Attempt to make the GATK tasks a) auto-detect GATK version and b) swi… #365
Conversation
|
@clintval You might be interested in this given your recent set of tweets. This isn't as elegant as it could be, but is an attempt to handle GATK 3.x vs. 4.x with a single set of task classes. |
|
I support this PR ❤️ |
|
I spent some time thinking about how I would solve this problem in a typed workflow DSL like Dagr. I originally found inspiration in a little library that helps with data class migrations. I dabbled with this problem in Snakemake but I never arrived on a working solution. I'm certain there is an elegant solution out there that would empower the user! I'm thinking a first-class feature upon Backward-in-time SupportDagr maintains a compatibility interface with older versions of the executable (similar to the aim of this PR). Forward-in-time SupportClient written compatibility that provides a way to migrate an existing My free 2¢ idea
Thanks for listening! |
| val jar = gatkJar | ||
| if (jar == GatkTask.GatkJarPath) GatkTask.GatkMajorVersion | ||
| else GatkTask.majorVersion(jar) | ||
| } |
There was a problem hiding this comment.
Ditto about this being something to bake into JarTask. We could also have a trait too, since not everything is a JAR.
| protected def addWalkerArgs(buffer: ListBuffer[Any]): Unit | ||
|
|
||
| /** Helper function to select an argument name for pre vs. post V4 naming. */ | ||
| protected def either(preV4Name: String, postV4Name: String): String = if (gatkMajorVersion < 4) preV4Name else postV4Name |
There was a problem hiding this comment.
This seems brittle if ever the GATK API changes. And also brittle if this becomes a pattern we adopt for other tools who change their command line API more frequently.
There was a problem hiding this comment.
Agreed ... I think I might just go back to having two separate argument generation sections for 3 vs 4 for now.
| @@ -50,9 +50,13 @@ class GenotypeGvcfs private (ref: PathToFasta, | |||
| val dbSnpVcf: Option[PathToVcf] = None) | |||
| extends GatkTask("GenotypeGVCFs", ref, intervals=intervals) { | |||
|
|
|||
| if (gatkMajorVersion >= 4 && gvcfs.length > 1) throw new IllegalStateException( | |||
tfenne
force-pushed
the
tf_gatk_multi_version_support
branch
from
f567c0a
to
1bd30ce
Compare
| case n if n >= 4 => | ||
| // Args for version 4 | ||
| buffer.append("-I", bam.toAbsolutePath) | ||
| buffer.append("-O", vcf.toAbsolutePath) |
There was a problem hiding this comment.
In GenotypeGvcfs you have a section before this block that adds common arguments:
// Args that are common to all versionsI'd recommend moving common args similarly, or not have common args in GenotypeGvcfs
There was a problem hiding this comment.
I almost did it, but in fact there's only 1 common arg, and it felt not worth it. Note that output is -o vs.-O. But I can move the -I argument out.
| case n if n < 4 => | ||
| buffer.append("--out", vcf.toAbsolutePath.toString) | ||
| case n if n >= 4 => | ||
| buffer.append("--output", vcf.toAbsolutePath.toString) |
There was a problem hiding this comment.
Do we need .toString here and elsewhere since we call it later?
There was a problem hiding this comment.
I have no idea why there were .toStrings there in the first place. Removed.
| @@ -51,6 +51,8 @@ class Mutect1(val tumorBam: PathToBam, | |||
| ) extends GatkTask(walker="MuTect", ref=ref, intervals=Some(intervals)) { | |||
|
|
|||
| override protected def addWalkerArgs(buffer: ListBuffer[Any]): Unit = { | |||
| require(gatkMajorVersion < 4, s"Mutect v1 is not supported in GATK v${gatkMajorVersion}.") | |||
There was a problem hiding this comment.
This require is in the addWalkerArgs method, but in IndelRealigner and RealignerTargetCreator it's in the constructor. The former will check when we run the task, while the latter when we build the task. Up to you.
There was a problem hiding this comment.
Good call. I'll move this one up because failing earlier seems better.
Codecov Report
@@ Coverage Diff @@
## master #365 +/- ##
=======================================
Coverage 91.95% 91.95%
=======================================
Files 31 31
Lines 1156 1156
Branches 65 65
=======================================
Hits 1063 1063
Misses 93 93Continue to review full report at Codecov.
|
…tch command lines accordingly.
@nh13 This isn't quite done yet - firstly I haven't tested any of this and secondly I've only gone through HC and GenotypeGVCF tasks so far to adapt for 4+. That said, I'd like feedback on the approach I've taken before I go much further.