Not a valid path value type: java.util.LinkedHashMap ([id:spikein_fasta])

[Gin-Wang]

Description of the bug

I met errors in using cutandrun pipeline. In 3.1 it goes well till TRIMGALORE and FASTQC and gets error 'NFCORE_CUTANDRUN:CUTANDRUN:ALIGN_BOWTIE2:BOWTIE2_TARGET_ALIGN (2)' Caused by: Not a valid path value type: java.util.LinkedHashMap ([id:spikein_fasta]).

The error occured in 3.0 version and could not run the pipeline. All input is reachable and readable by anyone. It is strange that I have a completed process of NFCORE_CUTANDRUN:CUTANDRUN:ALIGN_BOWTIE2:BOWTIE2_SPIKEIN_ALIGN, but it still got Not a valid path value type in spike in fasta.

I have posted this in the slack channels, and several people met the same error just like me. I wonder if there is a bug or just mistakes in the command line.

here the parameters I’m trying.

Command used and terminal output

INPUT=/.../cuttag/samplesheet.csv
OUTDIR=/.../cuttag/
FASTA=/home/Reference/GRCh38/GRCh38.primary_assembly.genome.fa
GTF=/home/Reference/GRCh38/gencode.v41.primary_assembly.annotation.gtf
BLACK=/home/Reference/GRCh38/hg38-blacklist.v3.bed
SPIKEIN_FASTA=/home/Reference/SpikeIn/Escherichia_coli_K_12_DH10B/NCBI/2008-03-17/Sequence/WholeGenomeFasta/genome.fa
SPIKEIN_INDEX=/home/Reference/SpikeIn/Escherichia_coli_K_12_DH10B/NCBI/2008-03-17/Sequence/Bowtie2Index
 
nextflow run /.../nf-core-cutandrun-3.1/ --input ${INPUT} --outdir ${OUTDIR} --fasta ${FASTA} --gtf ${GTF} --igenomes_ignore --max_memory 150.GB -profile singularity --macs_gsize 2700000000 --blacklist ${BLACK} --save_merged_fastq --replicate_threshold 2 --spikein_fasta $SPIKEIN_FASTA --spikein_bowtie2 ${SPIKEIN_INDEX} --save_reference --use_control false -bg

Relevant files

pipeline_report.txt
nextflow.log

System information

No response

[cjfields]

I had a very similar issue that occurred when I let the workflow index a given FASTA file, which seems to be the default for other workflows when giving a --fasta sequence and no index file. If I supply the bowtie2 index path using --bowtie2 it works fine, which makes me suspect there is some difference in the input channels (e.g. between when the workflow generates the index vs the index being provided on command line).

[Gin-Wang]

I have tested it as @cjfields mentioned, using the existing bowtie2 index instead of generating an index by the pipeline itself. It seems to go well now and have passed the alignment process.

[chris-cheshire]

Hi @Gin-Wang @cjfields , I have managed to reproduce and fix the error. It was a channel format issue. its fixed in the commit below and will be in the next release luslab@cdfc356

[cjfields]

Awesome. thx @chris-cheshire !

[sme229]

Hi @chris-cheshire I'm getting the same error trying to run scnanoseq pipeline