The crazy adventures continue. I need to call variants with some Illumina reads from tomatoes (Solanum lycopersicum) with the SLM-r2.1 reference from NCBI. I already noticed that I need to disable iGenomes, as there is (apparently) no tomato there, and it tries to inject too many human references. Does anyone have experience with plants in sarek, or otherwise knows which tweaks are necessary, and which steps will become impossible?

We talk with someone once that wanted to run on tomatoes

Alright, so if I have a genome FASTA, I only need to disable recalibration and duplicate marking, and the rest should run correctly?

It should not run these steps if no vcf are given

It's complaining, but

--skip_tools

should do the job

Let me know where it's complaining, I'll check it out

That one was fixed by explicitly skipping the offending steps, but now I'm getting a new one:

The sample-sheet only contains tumor-samples, but the following tools, which were requested by the option "tools", expect at least one normal-sample : haplotypecaller

The samplesheet contains one sample, and I added a status of 0 to it just to be sure, but it's still complaining

can you please post logs and samplesheets etc?

Command:

nextflow run nf-core/sarek -profile conda -params-file params.yaml -stub-run

Params:

input: samplesheet.csv
outdir: out
step: mapping

tools: freebayes,haplotypecaller,vep
skip_tools: baserecalibrator,markduplicates

igenomes_ignore: true
fasta: [...]/GCF_036512215.1_SLM_r2.1_genomic.fna.gz

snpeff_db: SLM_r2.1

Samplesheet:

"patient","sample","lane","fastq_1","fastq_2","status"
"SolPi","SRX8569566","LANE","[...]/fastq/SRX8569566_SRR12039813_1.fastq.gz","[...]/fastq/SRX8569566_SRR12039813_2.fastq.gz","0"

Stdout:

WARN: nf-core pipelines do not accept positional arguments. The positional argument `true` has been detected.
HINT: A common mistake is to provide multiple values separated by spaces e.g. `-profile test, docker`.

WARN: If GATK's Haplotypecaller, Sentieon's Dnascope or Sentieon's Haplotyper is specified, without `--dbsnp` or `--known_indels no filtering will be done. For filtering, please provide at least one of `--dbsnp` or `--known_indels`.
For more information see FilterVariantTranches (single-sample, default): <https://gatk.broadinstitute.org/hc/en-us/articles/5358928898971-FilterVariantTranches>
For more information see VariantRecalibration (--joint_germline): <https://gatk.broadinstitute.org/hc/en-us/articles/5358906115227-VariantRecalibrator>
For more information on GATK Best practice germline variant calling: <https://gatk.broadinstitute.org/hc/en-us/articles/360035535932-Germline-short-variant-discovery-SNPs-Indels->
[-        ] process > NFCORE_SAREK:PREPARE_GENOME:BWAMEM1_INDEX                         -
[-        ] process > NFCORE_SAREK:PREPARE_GENOME:GATK4_CREATESEQUENCEDICTIONARY        -
[-        ] process > NFCORE_SAREK:PREPARE_GENOME:SAMTOOLS_FAIDX                        -
[-        ] process > NFCORE_SAREK:PREPARE_INTERVALS:BUILD_INTERVALS                    -
[-        ] process > NFCORE_SAREK:PREPARE_INTERVALS:CREATE_INTERVALS_BED               -
[-        ] process > NFCORE_SAREK:PREPARE_INTERVALS:TABIX_BGZIPTABIX_INTERVAL_SPLIT    -
[-        ] process > NFCORE_SAREK:PREPARE_INTERVALS:TABIX_BGZIPTABIX_INTERVAL_COMBINED -
The sample-sheet only contains tumor-samples, but the following tools, which were requested by the option "tools", expect at least one normal-sample : haplotypecaller

Log:

[lots of irrelevant lines]
nextflow.exception.WorkflowScriptErrorException: The sample-sheet only contains tumor-samples, but the following tools, which were requested by the option "tools", expect at least one normal-sample : haplotypecaller
        at org.codehaus.groovy.vmplugin.v8.IndyInterface.fromCache(IndyInterface.java:321)
        [more traceback]

could be an issue with stub, could you try with test data?

-profile test

works normally with

-stub-run

, so it is probably not this. I will try the real data without stub anyway, although I wanted to use it to detect outstanding errors more quickly

that looks ok though, the empty fields should be for bams or vcfs

yeah I have a feeling this message is a red herring

Why do you want to skip

markduplicates

?

Tomato genome, and no known sites

BQSR obviously doesn't work, but Markduplicates doesn't need known sites?

Maxime said I should skip it, so I do

can you show the rest of the log please

Also you have this in the stdout:

WARN: nf-core pipelines do not accept positional arguments. The positional argument `true` has been detected.
HINT: A common mistake is to provide multiple values separated by spaces e.g. `-profile test, docker`.

SORRY

It took a while to copy from the VM, but here's the full log

Mar-17 13:44:38.296 [main] ERROR nextflow.cli.Launcher - @unknown
nextflow.exception.WorkflowScriptErrorException: This path is not available within annotation-cache.
Please check <https://annotation-cache.github.io/> to create a request for it.

I think you need to set

genome: null

and then for snpeff also set the cache location

let's try without snpeff for now

https://nf-co.re/sarek/3.8.1/docs/usage/#use-annotation-cache-for-snpeff-and-vep I am not sure we have the tomato cache preset in the annotation-cache bucket, so you might need to go ahead and fetch it form the remote cache location instead

Looks like disabling snpeff fixed the empty channel issue, it's creating Conda environments now

yep I think you'll need to set up the tomato cache locally. you can use downloadsnpeffcache pipeline for this and then point to that. I am pretty sure the pipeline tried to access the tomato cache from our s3 bucket where it is not set up

Alright, I will first try to get a full run without snpeff and then I will see about the cache