Support for AlphaFold database

[sirius777coder]

Hi,

I think Biotite is a really useful tool when I deal with a lot of biological data. Do your team has any plan to integrate AlphaFold database into the biotite API?

[padix-key]

Hi, currently it is not planned to add a direct interface to the AlphaFold Protein Structure Database (https://alphafold.ebi.ac.uk/). To my knowledge this database has no documentation for a REST API to search AlphaFold structures and fetch them. The RCSB PDB started to intgrate AlphaFold structures itself (https://www.rcsb.org/docs/general-help/computed-structure-models-and-rcsborg), so probably download of AlphaFold structures will be managed via biotite.database.rcsb in the foreseeable future.

[sirius777coder]

OK, look forward to seeing this!

[BradyAJohnston]

AlphaFoldDB officially released docs and better support for their API a few days ago. I was going to have a go at it, but I don't really know my way around APIs much at all. Thought I would share the links to potentially have it integrated into biotite.

https://www.ebi.ac.uk/about/news/updates-from-data-resources/alphafold-database-ux-update/

https://alphafold.ebi.ac.uk/api-docs

[dacarlin]

I would be happy to contribute an interface to the AlphaFold DB, I am familiar with the APIs. Is this something the project would welcome? I would be a first time contributor to biotite (long time user)

[padix-key]

Yes, indeed! In the moment there is already a contributor that was working in this in #465. However, the last commit is multiple months ago, so I expect the contributor cancelled the project. I would like to give the author another week to respond. Otherwise I would close his PR and give way to you to work on this feature.

I would imagine an interface that is similar to biotite.database.pdb (at least for fetch()). Although for license reasons you cannot simply copy the code from #465, you can still take advantage of the discussion as guideline.

[dacarlin]

That all sounds good to me 👍 Agree that the interface for biotite.database.pdb is a perfect model. How about I check back next week for an updated status?

[padix-key]

There was no response, yet, so I closed #465. Hence you can start if you like.

[padix-key]

Now @jonfunk21 also approved, that you may take on the issue.

[dacarlin]

That is wonderful, thanks for the update! I will plan on working on this over the next week or so. Will report back once I have something to show.

[dacarlin]

I’ve implemented fetching and included a few basic tests so far (see here https://github.com/dacarlin/biotite/blob/add-alphafold-db/tests/database/test_alphafold.py).

Shall I open a WIP merge request as I wrap up the changes and get the code ready for your comments? Currently, I have my changes in a branch that is in a recent fork

[padix-key]

Looks already quite good, feel free to open a PR. I wonder if the tests pass, since there are two issues, if I am not wrong.

content = file_response.text should not work for bcif because it is a binary format.

from .check import assert_valid_response should give an error because there is no check.pymodule in this subpackage.

[dacarlin]

Yes, thanks! You are definitely correct, I ran into issues before I finished writing the tests which led me down a rabbit hole.

Briefly, I created a new Conda environement using the environment.yml but I still got some errors from Cython that are preventing me from installing with pip install . and pip install -e .. I put the error log in this gist https://gist.github.com/dacarlin/bac823ef62aef6d3df7261ad8dd6d76a in case it's helpful. I ended up adding explicit casts to int in 4-5 lines of src/biotite/structure/io/pdb/hybrid36.pyx to get the install to work.

Good news is now I can write proper tests!

[padix-key]

I suppose you probably have the brand new Cython 3.0 installed? I just saw that the environment.yml does not restrict the upper bound of the Cython version so you probably have installed version 3.0, that introduced some breaking language changes. It quite relieves me to hear, that only a few lines of code seem to require fixing. I will create an issue for the changes introduced by Cython 3.0.

[dacarlin]

Glad that it's an easy fix. This is the output of conda env export --from-history in case it helps troubleshoot

name: biotite-dev
channels:
  - defaults
dependencies:
  - sphinx-gallery=0.11.1
  - tantan
  - matplotlib[version='>=3.3']
  - networkx[version='>=2.0']
  - muscle=3
  - setuptools[version='>=30.0']
  - mafft
  - sphinx[version='>=5.0']
  - numpy[version='>=1.15']
  - numpydoc[version='>=0.8']
  - requests[version='>=2.12']
  - sra-tools
  - dssp
  - wheel[version='>=0.30']
  - autodock-vina
  - scipy[version='>=1.8.0']
  - pytest[version='>=5.2']
  - scikit-learn[version='>=0.18']
  - pydot[version='>=1.4']
  - mdtraj[version='>=1.9.3']
  - msgpack-python[version='>=0.5.6']
  - clustalo
  - cython[version='>=0.29']
  - viennarna[version='>=2.5.0']
  - python=3.9
  - sphinxcontrib-bibtex[version='>=2.3']
  - pip[version='>=10.0']
  - cmake

Meanwhile, I'll create a PR for this feature branch 😄