freeseek/mochawdl

MoChA WDL pipelines

mochawdl's Issues

reference for minimac4

Dear

Recently, we have been building a pipeline for imputation using Cromwell. However, the Warp version we have only supports HG19. Would it be possible to use your reference for Warp imputation? Additionally, we need to use HG38 version Minimac4 and WDL . Do you have any experience optimizing this workflow?

Begginner question

Hi, and thanks in advance for the help.

I'm very new with wdls and its interactions with docker. I'll tell everything I've run, then my doubts and problems.
I'm running the wdl in a local server, that's got plenty of resources (disk space, RAM, processor). My sample consists on 287 individuals, Illumina GSA array, iScan genotyping.
I've downloaded the dockerfiles from due section, and built them locally. I've set the input json to my inputs (mocha_inputs.json), I've created the tsvs, I've set the image names, and downloaded and extracted the reference files.

Ok, all set, and the pipeline run with a success stats:

[2024-01-07 12:13:14,81] [info] WorkflowManagerActor: Workflow actor for 51934643-1bec-4f38-809b-d2911f4937e6 completed with status 'Succeeded'. The workflow will be removed from the workflow store.
[2024-01-07 12:13:18,20] [info] SingleWorkflowRunnerActor workflow finished with status 'Succeeded'.

But some of the results were weird, to say the least. For example, the pileup.pdf is empty. No plots were generated. mocha_stats_tsv shows me genotyped with call_rates of ~0.33, all computed_gender are "U". I've just finished checking other analysis, and unfortunately, those samples were poorly chosen for the test, since those results match the ones I've had running the dragen pipeline to call genotypes. vcf_qc shows an error that all samples were removed during subsetting (And checking a little further, I've seen that's because of low call rate).

While pipeline was running, I've noticed some feedbacks that goes like this:

[2024-01-07 12:12:55,99] [warn] BackendPreparationActor_for_51934643:mocha.mocha_plot:0:1 [51934643]: Docker lookup failed
java.lang.Exception: Unauthorized to get docker hash rggplot2:latest
        at cromwell.engine.workflow.WorkflowDockerLookupActor.cromwell$engine$workflow$WorkflowDockerLookupActor$$handleLookupFailure(WorkflowDockerLookupActor.scala:221)
        at cromwell.engine.workflow.WorkflowDockerLookupActor$$anonfun$3.applyOrElse(WorkflowDockerLookupActor.scala:88)
        at cromwell.engine.workflow.WorkflowDockerLookupActor$$anonfun$3.applyOrElse(WorkflowDockerLookupActor.scala:73)
        at scala.runtime.AbstractPartialFunction.apply(AbstractPartialFunction.scala:38)
        at akka.actor.FSM.processEvent(FSM.scala:707) 

Many many times. One thing I want to point out, is that Broad's score 1.17 is currently down, so I've commented out its installation in the docker images, for a test drive of the pipeline.
There are many many things that can be responsible for the weird results I've got, I know. But in short, I'd like to know:

1. Am I using the dockerfiles, and the docker declaration of the images correct?
2. I've tried sudo usermod -aG docker "my-user", but no success, results still the same.
3. I dont know why the pipeline seems to run, and seems to succeed, even though things dont seem to be calculated.
4. Is it ok to use score 1.16? It's up by the time I'm writing this.
5. I'm going to try with better samples, and I'll come back here if things work out.

Sorry for the long issue, and the stupidity, I hope to bring good news soon.