Cannot invoke method minus() on null object

[MikeWLloyd]

I have encountered an issue with Nextflow 25.10.3 (25.10.3.10983), where a module that has historically worked without issue is now reporting an error: Cannot invoke method minus() on null object -- Check script

I have an nf-test case that reproduces the issue:

🚀 nf-test 0.9.3
https://www.nf-test.com
(c) 2021 - 2024 Lukas Forer and Sebastian Schoenherr


Test Process GATK_GATHERBQSRREPORTS

  Test [3a4e9bb0] 'Main Function'
    > N E X T F L O W  ~  version 25.10.3
    > Launching `.nf-test-3a4e9bb04f6401697f6ec6f0d7996d.nf` [goofy_heyrovsky] DSL2 - revision: 9f3d9015f2
    > ERROR ~ Error executing process > 'GATK_GATHERBQSRREPORTS (testSample_42)'
    >
    > Caused by:
    >   Cannot invoke method minus() on null object -- Check script 'modules/gatk/gatk_gatherbqsrreports.nf' at line: 20
    >
    >
    > Source block:
    >   String my_mem = (task.memory-1.GB).toString()
    >   my_mem =  my_mem[0..-4]
    >   inputs = reports.collect { "-I $it" }.join(' ')
    >   """
    >   mkdir -p tmp
    >   gatk --java-options "-Xmx${my_mem}G -Djava.io.tmpdir=`pwd`/tmp" GatherBQSRReports \
    >   ${inputs} \
    >   -O ${sampleID}_recal_data.table
    >   """
    >
    > Container:
    >   /projects/omics_share/meta/containers/broadinstitute-gatk-4.2.4.1.img
    >
    > Tip: you can try to figure out what's wrong by changing to the process work dir and showing the script file named `.command.sh`
    >
    >  -- Check '.nf-test/tests/3a4e9bb04f6401697f6ec6f0d7996d/meta/nextflow.log' file for details
    > Execution cancelled -- Finishing pending tasks before exit
    FAILED (11.696s)

  Assertion failed:

  assert process.success
         |       |
         |       false
         GATK_GATHERBQSRREPORTS

...

FAILURE: Executed 1 tests in 11.748s (1 failed)

when I run the same test with version 24.10.6.5937, I have no issue.

🚀 nf-test 0.9.3
https://www.nf-test.com
(c) 2021 - 2024 Lukas Forer and Sebastian Schoenherr


Test Process GATK_GATHERBQSRREPORTS

  Test [3a4e9bb0] 'Main Function' PASSED (54.648s)


SUCCESS: Executed 1 tests in 54.724s

Thoughts on what could cause this?

I'm happy to provide more details if needed, just let me know what.

[bentsherman]

Based on the error I assume it's this line:

task.memory - 1.GB

Likely task.memory is null because it wasn't set in the config. Can you confirm whether you set it for this process?

Maybe you're using a modified or older version of the module, but I checked the latest version of nf-core/gatk4/gatherbqsrreports and it has an extra check for task.memory being null:

https://github.com/nf-core/modules/blob/e4ddbdf46b37c889dfce35c96e87ace2e790b1ef/modules/nf-core/gatk4/gatherbqsrreports/main.nf#L26-L31

[MikeWLloyd]

I am using modules that we have built not NF-core: https://github.com/TheJacksonLaboratory/jds-nf-workflows/blob/main/modules/gatk/gatk_gatherbqsrreports.nf.

Historically in version 24.10.6.5937 (and earlier) that logic resolved properly as can be seen in the .command.sh file:

 cat .command.sh
#!/bin/bash -ue
mkdir -p tmp
gatk --java-options "-Xmx39G -Djava.io.tmpdir=`pwd`/tmp" GatherBQSRReports     -I A2_chr10_recal_data.table -I A2_chr1_recal_data.table -I A2_chr2_recal_data.table -I A2_chr3_recal_data.table -I A2_chr4_recal_data.table -I A2_chr5_recal_data.table -I A2_chr6_recal_data.table -I A2_chr7_recal_data.table -I A2_chr8_recal_data.table -I A2_chr9_recal_data.table     -O testSample_42_recal_data.table

With the newer Nextflow version and error, it doesn't generate the .command.sh file, or for that matter it doesn't make the work/cache for the step and just crashes immediately.

Weirdly, in a production setting, I was able to get it to clear the step on resume, and it moved on to the next step. It still crashed on that step because of the same logic. This logic block has been in most (if not all of our) GATK based modules for awhile now.

[MikeWLloyd]

I just tried:

    script:
    // String my_mem = (task.memory-1.GB).toString()
    // my_mem =  my_mem[0..-4]

    inputs = reports.collect { "-I $it" }.join(' ')

    """
    echo ${task.memory}
    mkdir -p tmp
    gatk --java-options "-Xmx38G -Djava.io.tmpdir=`pwd`/tmp" GatherBQSRReports \
    ${inputs} \
    -O ${sampleID}_recal_data.table
    """

and yes, it is indeed reporting null

$ cat .command.log
INFO:    underlay of /etc/localtime required more than 50 (94) bind mounts
null
14:21:55.563 INFO  NativeLibraryLoader - Loading libgkl_compression.so from jar:file:/gatk/gatk-package-4.2.4.1-local.jar!/com/intel/gkl/native/libgkl_compression.so
Feb 05, 2026 2:21:55 PM shaded.cloud_nio.com.google.auth.oauth2.ComputeEngineCredentials runningOnComputeEngine
...

We set the task memory with

    cpus = 1
    memory = 40.GB
    time = '03:00:00'

Did the internal task reference logic / variable change with the newer version, and I missed that?

[bentsherman]

Did the internal task reference logic / variable change with the newer version, and I missed that?

No, it's still just task.memory

My guess is that you have some extra config somewhere that is unsetting the memory directive, but I'm not certain. Try to create a minimal reproducible example, then we should be able to isolate the issue

[bentsherman]

Just saw the module you linked. I see you are setting the directives in the process. In this case there should be no equals sign:

    cpus 1
    memory 40.GB
    time '03:00:00'

I' don't know' if that is the root cause, just wanted to call it out