Summary

There are two implementations of get_analyses_by_coordinate in the Studyset class with different signatures and behaviors. The latter definition overrides the former, making the first implementation effectively dead code.

Details

The first implementation:

• Accepts xyz and radius r
• Uses scipy.spatial.distance.cdist

The second implementation:

• Accepts xyz with either r (radius) or n (number of nearest analyses)
• Uses manual Euclidean distance computation
• Overrides the first definition entirely

This leads to:

• Dead/unreachable code
• Potential confusion for contributors and users
• Inconsistent API design

Proposed Solution

• Remove the duplicate implementation
• Keep a single unified method supporting both:
  • radius-based filtering (r)
  • nearest-neighbor selection (n)
• Add validation to ensure only one of r or n is provided
• Update docstrings accordingly

Additional Suggestions

• Add tests for both modes (r and n)
• Consider deprecating older behavior if needed

Why this matters

This improves:

• API consistency
• Code maintainability
• Developer experience

Happy to open a PR for this if the approach is approved.