Problem with installation

[brendaferrari]

I am writing this issue just to let you know about a problem a encountered when installing POVME.

I followed the instructions in the README.md file. When I tried to run the detect workflow with:

povme detect -c config.yaml -i protein.pdb -o output/

I got the following output in the terminal:

Traceback (most recent call last):
  File "<frozen runpy>", line 198, in _run_module_as_main
  File "<frozen runpy>", line 88, in _run_code
  File "C:\devel\anaconda3\envs\spherecon\Scripts\povme.exe\__main__.py", line 4, in <module>
  File "C:\devel\anaconda3\envs\spherecon\Lib\site-packages\povme\cli.py", line 4, in <module>
    from povme.pocket.detect import PocketDetector
  File "C:\devel\anaconda3\envs\spherecon\Lib\site-packages\povme\pocket\detect.py", line 5, in <module>
    from pymolecule import Molecule
ModuleNotFoundError: No module named 'pymolecule'

I was able to find the module in one of your repositories. I installed it separately and now POVME was able to generate the pocket files.

However, at the end of the run, the terminal gave me the following output

Removing hydrogen atoms...
Calculating the convex hull of the PDB file...
Adding points...
C:\devel\anaconda3\envs\spherecon\Lib\site-packages\scipy\_lib\_util.py:440: UserWarning: One of the clusters is empty. Re-run kmeans with a different initialization.
  return fun(*args, **kwargs)
C:\devel\anaconda3\envs\spherecon\Lib\site-packages\numpy\_core\fromnumeric.py:3860: RuntimeWarning: Mean of empty slice.
  return _methods._mean(a, axis=axis, dtype=dtype,
C:\devel\anaconda3\envs\spherecon\Lib\site-packages\numpy\_core\_methods.py:137: RuntimeWarning: invalid value encountered in divide
  ret = um.true_divide(

I am reporting this as I do not know if this is the expected behaviour.

Please let me know if you need any further info about the outputs.

Thanks in advance!

[aalexmmaldonado]

Hey @brendaferrari, thanks for letting us know about the missing pymolecule package. I have fixed that and updated the README. Could you please provide me with the configuration and PDB files you are using?

[brendaferrari]

Hi @aalexmmaldonado thank you very much for your response! The PDB file I used when this happened was this one: 5CSN and the configuration file was the same one that is provided in one of the tutorials:

pocket_detection_resolution: 4.0
pocket_measuring_resolution: 1.0
clashing_cutoff: 3.0
n_neighbors: 4
n_spheres: 5
sphere_padding: 5.0
use_ray: false
n_cores: 4

I ran the workflow with other PDB files but this prompt about one of the clusters being empty just appears for this 5CSN structure.

[aalexmmaldonado]

I could reproduce the same warnings with your structure and configuration file.

These warnings occur when we try to cluster the points into n_spheres groups with K-means when one of the groups is empty. In these cases, I retry K-means with a different initialization algorithm. If that still fails, we use all the points as the spheres. This only happens with tiny pockets that are unlikely relevant, so this should not tangibly impact your results.

Please let me know if there is anything else you ran into that I can improve. This package will be a priority for adding new features soon. Thanks!

[brendaferrari]

Thank you very much for looking into this issue and explaining it to me.

I am glad to know that you will continue to maintain this package, if I observe anything else I will let you know.