NikdAK
commented
1 year ago Hi, we never used singularity for this, but I can try to help you on some of the docker-specific stuff. As a note, there is also a branch for nextflow, however this is highly experimental, not fully tested and we can not provide support for it: https://github.com/roland-rad-lab/MoCaSeq/tree/human-pipeline-nextflow
So for me it sounds like the folders are not correctly mounted/bound and the container is trying to generate new directories somewhere it should not.
"/var/pipeline/" is basically just the working/home/base directory within docker, while "/var/" in this case is just a dummy/username, but hardcoded within the pipeline. It should work if you mount/bind the volumnes using --bind. You could bind all your local folders to the requires locations within the container:
-B ${working_directory}:/var/pipeline/ # in this folder the individual folders for each sample will be generated
-B ${temp_directory}/temp/:/var/pipeline/temp/ # in this folder the temporary files will be placed and removed, this can be a scratch or some larger folder.
-B ${ref_directory}:/var/pipeline/ref/ # here the references should be. After successfully downloading it you can store it somewhere else and just bind it here
-B ${script_directory}:/opt/MoCaSeq/ # optional! like this you can easily modify the code on your machine (e.g. clone the github repo), because the local version (in ${script_directory}) will overwrite the container version (located in /opt/MoCaSeq/ within the container)
-v ${fastq_directory}:/var/pipeline/raw/ # in case you want to store the FASTQs somewhere else. You would pass the file arguments like this: -tf /var/pipeline/raw/TUMOR.R1.fastq.gz -tr /var/pipeline/raw/TUMOR.R2.fastq.gz
In docker you have "-it --entrypoint=/bin/bash", which allows you to enter the container and check the paths manually. I believe this is the singularity version: https://docs.sylabs.io/guides/3.1/user-guide/cli/singularity_shell.html singularity shell MoCaSeq.sif
Good luck!
dpcook
Hi there,
I'm hoping to run the pipeline on a cluster without root access, so was hoping to convert the Docker image for Singularity. I believe that works fine, but I can't seem to have the pipeline work successfully. I think I just don't understand what
/var/pipelineis.Successfully starts running the pipeline and builds the reference, but then throws the first error:
Fastq preprocess all goes as fine, but then get an error related to the one above:
I do see the .conf file in ${working_directory}/ref/GRCm38.p6/, but not in sample output directory's pipeline subdirectory.
The rest of the pipeline runs, but outputs largely empty files.
I figure I need the equivalent of the Docker flag
-v ${working_directory}:/var/pipeline/, but don't really know what that is (is var the sample name? I see the 'pipeline' subdirectory made in the output folder). From my limited knowledge of singularity, I thought the -B option gives directory permissions, but it doesn't recognize /var/, so I'm not sure how to navigate it.Sorry if the question is naive! Any help is appreciated