Vs 1063 bit pack ref ranges data into a more efficient representation #8543
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…o expand out the packed ref field so it looks like what the prepare tables are expecting
…hat I copied here
…sed ref extract. But, until I sort out a better way to load them in advance, this should work.
… integration used compressed refs
mcovarr
requested changes
Oct 10, 2023
There was a problem hiding this comment.
Needs documentation, unit tests over core logic, WDL fix.
EDIT: Also as discussed in standup 2023-10-10, the Hail integration tests which effectively represent the AoU path should switch this on by default as we're currently planning to use this feature from Echo onward.
src/main/java/org/broadinstitute/hellbender/tools/gvs/ingest/CreateVariantIngestFiles.java
Show resolved
Hide resolved
src/main/java/org/broadinstitute/hellbender/tools/gvs/ingest/RefRangesAvroWriter.java
Show resolved
Hide resolved
src/main/java/org/broadinstitute/hellbender/tools/gvs/common/SchemaUtils.java
Show resolved
Hide resolved
src/main/java/org/broadinstitute/hellbender/tools/gvs/common/SchemaUtils.java
Show resolved
Hide resolved
src/main/java/org/broadinstitute/hellbender/tools/gvs/common/SchemaUtils.java
Show resolved
Hide resolved
gbggrant
reviewed
Oct 11, 2023
gbggrant
reviewed
Oct 11, 2023
There was a problem hiding this comment.
Looks good. A couple of comments.
Also, please put in the code / documentation somewhere the representation of the packed ref_ranges table (the one in the spike: https://broadworkbench.atlassian.net/browse/VS-973
…to lock the bit packing behavior in place
…ate the gatk docker
…plain-er English for easier comprehension
mcovarr
approved these changes
Oct 12, 2023
gbggrant
approved these changes
Oct 12, 2023
flipping defaults back
koncheto-broad
deleted the
VS-1063-bit-pack-ref-ranges-data-into-a-more-efficient-representation
branch
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Adding in the option for using a compressed representation of our reference data. The flag use_compressed_references needs to be passed into ingest and also export, and doing so causes us to use a different internal schema for the ref_ranges table that can save ~40% on space. On the VCF path, the packed data is expanded while we populate the prepare tables before extract. On the VDS path, the packed data is expanded while we extract from the tables to create the Avro files.
I updated integration tests to take use_compressed_refs as an option, and saw what I needed. Integrations tests for the VCF and VDS paths go to completion and fail, but in the expected way. AssertIdenticalOutputs, the important part, passes. AssertTableSizesAreExpected fails because the ref_ranges table is so much smaller (255884464 vs 431805033 expected) and AssertCostIsTrackedAndExpected fails because the PrepareRangesCallsetTask.GvsPrepareRanges.BigQuery Query Scanned is similarly smaller (340787200 vs 515899392). Given the nature of the change and the fact that it is an optional flag that defaults to zero, updating the integration tests to cover this path did not seem sound at this time. But using them to verify that the content of the final vcfs wasn't affected by the change makes more sense.
VCF integration run: https://job-manager.dsde-prod.broadinstitute.org/jobs/e9896786-9d3f-48b9-ab6f-9a76d9fafafe
Hail integration run: https://job-manager.dsde-prod.broadinstitute.org/jobs/dbdd0934-50c3-4477-a191-3282341eacd3
another post-merge integration run with same results: identical outputs, failure due to lower table sizes and query scan costs
https://job-manager.dsde-prod.broadinstitute.org/jobs/edabdac3-1856-4e2a-aadc-3bcd4c353956