Normalizing autosomes part of predict command not working

[Stikus]

Hello again, we have another problem.

Installed packages:

joblib-1.4.2
numpy-2.1.1
pandas-2.2.3
pysam-0.22.1
python-dateutil-2.9.0.post0
pytz-2024.2
scikit-learn-1.5.2
scipy-1.14.1
threadpoolctl-3.5.0
tzdata-2024.2
wisecondorx-1.2.7 

WisecondorX is the latest version, but log show 1.2.7 (issue #126)

Error log:

  Command: 'WisecondorX predict /outputs/inputs/NPZ/E07002_normal_alignment.npz /outputs/reference.npz /outputs/E07002_normal_alignment --plot --bed'.
  PID=429 (last job)
[INFO - 2024-10-03 18:02:48]: Starting CNA prediction
[INFO - 2024-10-03 18:02:48]: Importing data ...
[INFO - 2024-10-03 18:02:48]: Normalizing autosomes ...
Traceback (most recent call last):
  File "/usr/local/bin/WisecondorX", line 8, in <module>
    sys.exit(main())
  File "/usr/local/lib/python3.10/dist-packages/wisecondorx/main.py", line 487, in main
    args.func(args)
  File "/usr/local/lib/python3.10/dist-packages/wisecondorx/main.py", line 187, in tool_test
    results_r, results_z, results_w, ref_sizes, m_lr, m_z = normalize(
  File "/usr/local/lib/python3.10/dist-packages/wisecondorx/predict_control.py", line 32, in normalize
    sample = project_pc(sample, ref_file, ap)
  File "/usr/local/lib/python3.10/dist-packages/wisecondorx/predict_tools.py", line 61, in project_pc
    transform = pca.transform(np.array([sample_data]))
  File "/usr/local/lib/python3.10/dist-packages/sklearn/utils/_set_output.py", line 316, in wrapped
    data_to_wrap = f(self, X, *args, **kwargs)
  File "/usr/local/lib/python3.10/dist-packages/sklearn/decomposition/_base.py", line 139, in transform
    xp, _ = get_namespace(X, self.components_, self.explained_variance_)
AttributeError: 'PCA' object has no attribute 'explained_variance_'

Can anyone help?

[Stikus]

After some debug we've found, that with scikit-learn<1.4.0 all tests are good, with scikit-learn<1.5.0 there are small numerical changes (rounding) and with scikit-learn==1.5.0 we got this error.

I assume that some breaking changes for this code happened in v1.5.0. Maybe these: https://scikit-learn.org/stable/whats_new/v1.5.html#id9

I think that repository need either pin scikit-learn<1.5.0 or update code to modern standards.