Container for scsplit

[IreneRobles]

I am building a nf-core module for scsplit Trying to follow guidelines, I would like to add a scsplit docker to biocontainers, so it can be imported by my module

Software URL: https://github.com/jon-xu/scSplit

This docker image would work:

FROM python:3.9 as base

ARG SCSPLIT_VERSION="1.0.8"

RUN apt-get update && apt-get install -y python3-pip git

RUN pip install "setuptools==57.5.0" wheel numpy pandas pysam pyvcf3 scikit-learn scipy statistics 
RUN wget https://github.com/jon-xu/scSplit/archive/refs/tags/v${SCSPLIT_VERSION}.tar.gz -O /opt/scsplit.gz && \
    tar xvzf /opt/scsplit.gz -C /opt && \
    rm /opt/scsplit.gz

ENV PATH="${PATH}:/opt/scSplit-${SCSPLIT_VERSION}"

CMD [ "bash" ]

[IreneRobles]

Attempting to suggest a docker image, I created this pull request: https://github.com/BioContainers/containers/pull/564

[mboudet]

Ok, it seems this package already exists in bioconda (https://bioconda.github.io/recipes/scsplit/README.html)

As such, there is already a docker image available here: https://quay.io/repository/biocontainers/scsplit

[IreneRobles]

Yes, however, I cannot use the tool as described in https://github.com/jon-xu/scSplit?tab=readme-ov-file#3-building-allele-count-matrices with the current docker recipe.

If you log into the current biocontainer docker:

docker run -it quay.io/biocontainers/scsplit:1.0.8.2--pyh7cba7a3_0

And you try to run the tool:

$scSplit
sh: scSplit: command not found

If you try to run the tool with my docker recipe:

$scSplit
usage: scSplit <command> [<args>]
Commands:
   count     Generate REF/ALT count matrices from pooled BAM file
   run       Demultiplex the scRNA-Seq using REF/ALT count matrices
   genotype  Generate sample genotypes in VCF format
scSplit: error: the following arguments are required: command

Therefore, the docker image needs to change.

On Fri, 29 Mar 2024 at 15:33, mboudet @.***> wrote:

Ok, it seems this package already exists in bioconda ( https://bioconda.github.io/recipes/scsplit/README.html)

As such, there is already a docker image available here: https://quay.io/repository/biocontainers/scsplit

— Reply to this email directly, view it on GitHub https://github.com/BioContainers/containers/issues/565#issuecomment-2027326095, or unsubscribe https://github.com/notifications/unsubscribe-auth/AEY4VEA6SFWEBCCPYVE5WXTY2VUQZAVCNFSM6AAAAABFDRM4TKVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDAMRXGMZDMMBZGU . You are receiving this because you authored the thread.Message ID: @.***>

[mboudet]

Hmm, seems like the conda install is missing some dependencies (like numpy) anyway. Beyond that, the docker image is not broken, it's just how the tool works when you install it with pip (either you import it in python, or you manually call '/usr/local/lib/python3.12/site-packages/scSplit/scSplit'. I suppose the conda recipe could have added this file in the path.

In any case, to avoid confusion, we do not add tools already existing in bioconda in this repository, so you'll have to check on the bioconda side to fix the recipe (and get a working docker image)