Hi, I'm have trouble pinpointing an issue with GAT...
# sarek
a
Hi, I'm have trouble pinpointing an issue with GATK4_BASERECALIBRATOR. Nextflow logs says
Copy code
DEBUG nextflow.Session - Unexpected error while dumping DAG status
java.lang.IndexOutOfBoundsException: Index: -1, Size: 1124
. The log in work folder shows the job is killed by slurm. Our sysadmins says the killing signal is sent by nextflow. Thanks for the help !
With nextflow version 24.08.0-edge.5922 and sarek 3.5.1
Update: I cannot reproduce this issue fully
t
Why are you using 24.08.0-edge.5922? It is neither brand new, nor meant to be a stable release, so you are combining the disadvantages of old software and cutting-edge software. If there is indeed a bug in the pipeline somewhere (rather than in that
nextflow
version), the most useful information for the developers would be how it works with the newest stable version (currently 25.10.3). Nextflow is easy to install (especially if a reasonably recent version of Java is already available), so you don't have to rely on sysadmins to do that, so that would be my recommendation.
a
Thanks for your reply. I'm using the version packaged with nix for reproducibility. I haven't had this issue so far with many runs, which I found suprising to get now. Thanks also for the suggestion, trying the latest stable version for nextflow (outside) and sarek was next on my todo list.
🤞 1
👍 1
Follow-up : using latest stable (25.10.3) did not fix the issue. There were errors about being unable to write trace or timeline.
Copy code
timeline.overwrite = true
trace.overwrite = true
dag.overwrite = true
report.overwrite = true
decreased the error down to 2
Copy code
java.lang.NullPointerException: Cannot invoke "nextflow.dag.MermaidRenderer$Node.getInputs()" because "target" is null at nextflow.dag.MermaidRenderer.getNodeLookup(MermaidRenderer.groovy:136)
and
Copy code
Feb-05 14:33:27.034 [SIGHUP handler] DEBUG nextflow.Session - Unexpected error while dumping DAG statusjava.lang.IndexOutOfBoundsException: Index: -1, Size: 1077 at java.base/java.util.ArrayList.rangeCheckForAdd(ArrayList.java:756)
My setup is to install dependencies with nix outside sarek. I haven't got this issue despite multiple runs. Also, I'm trying to resume a previous workflow
t
My setup is to install dependencies with nix outside sarek.
What are these dependencies? Also, have you tried the newest sarek version (3.7.1)? It would be interesting to know which piece of software is causing the errors.
a
All CLI tools needed by sarek are installed through nix (shameless plug https://github.com/apraga/reproducible-sarek-germline). Running the same analysis with the same nextflow and sarek version than 3 weeks ago now fails. So I'm not sure it's not an issue on our cluster ... but hard to pinpoint it. I didn't use the latest sarek as it requires dbsnp filename to ends with vcf.gz
t
So you are using nix essentially in place of Conda or a container engine (Singularity, Docker, etc.)? If that is the case, I could see that external events might screw something up ... To be honest, the error message does not scream of an error with any tool(s), so I don't know. But I would try to run sarek like it was designed, ideally with a container engine. If that fixes it, you have an indication what is going wrong.
a
I agree. I'm waiting of feedback from sysadmins as something is off. Trying sarek « as it was intended » is my next step :)
👍 1