ScatterIntervals produces interval_list instead of intervals #5392
Conversation
|
@lbergelson Don't forget to change the WDLs, e.g., the CNN WDL task still globs Is there any difference between the functionality of this tool and that provided by Picard's IntervalListTools, @davidbenjamin @lucidtronix? I only ask now because @yfarjoun recently added some functionality to IntervalListTools that will allow us to scatter by interval count (rather than genomic length), which will replace a hacky bit of bash code that currently is responsible for doing this scatter in the gCNV WDL when I get around to it. So it might be worth revisiting. |
|
@samuelklee From ~10 minutes of reading code it looks like the functionality is the same. It's been a while but I suspect we created the GATK tool to get rid of a Picard jar dependency. |
|
@lbergelson you beat me because I was stuck trying to actually run a Picard tool in the integration test. (For future reference, that needs a workaround because the test running adds the ERROR level logging to all command lines and Barclay can't parse that for Picard tools for some reason.) The big reason I was using this instead of IntervalListTools is because the Picard version creates a terrible output file structure that I was having trouble capturing with a simple glob in WDL. I agree that the functionality here is largely redundant, but it was helping me get my workflow working faster at the moment. |
| public void testOneIntervalAlternateExtension() { | ||
| final int scatterCount = 5; | ||
| final File outputDir = createTempDir("output"); | ||
| final String extension = "-scattered.with.a.wierd.extension"; |
There was a problem hiding this comment.
Obviously Picard won't read this in, but would the GATK engine? I'm a little worried about giving users the (BB) gun to shoot themselves in the foot.
There was a problem hiding this comment.
I figured we'd want a way to let users specify the old pattern if they want.
* This matches what picard expects an interval list to be named. * Added a new --extension argument to allow changing the extension. * Fixes #5390
lbergelson
force-pushed
the
lb_split_intervals_extension
branch
from
6d4847f to
1ac0038
Compare
Codecov Report
@@ Coverage Diff @@
## master #5392 +/- ##
===============================================
+ Coverage 86.981% 86.983% +0.003%
- Complexity 30420 30421 +1
===============================================
Files 1850 1850
Lines 140932 140944 +12
Branches 15505 15505
===============================================
+ Hits 122584 122598 +14
+ Misses 12709 12707 -2
Partials 5639 5639
|
…stitute#5392) * This matches what picard expects an interval list to be named. * Added a new --extension argument to allow changing the extension . * Fixes broadinstitute#5390
@ldgauthier