VS-776. Update to latest version of VQSR Lite. #8269
Conversation
…c annotations right now.
Correct for the newly renamed output file.
…ateToLatestVQSRLite # Conflicts: # .dockstore.yml # scripts/variantstore/wdl/GvsCreateFilterSet.wdl
Codecov Report
Additional details and impacted files@@ Coverage Diff @@
## ah_var_store #8269 +/- ##
================================================
Coverage ? 42.749%
Complexity ? 23842
================================================
Files ? 2197
Lines ? 167119
Branches ? 18006
================================================
Hits ? 71442
Misses ? 90265
Partials ? 5412 |
…ateToLatestVQSRLite
gbggrant
changed the title
…ateToLatestVQSRLite
|
WRONG! FYI - the It was failing because I hadn't ported the test data. Now it passes! |
…ateToLatestVQSRLite
| gatk_docker = "us.gcr.io/broad-gatk/gatk:4.4.0.0", | ||
| annotations = ["AS_QD", "AS_MQRankSum", "AS_ReadPosRankSum", "AS_FS", "AS_MQ", "AS_SOR"], | ||
| resource_args = "--resource:hapmap,training=true,calibration=true gs://gcp-public-data--broad-references/hg38/v0/hapmap_3.3.hg38.vcf.gz --resource:omni,training=true,calibration=true gs://gcp-public-data--broad-references/hg38/v0/1000G_omni2.5.hg38.vcf.gz --resource:1000G,training=true,calibration=false gs://gcp-public-data--broad-references/hg38/v0/1000G_phase1.snps.high_confidence.hg38.vcf.gz --resource:mills,training=true,calibration=true gs://gcp-public-data--broad-references/hg38/v0/Mills_and_1000G_gold_standard.indels.hg38.vcf.gz --resource:axiom,training=true,calibration=false gs://gcp-public-data--broad-references/hg38/v0/Axiom_Exome_Plus.genotypes.all_populations.poly.hg38.vcf.gz", | ||
| extract_extra_args = "-L ${interval_list} --use-allele-specific-annotations", | ||
| score_extra_args = "-L ${interval_list} --use-allele-specific-annotations", | ||
| extract_runtime_attributes = {"command_mem_gb": 27}, | ||
| train_runtime_attributes = {"command_mem_gb": 27}, | ||
| score_runtime_attributes = {"command_mem_gb": 15}, |
There was a problem hiding this comment.
There's a lot more hard-coded here than I would have expected, are we confident we'll never want to override any of this?
There was a problem hiding this comment.
This is a good point.
There was a problem hiding this comment.
All set - added ability to pass runtime block as input.
Added test files.
|
|
||
| # These are the SNP and INDEL annotations used for VQSR Classic, the order matters. | ||
| Array[String] vqsr_classic_indel_recalibration_annotations = ["AS_FS", "AS_ReadPosRankSum", "AS_MQRankSum", "AS_QD", "AS_SOR"] | ||
| Array[String] vqsr_classic_snp_recalibration_annotations = ["AS_QD", "AS_MQRankSum", "AS_ReadPosRankSum", "AS_FS", "AS_MQ", "AS_SOR"] | ||
|
|
There was a problem hiding this comment.
nit: since these are being hard-coded, maybe don't even pass them as inputs, just have them hard-coded in the task?
There was a problem hiding this comment.
Yeah - I like that. It wasn't really clear to me why these were parameterized. It would seem unlikely that we would change this sort of thing.
|
|
||
| # reference files | ||
| # Axiom - Used only for indels | ||
| # Classic: known=false,training=true,truth=false |
There was a problem hiding this comment.
what does this comment (and the ones like it, below) mean?
Classic: known=false,training=true,truth=false
There was a problem hiding this comment.
That was from when I was trying to make sure that I had the right settings for both classic and lite. I don't think it's informative here (it was more of a one time idiot check I did), so will get rid of them.
Update to latest version of VQSR Lite
Refactor GvsCreateFilterSet.wdl to move VQSR Classic code to its own WDL