CellTune: an integrative software for accurate cell classification in spatial proteomics

Area of Science:

• Proteomics
• Computational Biology
• Cell Biology

Background:

• Spatial proteomics captures tissue complexity but faces challenges in cell classification.
• Lack of efficient algorithms, annotation tools, and labeled datasets hinders computational method benchmarking.

Purpose of the Study:

• Introduce CellTune, an integrated software for precise cell classification in large spatial proteomics datasets.
• Provide a user-friendly, code-free interface for advanced data analysis.

Main Methods:

• Developed CellTune, an active learning workflow for human-in-the-loop cell classification.
• Created CellTuneDepot, a resource with 40,000 manually annotated and 3.5 million labeled cells across 60 cell types.

Main Results:

• CellTune achieves classification accuracy comparable to human performance.
• The software enables increased classification resolution and discovery of novel cell types.
• CellTune outperforms alternative computational methods.

Conclusions:

• CellTune and CellTuneDepot offer a powerful tool for state-of-the-art cell classification at scale.
• This resource drives biological insights by enhancing accuracy and resolution in spatial proteomics analysis.