Can't find reference genomes for ngsplot

[orzechoj]

When I run on ngs.plot.r (using the hg19 genome), I get the following error:

 Error in CheckRegionAllowed(reg2plot, default.tbl) :
    Unknown region specified. Must be one of: bed
  Execution halted

It's the command

 ngs.plot.r \
      -G hg19 \
      -R genebody \
      -C ngsplot_config \
      -O Genebody \
      -D ensembl \
      -FL 300

throwing the error.

Some poking around, from within the singularity container

Running ngsplotdb.py shows that no genomes are installed for ngs.plot:

> ngsplotdb.py list
ID   Assembly Species  EnsVer   NPVer    InstalledFeatures    
>

But loading the nfcore-chipseq-1.0dev environment doesn't solve this, even though the hg19 genome is included the environment:

> source activate nfcore-chipseq-1.0dev
(nfcore-chipseq-1.0dev) > ngsplotdb.py list
ID   Assembly Species  EnsVer   NPVer    InstalledFeatures    
(nfcore-chipseq-1.0dev) > conda list
...
r-ngsplot                 2.63                          1    bioconda
r-ngsplotdb-hg19          3.00                          2    bioconda
r-ngsplotdb-hg38          3.00                          2    bioconda
r-ngsplotdb-mm10          3.00                          1    bioconda
...

[apeltzer]

Looks like I need to have a look at the conda package again...

[apeltzer]

Package updated, I think the reason was simply the missing -y parameter when installing the databases. Lets see if that helps. There also might exist some issues with Python, as this package requires strictly python2.7 to be present and uses python in the respective script headers, so I might need to update that too - I will test this first and then fix the remaining issues on the go :-)

[ewels]

Sounds good, thanks! 👍

I'm slightly paranoid that this could also be affected by our conda ENV hack (no source activate so no pre/post- hook activations). But hopefully it'll be fine..

[apeltzer]

To be 100% sure, I also added the python <3.0a0 requirement to the file. It just feels safer and better in this context.

Yeah, I also had that feeling. Which could make things a bit more complicated in the installation unfortunately :-(

[ewels]

If anything like that does come up, then we should be able to do a workaround by mimicking those actions in the Dockerfile and Singularity file I figure.

[apeltzer]

Yup, I think that should be possible!

[chuan-wang]

Hello,

I found the following solution in case it is of any help:

In the ngsplotdb.py and setTableDefaults.py files change the first line from
#!/usr/bin/env python

to

#!/usr/bin/env python2

/chuan

[apeltzer]

Hi @chuan-wang ,
I just merged in the required changes in bioconda, which should hopefully fix things without requiring that change (enforcing python2.7 to be used for the package there).

bioconda/bioconda-recipes#10406

This might take a while to be present in bioconda, then requiring us to rebuild the Docker Image and the Singularity Hub image, but should install fine after that. At least I was able to install properly after that :-)

[chuan-wang]

@apeltzer Good to know! Thanks for the work! XD

[apeltzer]

Still need to do some fixes with bioconda. Slighly more complex as they blacklisted all R=3.4.1 packages right now and I had to get around this first of all.
Will keep you posted here and its high up my priority list...

[chuan-wang]

No hurry! We are still working on some bugs in the pipeline.

[chuan-wang]

Hi @apeltzer, just let you know that we are replacing the function of ngsplot with deepTools in PR #51. Hopefully you haven't spent too much time on fixing the ngsplot issue...Sorry that I should have told you earlier.

[apeltzer]

Hi @chuan-wang , no worries - I'M skipping fixing that package now then and am more than happy to be able to do so...

[chuan-wang]

@apeltzer OK! XD