Conda - use Python 3

[ewels]

We noticed in nf-core/methylseq#123 that the conda environments we're using everywhere are still using Python 2.7. This PR shifts the conda environments to Python 3.7

• Update base Dockerfile to use miniconda3, and update to conda 4.7.12
• Add Python 3.7 dependency to the template environment.yml file

Note that this is quite a big change! It could impact a lot of pipelines that are using Python tools.. However, if pipelines still require Python 2.7, they can change the dependency in the environment.yml file to 2.7 and stuff should still work.

[drpatelh]

Think it's probably the right time to bump this up 👍 Note though I am still able to get a Python 3 conda environment with the current set-up. Had to deal with the switch when updating all the software for the chipseq and atacseq pipelines.

[drpatelh]

If I remember correctly this is the only line I had to change in the template to get things running with Python 3:
https://github.com/nf-core/chipseq/blob/1939e4842dbc17e7f5b370c771802948271dde11/bin/scrape_software_versions.py#L57

[drpatelh]

Maybe worth creating a dummy pipeline from the template and testing though?

[ewels]

I think it's ok as long as we're clear about what people should do if stuff breaks...

dependencies:
  - python=2.7
  #..everything else as before

This should still work, even if the base environment is python 3 ^

[ewels]

(but yes, we should make sure that at least the template scripts work with py3 and preferably both if they don't already)

[apeltzer]

Just opened PR #476 to fix the issue - then things work fine even in miniconda3 and this PR should be good t o go I think?

[apeltzer]

LGTM now!