Subsetting for A-length annotations (e.g. AF, PRI, F1R2, F2R1),

[ldgauthier]

accounting for incorrect input lengths

[gatk-bot]

Travis reported job failures from build 38115 Failures in the following jobs:

Test Type	JDK	Job ID	Logs
cloud	openjdk8	38115.1	logs
cloud	openjdk11	38115.14	logs
unit	openjdk11	38115.13	logs
integration	openjdk11	38115.12	logs
unit	openjdk8	38115.3	logs
variantcalling	openjdk8	38115.4	logs
integration	openjdk8	38115.2	logs

[LiviaMoura]

Hello @ldgauthier.

Just to let you to know what we tested so far, and what was working....

1 - Reblock with the snapshot + pipe with gnarly SNAP --> working on diploids (doesn`t work with sexual chromosomes)

2 - Reblock with the snapshot + pipe without gnarly SNAP --> not working, it broke at GenotypeGVCF

3 - Reblock with GATK 4.2.5 + pipe with gnarly 4.2.5 --> it doesn`t work

4 - Reblock with GATK 4.2.5 + pipe without gnarly 4.2.5 --> it doesn`t work

5 - Reblock with the snapshot + pipe without gnarly 4.2.5 --> It works even with sexual chromosomes

The 5th is the way to go... It seems Reblock from GATK 4.2.5 has a bug that the SNAPSHOP doesn`t have anymore.

I hope this help

[ldgauthier]

Yeah, there is a reblocking bug in 4.2.5.0. It's fixed in master, but we are waiting on a Google NIO fix for requester-pays buckets before we release a new version.

If the number of samples you have works with GenotypeGVCFs, then there's certainly nothing wrong with that. I also plan on taking a look at haploid genotypes in Gnarly this week, but I understand if you don't have a lot of faith in that considering how long the other fix has taken. :-D

[LiviaMoura]

Haha.. It's not about faith.. For now, I really don't know why I should use gnarly as a step, instead of GenotypeGVCF right away (with usegnarly = false). As gnarly says it's a "dirty" method, why should I use the "dirty" if the other method, without gnarly, is working?

If you can explain which is the difference, I'd be glad.

The only thing I know (because I read on an issue) is that we need Reblock to work with Dragen samples, so I'm following these tips

[ldgauthier]

This is a hot topic recently, so I already have a doc to compare and contrast: https://docs.google.com/document/d/1qws0owSEc0XGcZGAcxmBOEk8fiWS1Dnv4tvHNgC_xVU/edit?usp=sharing

Gnarly is still a "beta" tool. I wanted to add some way to reduce the number of alternate alleles, but that may be easier to do after this recent GenomicsDB update.

[LiviaMoura]

Excelent... good job I'm glad that I read "is more sensitive to rare alleles at common sites". As we are working with rare diseases, this is gold. Ok... I'm convinced that I need to wait for Gnarly fix xD (or I'll use chr1-22 on it, and sexual without it for now...) Thank you very much

[gatk-bot]

Github actions tests reported job failures from actions build 2203889963 Failures in the following jobs:

Test Type	JDK	Job ID	Logs
cloud	8	2203889963.10	logs
cloud	11	2203889963.11	logs
unit	11	2203889963.13	logs
integration	11	2203889963.12	logs
variantcalling	8	2203889963.2	logs
unit	8	2203889963.1	logs
integration	8	2203889963.0	logs

[codecov[bot]]

Codecov Report

Merging #7725 (dc2d48f) into master (b6a28d1) will decrease coverage by 73.022%. The diff coverage is 7.282%.

@@               Coverage Diff                @@
##              master     #7725        +/-   ##
================================================
- Coverage     86.941%   13.919%   -73.022%     
+ Complexity     36860      7450     -29410     
================================================
  Files           2211      2219         +8     
  Lines         173376    173843       +467     
  Branches       18710     18795        +85     
================================================
- Hits          150734     24197    -126537     
- Misses         16055    146992    +130937     
+ Partials        6587      2654      -3933     

Impacted Files	Coverage Δ
...bender/tools/walkers/variantutils/ReblockGVCF.java	0.000% <0.000%> (-80.711%)	:arrow_down:
...der/tools/walkers/variantutils/SelectVariants.java	40.212% <0.000%> (-40.955%)	:arrow_down:
...lbender/utils/variant/GATKVariantContextUtils.java	26.945% <0.000%> (-60.279%)	:arrow_down:
...s/variant/writers/ReblockingGVCFBlockCombiner.java	0.000% <0.000%> (-77.083%)	:arrow_down:
...lkers/genotyper/AlleleSubsettingUtilsUnitTest.java	1.709% <0.000%> (-97.009%)	:arrow_down:
...lkers/variantutils/ReblockGVCFIntegrationTest.java	1.010% <0.000%> (-96.664%)	:arrow_down:
...ools/walkers/variantutils/ReblockGVCFUnitTest.java	2.333% <0.000%> (-96.667%)	:arrow_down:
...tools/walkers/genotyper/AlleleSubsettingUtils.java	26.871% <23.333%> (-56.252%)	:arrow_down:
...nder/tools/walkers/genotyper/GenotypingEngine.java	49.686% <50.000%> (-36.478%)	:arrow_down:
.../org/broadinstitute/hellbender/utils/IGVUtils.java	0.000% <0.000%> (-100.000%)	:arrow_down:
... and 1954 more

[gatk-bot]

Github actions tests reported job failures from actions build 2340788459 Failures in the following jobs:

Test Type	JDK	Job ID	Logs
unit	11	2340788459.13	logs
integration	11	2340788459.12	logs
unit	8	2340788459.1	logs
variantcalling	8	2340788459.2	logs
integration	8	2340788459.0	logs