No channel created for ch_fasta_for_rsem_reference

[ewels]

Originally posted by @kviljoen in #375 (comment)

---

Unknown variable 'ch_fasta_for_rsem_reference' where it looks like there was no channel created for ch_fasta_for_rsem_reference.

[ewels]

@kviljoen - if you're able to post any more details about this problem in this issue, that would be really helpful (for example, log output). Thanks!

[kviljoen]

From the .nextflow.log I have: '[main] DEBUG nextflow.Session - Session aborted -- Cause: Unknown variable 'ch_fasta_for_rsem_reference'

[ewels]

Could you please post the whole log file? There may be other clues, and it will tell us which exact version of the pipeline you're using. The command you used, the input references you used and the different pipeline options may also be important.

[ewels]

The command / config used especially is likely to be helpful. The ch_fasta_for_rsem_reference channel is created in the combine_genome_annotations channel and used in the makeRSEMReference channel, and the automated tests are working fine, so it's likely to be down to the specific combination of input parameters that you're using.

[ewels]

Copied from #375 (comment):

---

Attached log file from dev run. Please note that this is a result of merging with my own version of the pipeline because we've had to make some small tweaks particularly with java-related process memory settings in the main.nf to get it to run optimally on our cluster so may be specific to my version of pipeline. Log file attached
testrun.log

[ewels]

Hi @kviljoen,

Thanks for this! For future reference, the launch command was:

nextflow run kviljoen/RNAseq --reads '/scratch/kviljoen/Sonwabile_RNAseq/process/bbmap_QC/2020-06-10-fastq_QC/bbduk/*_R{1,2}.fq' --igenomes_base /bb/DB/bio/rna-seq/references/ --forwardStranded --removeRiboRNA -profile uct_hpc -r dev

First though is that you're using quite an old version of Nextflow, 19.04.1 - see the warning that is generated:

Nextflow version 19.04.1 does not match workflow required version: >=19.10.0 -- Execution will continue, but things may break!

I doubt that this is causing this problem, but the fact that we've pinned 19.10.0 as the minimum required version of Nextflow makes me a little nervous.

Do you get the same issue running with the test data profile? So, this command:

nextflow run kviljoen/RNAseq --forwardStranded --removeRiboRNA -profile test,uct_hpc -r dev

Phil

[ewels]

Side note:

Editing the pipeline code makes debugging and updating super difficult - it's much much better to use nf-core/configs for the kinds of changes you've introduced there: https://nf-co.re/usage/configuration

• uct_hex is already in nf-core/configs, so doesn't need to be in the pipeline
• cache 'deep' and label changes may require some modification of the pipeline, but it's better to introduce these changes as options in the main pipeline so that everyone can use them if they want to

Tweaking the pipeline to run optimally without changing the main pipeline code is one of the main strengths of Nextflow & nf-core in my mind 😄

[kviljoen]

Hi Phil,

Thanks I'll try get our cluster nextflow updated. I don't get the same error when running nextflow run kviljoen/RNAseq --forwardStranded --removeRiboRNA -profile test,uct_hpc -r dev The pipeline does however fail at process STAR with Fatal INPUT FILE error, no valid exon lines in the GTF file: GRCh37__gfp.gtf Solution: check the formatting of the GTF file. Most likely cause is the difference in chromosome naming between GTF and FASTA file. Attached log
nextflow.log

Regarding the pipeline editing, of course you are right, I'll try keep changes to the config or feature requests if necessary. We are no longer using uct_hex which was PBS based, the new uct_hpc cluster is SLURM-based.

[ewels]

Weird, what is GRCh37__gfp.gtf? I can't see any reference of that file anywhere.

If uct_hex is dead, should we remove that from nf-core/configs and add uct_hpc instead?

Phil

[kviljoen]

I actually have no idea what GRCh37__gfp.gtf is. I have the following files under our saved references ../rna-seq/references/Homo_sapiens/Ensembl/GRCh37/Sequence/STARIndex/

Would be great if you could update with uct_hpc config, thanks @ewels :)

[ewels]

nf-core/configs#161

[apeltzer]

In theory, it could be that this was based on the recent introductions made in here #419 but local testing of this here:

nextflow run nf-core/rnaseq -r dev -profile test,docker also didn't get that error and just ran through. Maybe it would be a good idea indeed to get the config updated and then switch over to using the original nf-core/rnaseq pipeline?

[kviljoen]

Hi Alexander, Phil

I tried again after updating to Nextflow 20.04 and with the nf-core/rnaseq

nextflow run nf-core/RNAseq --forwardStranded --removeRiboRNA -profile test,uct_hpc -r dev

but still get the same error: testrun.log
Fatal INPUT FILE error, no valid exon lines in the GTF file:
GRCh37__gfp.gtf

Solution: check the formatting of the GTF file. Most likely cause is the
difference in chromosome naming between GTF and FASTA file.

If I try a run on test data:

nextflow -log /scratch/kviljoen/Sonwabile_RNAseq/process/testrun_dev_branch/testrun.log run nf-core/RNAseq --reads '/scratch/kviljoen/Sonwabile_RNAseq/process/bbmap_QC/2020-06-10-fastq_QC/bbduk/*_R{1,2}.fq' --igenomes_base /bb/DB/bio/rna-seq/references/ --forwardStranded --removeRiboRNA -profile uct_hpc -r dev

I get error:

No such variable: ch_fasta_for_rsem_reference

Check script '/home/kviljoen/.nextflow/assets/nf-core/RNAseq/main.nf'
at line: 910 or see
'/scratch/kviljoen/Sonwabile_RNAseq/process/testrun_dev_branch/testrun_2.log'
file for more details
testrun_2.log

[apeltzer]

Hi Katie,

I'll have a sit with some of the newer issues with the RNAseq pipeline these days after work to get this fixed. Can't work on it during work atm unfortunately, so have to shift this into spare time.

Thanks for log & error details, that helps a lot in debugging!

[apeltzer]

Should be fixed as of #436