CellColoc is an open-source Python workflow designed to standardize and make transparent common microscopy image-analysis tasks that are often performed via ad hoc approaches. The workflow focuses on segmentation-based cell colocalization, single-channel segmentation, and per-cell feature extraction for both 2D and 3D fluorescence microscopy images. By separating reusable analysis logic from project-specific configuration, CellColoc aims to provide an extensible foundation for community-driven microscopy workflows that require clear per-cell overlap classification and morphology readouts.
The authors emphasize that conventional pipelines frequently mix image loading, segmentation, region selection, manual inspection, and spreadsheet post-processing in ways that are difficult to transfer across projects and can obscure how intermediate results were produced. CellColoc addresses these issues by providing a modular, script-driven layer that consolidates reusable analysis steps while preserving transparency of intermediate artifacts.
CellColoc implements a modular workflow layer that integrates externally developed segmentation backends into reusable analyses. The project retains a core package with reusable functionality and expects project-specific analyses to be executed from concise user scripts. This design choice promotes reproducibility and reuse by making the analysis logic portable while keeping configuration and execution simple for distinct projects.
Important design elements described by the authors include a standardized results directory for writing intermediate artifacts and outputs, support for channel-wise selection of segmentation backends, and facilities for inspecting and refining intermediate masks. These elements are intended to make it easier to trace how final tabulated results were produced.
CellColoc integrates multiple segmentation backends. The abstract explicitly names Cellpose and threshold-based methods as supported backends. Users can select backends on a per-channel basis, enabling flexible combinations of approaches across multi-channel datasets.
The package provides a mechanism for cached post hoc refinement of Cellpose thresholds, allowing users to adjust segmentation parameters after initial runs without repeating all upstream steps. The workflow also supports reanalysis after manual mask editing, so human-curated corrections to segmentation masks can be incorporated into downstream quantitative outputs.
CellColoc supports interactive or reusable regions of interest (ROIs). This permits per-project choices between manual region selection and scripted reuse of the same ROI definitions across multiple images or batches.
For volumetric imaging, the pipeline offers z-cropping and z-projection options to accommodate common 3D-to-2D preprocessing strategies. Optional third-channel occupancy and cell-positivity analysis are available to support workflows that require evaluating additional markers or occupancy measures per cell.
Reanalysis after manual mask editing is explicitly supported, which helps integrate manual quality control steps without losing provenance for intermediate artifacts.
A central principle of CellColoc is producing inspectable intermediate artifacts and standardized structured outputs. The workflow writes ROI masks, per-channel label masks, positive-cell masks, and structured result tables to a standardized results directory. This makes intermediate data available for inspection and helps reproduce results from raw inputs through to final tables.
The authors note that this artifact-centric output design promotes transparent inspection, reproducibility, and reuse aligned with FAIR principles. By standardizing where and how intermediate files are stored, analyses performed with CellColoc should be easier to audit and re-run by others or by the original analysts at a later date.
CellColoc is distributed with public example datasets and a synthetic benchmark to support validation of the workflow and to illustrate typical usage patterns. Archived software releases accompany the package to facilitate reproducible software environments for downstream users. These resources are intended to help users assess performance and adopt the workflow in their own projects.
Analyses are executed from concise user scripts while the core package retains reusable functionality. This model separates project-specific analysis configuration from core processing logic and encourages modular reuse of functionality across projects. Channel-wise backend selection, cached parameter refinement, and a reanalysis pathway after manual edits are among the features designed to support iterative and practical microscopy workflows.
Because CellColoc integrates existing segmentation tools and exposes configuration at the script level, it is positioned as an extensible foundation for community-driven microscopy workflows rather than a closed, monolithic application.
The authors have declared no competing interests. The work is released as a preprint on bioRxiv and has not been peer-reviewed. The article lists authors and their affiliation with the German Center for Neurodegenerative Diseases (DZNE) in Bonn, and provides a DOI and posting date as part of the preprint record. Users should note the preprint status when interpreting the work.