markdown_to_html.r markdown built with different version : using local lib? about chipseq

[drpatelh]

Hi @AlastairKerr ! Thanks for running the pipeline. The process shouldnt need to try and install markdown because it will already be in the singularity image.

Can you send me the command you are using please? Im assuming you have singularity installed?

[drpatelh]

Hmmm...markdown is definitely in the container. I can load it in R once in the image:
singularity shell nfcore-chipseq-1.0.0.img

I wonder if its got something to do with an R environment variable lying around thats over-riding the version in the image.

[drpatelh]

Do any of these commands give you the same path as #<local path># in the above error trace? If so, it probably means that it needs to be unset.

R RHOME
echo $R_HOME
echo $PATH

[drpatelh]

Its always worth trying to get the test profile running if this is the first time you are using an nf-core pipeline.

So assuming you have Nextflow and Singularity on your path you could try and run:
nextflow run nf-core/chipseq -profile test,singularity

If you have already pulled and converted the container, and have it locally you could run:
nextflow run nf-core/chipseq -profile test,singularity -with-singularity /path/to/nfcore-chipseq-1.0.0.img

[AlastairKerr]

command I used :
nohup nextflow run nf-core/chipseq --design design.tsv --genome R64-1-1 -profile singularity &

nf-core was working and the image downloaded to my cache with no issues.

I've used the rnaseq pipeline in nf-core before with no errors.

Output using
nohup nextflow run nf-core/chipseq -profile test,singularity -with-singularity //nfcore-chipseq-1.0.0.img --design design.tsv --genome R64-1-1 &

I get the same error

• markdown built with a different version,
• attempting to download
• make error (not found in path)

[AlastairKerr]

Thanks, this has helped me track down the error.
.libPaths() was being set in my .Rprofile and being imported into the container. The libs in the container had less priority than the environment.

I'm not sure what would be the easiest solution. Maybe remove this file and try and set an environmental variable instead that I can unset in my nextflow.config.

[ewels]

I've had similar issues in the past with conda inheriting environment variables from outside the container. It's not great! It would be ideal if there was a way to totally isolate the running environment from the host env vars... 🤔