Make conda environment.yml about chipseq

[ewels]

Super cool! Yes, phantompeakqualtools is part of spp, so just installing that should do the job. Great work @apeltzer - thanks! That's quite a relief that we don't need to package stuff ourselves.

There are a few other issues that need testing / thinking about now... I've updated the original comment with some checkboxes.

[apeltzer]

Just checked and everything should work now (at least from what I see in main.nf @ewels I will send in a PR of bioconda branch to master and we'll see what the tests say?

[apeltzer]

I figured out 99% of all packages (I hope at least) in bioconda or conda-forge (see the bioconda branch which I committed to).

The missing one is "phantompeakqualtools", but if I'm not mistaken (please check this out @ewels ), this is also now packaged in a CRAN R package as "spp", therefore available in bioconda:

https://bioconda.github.io/recipes/r-spp/README.html

If this is the case, we just need to add:

  - bioconda::r-spp=1.15.2

and could be done with the new environment.yaml.

[apeltzer]

See also here for the last missing package : http://compbio.med.harvard.edu/Supplements/ChIP-seq/

[apeltzer]

Only very small tools missing:

• run_spp.R
• ngs.plot.R

These are highlighted in the bioconda branch. The docker image builds fine, but I don't know where to get the run_spp.R script from, as the bioconda package doesn't install that unfortunately.

/opt/conda/envs/nfcore-chipseq-1.4dev/lib/R/library/spp/

[ewels]

run_spp.R is from phantompeakqualtools. We were downloading this from https://code.google.com/archive/p/phantompeakqualtools/ but it looks like the latest version of the code is now at https://github.com/kundajelab/phantompeakqualtools

So me saying that "phantompeakqualtools is part of spp" was probably wrong sorry.

ngs.plot.R is found at https://github.com/shenlab-sinai/ngsplot

[ewels]

Installation script for these tools by @tiagochst in #10:

ENV NGSPLOT_VERSION="2.63"
RUN curl -fsSL https://github.com/shenlab-sinai/ngsplot/archive/${NGSPLOT_VERSION}.tar.gz -o /opt/ngsplot_${NGSPLOT_VERSION}.tar.gz && \
    tar xvzf /opt/ngsplot_${NGSPLOT_VERSION}.tar.gz -C /opt/ && \
    rm /opt/ngsplot_${NGSPLOT_VERSION}.tar.gz
ENV PATH=${PATH}:/opt/ngsplot-${NGSPLOT_VERSION}/bin
ENV NGSPLOT=/opt/ngsplot-${NGSPLOT_VERSION}/

RUN wget "https://drive.google.com/uc?export=download&id=0B5hDZ2BucCI6SURYWW5XdUxnbW8" -O ngsplotdb_hg19_75_3.00.tar.gz && \
    echo y | ngsplotdb.py install ngsplotdb_hg19_75_3.00.tar.gz && \
    rm -rf  ngsplotdb_hg19_75_3.00.tar.gz && \
    wget "https://drive.google.com/uc?export=download&id=0B5hDZ2BucCI6S3E4dVprdlF2YW8" -O ngsplotdb_hg38_76_3.00.tar.gz && \
    echo y | ngsplotdb.py install ngsplotdb_hg38_76_3.00.tar.gz && \
    rm -rf  ngsplotdb_hg38_76_3.00.tar.gz && \
    wget "https://drive.google.com/uc?export=download&id=0B5hDZ2BucCI6NXNzNjZveXdadU0" -O ngsplotdb_mm10_75_3.00.tar.gz && \
    echo y | ngsplotdb.py install ngsplotdb_mm10_75_3.00.tar.gz && \
    rm -rf  ngsplotdb_mm10_75_3.00.tar.gz 

RUN git clone https://github.com/kundajelab/phantompeakqualtools  && \
    mv phantompeakqualtools /opt/  && \
    echo 'library(caTools)' | cat - /opt/phantompeakqualtools/run_spp.R > temp && mv temp /opt/phantompeakqualtools/run_spp.R && \
    chmod 755 /opt/phantompeakqualtools/* && \
    echo 'alias run_spp.R="Rscript /opt/phantompeakqualtools/run_spp.R"' >> ~/.bashrc 
ENV PATH=${PATH}:/opt/phantompeakqualtools 

Need to think hard about how to handle the reference genomes for ngsplot.

[ewels]

The three ngsplot references are actually not super huge:

35M ngsplotdb_hg19_75_3.00.tar.gz
34M ngsplotdb_hg38_76_3.00.tar.gz
21M ngsplotdb_mm10_75_3.00.tar.gz

That's 91M in total, a lot less than I was expecting. This is small enough that it may be possible to add these to the bioconda recipe (in discussion with that community). Alternatively, we could consider adding these to the pipeline repo. Finally, it would be very easy to add these to the AWS-iGenomes reference.

[ewels]

xref #10 (comment):

We just need to keep in mind that the run_spp.R has a line missing (library(caTools)) to make it work.

This is something that we can add in using a patch in the bioconda release.

@apeltzer - is this packaging something that you're keen / able to work on?

[apeltzer]

Hi everyone! I just asked in the bioconda gitter whether they feel its alright and will do it once confirmed that its okay to package something like this!

Absolutely happy to create a bioconda package for that specific purpose - shouldn't be too difficult to do that!

[apeltzer]

So all conda packages are there and I added run_spp.R to the bin folder of the bioconda branch.
Will adjust calls to tools now!

[ewels]

Sounds good 👍