, for example when subsetting and trimming the alleles.

[ldgauthier]

@nh13 I did get started on this. It works, but needs formal tests. I might be able to finish it up this week, but if your need is urgent and you trust me without tests, then there's a branch.

[nh13]

@ldgauthier will you also be able to fold in the changes in #5698 (comment)

[ldgauthier]

Can do!

…

On Thu, Mar 21, 2019 at 5:15 PM Nils Homer ***@***.***> wrote: @ldgauthier <https://github.com/ldgauthier> will you also be able to fold in the changes in #5698 (comment) <#5698 (comment)> — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#5704 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AGRhdAYT0Z-FJo8AP5GhMNGjRT_g0drKks5vY_ZhgaJpZM4bIBwm> .

-- Laura Doyle Gauthier, Ph.D. Associate Director, Germline Methods Data Sciences Platform gauthier@broadinstitute.org Broad Institute of MIT & Harvard 320 Charles St. Cambridge MA 0214