luntergroup / octopus

Bayesian haplotype-based mutation calling
MIT License
302 stars 38 forks source link

Make forest files more accessible #109

Closed tfenne closed 3 years ago

tfenne commented 4 years ago

Would it be possible to make the pre-build forest files more accessible please? If not doing the full install with scripts/install.py then it's rather hard to figure out where the forests are and download them.

It would be great if one of the following could be done:

  1. Have a mode in install.py that just downloads the forest files and isn't dependent on having run the make/cmake build already
  2. Make the google bucket/folder where the forest files are publicly listable so it's easy to view in a browser and grab the right files
  3. Have up to date links in the readme to the forest files for the last 2-3 version

My tendency has been to install/build octopus semi-manually using the OS's package manager to install dependencies and avoiding homebrew, but then I get stuck trying to guess the right filenames for the forest files and can't tell if I'm guessing wrong or if there isn't a forest file for the version currently on develop. Based on the version in CMakeLists.txt I was expecting this path to exist but it does not: https://storage.googleapis.com/luntergroup/octopus/forests/germline.v0.7.0.forest.

dancooke commented 4 years ago

Hi Tim, I've made the 'octopus' bucket on Google Cloud public. I'll improve the docs with the next release.

tfenne commented 4 years ago

Thanks @dancooke !

DBS-Max commented 4 years ago

Should I be using the 0.6.3b forest with the current develop branch? Or should I build from the 0.6.3b branch? The 0.7.0 forest files that the develop branch says it downloads aren't actually downloaded.

dancooke commented 4 years ago

@DBS-Max The 0.6.3b forests won't work with the develop branch version, and the v0.7.0 forests aren't available yet. So either you can train your own forest with the develop branch version, or just stick with v0.6.3b until v0.7.0 is released.

DBS-Max commented 4 years ago

Sounds good, thanks for clarifying.