No need to sort variants in HaplotypeCallerSpark. #5909
Merged
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
Codecov Report
@@ Coverage Diff @@
## master #5909 +/- ##
===============================================
+ Coverage 80.117% 80.119% +0.002%
- Complexity 30673 30674 +1
===============================================
Files 1991 1991
Lines 149341 149342 +1
Branches 16481 16482 +1
===============================================
+ Hits 119647 119651 +4
+ Misses 23892 23890 -2
+ Partials 5802 5801 -1
|
jamesemery
approved these changes
May 1, 2019
| writeVariants(ctx, outputFile, variants, header, writeTabixIndex, true); | ||
| } | ||
|
|
||
| public static void writeVariants( |
| @@ -184,11 +184,10 @@ private static void processAssemblyRegions( | |||
|
|
|||
| final JavaRDD<VariantContext> variants = rdd.mapPartitions(assemblyFunction(header, referenceFileName, hcArgsBroadcast, annotatorEngineBroadcast)); | |||
|
|
|||
| variants.cache(); // without caching, computations are run twice as a side effect of finding partition boundaries for sorting | |||
There was a problem hiding this comment.
huh, does this have to do with the change in sorting? what sort of difference in runtime does this branch have?
There was a problem hiding this comment.
The cache call was added to avoid repeat work. I haven't tried re-running benchmarks yet but I suspect it will give some speed up.
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Outdated suggestions cannot be applied.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.
VariantsSparkSinkwill always sort variants before writing them out. However,HaplotypeCallerSparkalways processes reads in coordinate-sorted order, and produces variants in the same order, so there is no need forVariantsSparkSinkto sort variants. (In fact, in GVCF mode the sort is prohibitive since the engine creates a variant for every locus over the interval of interest, which go through the sort step before being merged into GVCF bands.)This PR removes the sort step for
HaplotypeCallerSpark(andPrintVariantsSpark, which doesn't need it either). All of the concordance unit tests pass, and as an additional sanity check I compared the GVCF output from running regularHaplotypeCalleron a large input BAM toHaplotypeCallerSpark(with and without variant sorting). Removing variant sorting actually made the GVCF output more similar to regularHaplotypeCaller- it reduced the number of differences from three to one. (The one difference is a minor difference in QUAL due to a boundary artifact.) See VCFs in vcfs.zip.