volkamerlab / kissim Goto Github PK

Add text (i.e. angle) to CA atoms using add_label. Not needed any more, since we are color-coding residues by categories (e.g. based on angles).

Change default for

Fingerprint.values_array(True, False. True)

to

Fingerprint.values_array(True, True, True) 

Hi. I'm hyunseung kong.

https://kissim.readthedocs.io/en/latest/tutorials/api.html#Run-encode-function
Here I found what appears to be a typo in the example code.

I think,
"../../kissim/tests/data/KLIFS_download/"
needs to be change
"kissim/tests/data/KLIFS_download/"

This is a minor, but I'm informing you.

Use interrogate to check docstring coverage; improve where needed.
https://calmcode.io/shorts/interrogate.py.html

Add interrogate badge to README :)
https://interrogate.readthedocs.io/en/latest/#other-usage

Currently e.g. empty fingerprints when structure contains modified residues (must throw an error somewhere). Catch these errors appropriately.

Thought this was fixed already - revise code!

Description

Bump up env/CI to Python 3.8, 3.9, and 3.10.
Triggered by new pandas release 2.0.0, which is only available for Python >= 3.8, see https://pandas.pydata.org/docs/getting_started/install.html#python-version-support

Todos

• Apply latest cookiecutter-cms template
  • TBD: Bump up CI to Python 3.8, 3.9, and 3.10.
  • TBD: Bump up env to >= 3.8
• Update pandas API to 2.0.0: https://pandas.pydata.org/docs/dev/whatsnew/v2.0.0.html
  • pd.append to pd.concat

Use only a subset of residues in the kissim fingerprint in the comparison step.

For instance, the subset could contain ligand-interacting residues only:

• For each kinase-ligand combo calculate the average contacts
• Based on (a) DFG-in, (b) DFG-out, or (c) all DFG profiles, extract residues that are in contact with any ligand in at least 1% of the kinases (or structures/monomers?)

List provided by @AJK-dev:

DFG-in structures
1,2,3,4,5,6,7,8,9,10,11,12,13,15,16,17,19,20,21,24,25,27,28,36,37,38,43,44,45,46,47,48,49,50,51,52,54,55,59,70,72,74,75,76,77,80,81,82,83,84,85

DFG-out structures
3,4,5,6,7,8,9,11,12,13,15,16,17,19,20,21,23,24,25,27,28,31,35,36,37,38,43,44,45,46,47,48,49,50,51,52,54,55,60,61,64,66,67,68,69,70,74,75,77,79,80,81,82,83,84,85

Min 1% in any DFG profile
1,2,3,4,5,6,7,8,9,10,11,12,13,15,16,17,19,20,21,23,24,25,27,28,31,35,36,37,38,41,43,44,45,46,47,48,49,50,51,52,53,54,55,56,59,60,61,64,66,67,68,69,70,72,74,75,76,77,79,80,81,82,83,84,85

Available:

FingerprintDistanceGenerator.from_feature_distances_generator()
FingerprintDistanceGenerator.from_structure_klifs_ids()

Add

FingerprintDistanceGenerator.from_fingerprint_generator()

Although this data is not used any more, update file (according to corrections in side chain angle calculation).

Ignore unit tests in CodeCov.

Currently, we generate fingerprints from structure KLIFS IDs.
Add API to generate fingerprint from a PDB ID + indices for 85 binding site residues.

• Rename kissim.io.data.KlifsToKissimData to KissimData with the following methods
  • KissimData.from_structure_klifs_id(structure_klifs_id, klifs_session)
  • KissimData.from_pdb(pdb_id_or_file, chain_id, alternate_model_id, residue_ids, residue_ixs, kinase_name)
  • Update kissim.io.biopython.PocketBioPython and kissim.io.dataframe.DataFrame
• Rename KissimData attribute structure_klifs_id > structure_id
• Add Fingerprint.from_pdb
• Add FingerprintGenerator.from_pdb
• Add kissim.encoding.features.BaseFeature.from_pdb

In order to save some resources, schedule automatic GHA CI for Tuesdays (not daily).

cron: "0 0 * * *" > cron: "0 0 * * TUE"

See this issue: volkamerlab/opencadd#126

Update docs here: https://kissim.readthedocs.io/en/latest/installing.html

I think my ipywidgets pinning (see #113) has no effect (due to package order in env file) - revise this and understand why the kissim tests are not failing.

See issue here: volkamerlab/cicag_klifs_workshop#2

Change default value for feature_weights in FingerprintDistanceGenerator.from_feature_distances_generator() to our defaults (TBD).

Remove structures with reference point outliers (front pocket, DFG). These are other structures than those removed in postprocessing (distances >35) - those had no actual pocket any more.

Add the default (1) kinase matrix, (2) kinome tree and (3) tree-related kinase annotation as output files to the compare API/CLI.

Add unit tests for the cli module!

I have not figured out yet how to test the CLI functions properly with pytest.

Existing tests caused problems with codecov and I do not understand why, so I am not trusting them for now, see PR #33.

CI fails for windows, however it seems to be a connection problem (bravado / KLIFS). Remove windows for now.

Broken notebooks using ipywidget -> nglview (AttributeError)

Same issue as in volkamerlab/teachopencadd#262 and volkamerlab/kissim_app#60:

• Wait for nglview to pin the ipywidget version
• Pin it as we did in teachopencadd: nglviewer/nglview#1033 (comment)
• Unpin for new nglview release with their update

Fix diagonal values for

fingerprint_distance_generator.kinase_distance_matrix()

Do not set to 0.0!

Add cutoffs to definitions.py:

Generating some fingerprints fails because of an error (maybe a bug?) in Bio.PDB.HSExposure as reported here:
biopython/biopython#3469

For now, remove these structures from fingerprints generation (check how many structures are affected).

With latest matplotlib version 3.4, the tick labels for the cmap colorbar for categorial features are mixed up:

• Number of colors in cbar incorrect (see image); matplotlib/matplotlib#21870
• For now, pin matplotlib to <3.5; #101
• Formatter must be used together with locator: /Users/dominique/Documents/github/kissim/kissim/viewer/base.py:451: UserWarning: FixedFormatter should only be used together with FixedLocator ax.set_xticklabels(xticklabels)

Currently, standalone calculation (from scratch). Check dtypes in pandas.DataFrame (centroid distances float64, other distances object)

See cimm-kzn/3D-MIL-QSAR#20

Follow instructions: https://stevenmortimer.com/5-steps-to-change-github-default-branch-from-master-to-main/