UNHUDDLE

UNHUDDLE (Uncovering Neighborhood Heterogeneity Using Deterministic Normalization and Local Equilibrium) resolves signal in densely packed tissue regions, or “cell huddles,” in multiplex spatial proteomics.

It uses each cell’s local neighbourhood to estimate competing claims on shared border-pixel intensity and reallocates that signal to the most likely source. The pipeline also provides optional cohort-level denoising, configurable normalization, morphometrics, quality control and export to an analysis-ready AnnData object.

The repository includes a command-line interface, tutorials, demonstration data, optional DeepCell–Mesmer segmentation integration, PENGUIN preprocessing support, benchmarking material and a JOSS manuscript draft.

Version: 1.0.0
License: MIT
Persistent identifier: 10.5281/zenodo.21970599
Version DOI: 10.5281/zenodo.21970600