search

YeoLab / eCLIP

Other
38 stars 26 forks source link

Docker workflow for unsupported species in clipper #12

Closed peaceben closed 4 years ago

peaceben commented 4 years ago

Is there any way to use the Docker-based workflow (using e.g. ./paired_end_clip.yaml) for a species that is not natively supported by clipper?

I read the clipper documentation and build the required annotation files for Arabidopsis in my case. A new installation with the custom annotation files should provide support for the species of interest. But this probably doesn't help when using Docker images, if i understood it correctly?

What approach would you recommend in this particular case? :+1:

Best regards, Ben

byee4 commented 4 years ago

You will have to add the new species annotation to Clipper and build a new Docker image and specify that version instead.

peaceben commented 4 years ago

Hi!

Any chance you could provide the Dockerfile for brianyee/clipper:997fe25532a5bdcf5957f2a467ca283bbd550303 ? As the Dockerfile in the repository is not working when i try to build a new image.

Best, Ben

byee4 commented 4 years ago

Dockerfile.txt If that doesn't work you can probably get away with using the latest version as the base image. Otherwise, might need to install the python3 Clipper

peaceben commented 4 years ago

Hi thanks! What worked for me was this modified version of the clipper3 dockerfile:

Dockerfile.txt

peaceben commented 4 years ago

But i still end up with the error: [job step_clipper] Job error: Error collecting output for parameter 'output_tsv':

But also with these files in /.tmp/... .peakClusters.bed .peakClusters.bed.pickle

Seems to me like an error related to CWL/Docker as Clipper generates output in the previous mentioned files.

I should mention i had to fix another thing before in clipper/src/main.py line 146; as in python3 binary has to be defined here:

with open(options.outfileF + ".pickle", 'w') as f: to with open(options.outfileF + ".pickle", 'wb') as f:

Do you have any approaches in mind? I would be very grateful, as i have the feeling that everything is working properly now.

Best, Ben

peaceben commented 4 years ago

What worked for me was removing the tsv output in the clipper.cwl file. The eclip pipeline is completing now without error. Thanks!