fix: bug in compute_transition_probabilities

[ludvigla]

Description

The transition probabilities are invalid at k > 1 in compute_transition_probabilities which can influence the computation of local G scores. This PR contains a fix for this bug along with some additional minor fixes:

Changes

• correctly row-normalize W_out matrix after matrix power computation
• handle situation where use_weights=False to match R implementation
• set normalize_counts=False as default to match R implementation

Fixes: exe-1867

Type of change

Please delete options that are not relevant.

• Bug fix (non-breaking change which fixes an issue)

PR checklist:

• This comment contains a description of changes (with reason).
• I have performed a self-review of my own code
• I have made corresponding changes to the documentation
• My changes generate no new warnings
• I have added tests that prove my fix is effective or that my feature works
• If a new tool or package is included, I have updated poetry.lock, and cited it properly
• I have checked my code and documentation and corrected any misspellings
• I have documented any significant changes to the code in CHANGELOG.md

[codecov]

Codecov Report

Attention: Patch coverage is 44.44444% with 5 lines in your changes missing coverage. Please review.

Project coverage is 81.78%. Comparing base (4404d87) to head (dbef2f4).

Files	Patch %	Lines
src/pixelator/graph/node_metrics.py	44.44%	5 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##              dev     #159      +/-   ##
==========================================
- Coverage   81.78%   81.78%   -0.01%     
==========================================
  Files         121      121              
  Lines        6787     6795       +8     
==========================================
+ Hits         5551     5557       +6     
- Misses       1236     1238       +2     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[ambarrio]

LGTM. Only comment is how did you discover this? Can you in this case include maybe a test to collect this knowledge that needs to always pass?

[ludvigla]

LGTM. Only comment is how did you discover this? Can you in this case include maybe a test to collect this knowledge that needs to always pass?

I discovered it when creating the local G tutorial for Python because the p-values didn't match with R. These things are quite difficult to detect which is why I think it's so important that more people play around with the method in their analysis.

I will add a test!

[johandahlberg]

LGTM