Fix KeyError in statistical analysis helper for large datasets

[Efdix]

Fix KeyError in cpdb_statistical_analysis_helper with large datasets

Description

This PR fixes a KeyError encountered when running cpdb_statistical_analysis_method.call on large datasets (e.g., ~500k cells).

Issue

When processing wide DataFrames (many columns/cells), counts.set_index('id_multidata', inplace=True, drop=True) in cpdb_statistical_analysis_helper.py was causing all columns (cell barcodes) to be lost/dropped from the DataFrame, leading to a KeyError in the subsequent step counts = counts[cells_names].

Traceback:

KeyError: "None of [Index(['...'], dtype='object', length=493019)] are in the [columns]"

Fix

Replaced the set_index(inplace=True) operation with a manual index assignment:

counts.index = counts['id_multidata']
counts.index.name = 'id_multidata'

This approach avoids the internal Pandas issue associated with set_index on very large DataFrames and correctly preserves the columns.

Verification

• Verified that the KeyError is resolved and the analysis proceeds correctly with the large dataset.
• Verified that the logic remains equivalent for smaller datasets.

Environment

• CellphoneDB version: 5.0.1
• Pandas version: 2.3.3
• Python version: 3.14.2
• Dataset size: ~500k cells