bbsplit fasta naming collision

[aiqc]

Description

When --bbsplit_fasta_list and --fasta point to the same file, it results in a naming collision error. As a workaround I had to copy and rename my reference genome, which is not ideal.

Details

run

nextflow run nf-core/rnaseq -r 3.12.0 -profile docker --input Test/sample_sheet.csv --outdir Test --skip_bbsplit false --bbsplit_fasta_list Test/bbsplit_list.csv --remove_ribo_rna --fasta Test/references/Homo_sapiens_copy/NCBI/GRCh38/Sequence/WholeGenomeFasta/human.fa --gtf Test/references/Homo_sapiens/NCBI/GRCh38/Annotation/Genes/genes.gtf

bbsplit

human,Test/references/Homo_sapiens/NCBI/GRCh38/Sequence/WholeGenomeFasta/human.fa
mouse,Test/references/Mus_musculus/NCBI/GRCm38/Sequence/WholeGenomeFasta/mouse.fa

output

ERROR ~ Error executing process > 'NFCORE_RNASEQ:RNASEQ:PREPARE_GENOME:BBMAP_BBSPLIT'

Caused by:
  Process `NFCORE_RNASEQ:RNASEQ:PREPARE_GENOME:BBMAP_BBSPLIT` input file name collision -- There are multiple input files for each of the following file names: human.fa

System information

nextflow version 23.04.2.5870
nf-core/rnaseq/release3.12.0
macOS 11.7.2

[aiqc]

stageAs workaround
https://stackoverflow.com/a/73662533/5739514

[pinin4fjords]

This functionality is working as intended. The main reference is passed from the workflow, 'contaminating' references via --bbsplit_fasta_list. It does not make sense to supply the same reference file in both places, since bbsplit will end up trying to choose mappings between two copies of the same reference.

If I've misunderstood the use case feel free to reopen, closing for now.