Rewrote docker containers using bioconda environment.yml ?

[maxulysse]

cf SciLifeLab/NGI-RNAseq#216

[maxulysse]

Tried to update our most complicated container (the one with the most tools: sarek) using this method.
First the pros:

• I like the simplicity of the method.
• No need to tailor the installation for each tools.

But the miniconda base image (234 MB) is quite larger than the base debian:8 image (53 MB).
So the resulting image is obviously larger (470 MB) than what we have (245 MB).

What do you think?

[apeltzer]

Personally, I'd say this is an improvement, as its much simpler and relies on packages that aare ultimately tested by everyone in the bioconda community, which should in general be more reproducible and reliable. I see your point with 200+MB more for the image, but wouldn't see this as the biggest issue here - its not 1GB+ and most people won't really care about 200MB more to download.

[marcelm]

You may be able to reduce the size somewhat by putting an && conda clean -a after the RUN conda install ...

[maxulysse]

Good point, I'll try that ;-)

[maxulysse]

@apeltzer
My concern is not the fact that it is 200+MB but more that it is almost twice the amount.
But regarding that I have other containers that are much more bigger on a higher order of magnitude, I can work with that.

[maxulysse]

@marcelm tried it, and it's now 387 MB so almost 100 MB down, which I can accept given the simplicity it will give us in installing all the tools.

[ewels]

For what it's worth, I've ported the MethylSeq and RNA pipelines to this method now and I'm totally in love with it 😄 Especially with the cool nf-core lint . command which warns you if there are newer versions of any packages available...

[maxulysse]

closed by #607