Hello, I have recently run into a new issue with module BAKTA_BAKTA that I did not have in the past. I have been making a pipeline for bacterial genome annotation and recently updated all process modules for a rework of many parts of the pipeline. I tried to replicate the errors manually by running the apptainer container directly, but it kept working just fine. Then, when I went back and tried to run the pipeline again, I got a completely new error. First error: tRNAscan-SE error code 2 Second error that is now in place of first error: PilerCR could not be executed(make sure it is installed or skip it using --skip-crispr) I am using the latest version of BAKTA_BAKTA and calling it from a subworkflow like this: // Run BAKTA bakta_out = BAKTA_BAKTA( ch_meta_fasta, params.bakta_db ? ch_meta_fasta.map{ _f -> file(params.bakta_db)} : file("${params.beav_dir}/databases/bakta_db").exists() ? ch_meta_fasta.map{ _f -> file("${params.beav_dir}/databases/bakta_db")} : ch_bakta_db_dl.combine(ch_meta_fasta).map{ _db, _meta, _f -> file("${params.beav_dir}/databases/bakta_db")}, [], [], [], [] ) The long second parameter is just to resolve where the bakta database is.

Hi! Both errors mean the tool is in the container but fails when Nextflow launches it. If PilerCR were missing, Bakta would say "not found"; "could not be executed" means

pilercr -options

returned an error. The error also changed between runs, so I suspect the task's environment rather than the module. A few things to try: 1. Reproduce Nextflow's exact run: in the failed task's work dir, run

bash .command.run

and check

.command.err

. 2. See the real tRNAscan-SE error: set

ext.args = '--verbose --debug'

for

BAKTA_BAKTA

. 3. Rule out host env leaking in: add

apptainer.runOptions = '--cleanenv'

. Leaked

LD_LIBRARY_PATH

,

PERL5LIB

or

TMPDIR

from HPC modules or conda can break both tools. 4. Clear the cached image: the module switched to a new Seqera container (Bakta 1.12) in April. Delete the cached

.sif

so Nextflow pulls it again, and make sure your manual test uses that same image. Also check whether your manual test ran on a login node while the tasks ran on compute nodes. Temporary workaround:

--skip-trna --skip-crispr

. If it still fails, could you share

.command.err

and your apptainer config?