Aegle is a 2D data process and analysis pipeline for 2D PhenoCycler images.
This pipeline is developed as part of the PennTMC project.
Aegle is named after the Greek goddess of radiant good health. Inspired by her name, this pipeline aims to bring clarity and precision to spatial biology by extracting meaningful insights from PhenoCycler images.
-
Image Preprocessing
- Tissue Extraction: If a scan contains multiple tissues,
scripts/run_extract_tissue.shextracts tissues from the whole slide image (.qptiff) and saves them as separate images (.ome.tiff). - Antibody Extraction: If antibody names are not provided with the scan,
scripts/run_extract_antibody.shretrieves antibody names from the metadata in.qptifffile and saves them as a CSV file.
- Tissue Extraction: If a scan contains multiple tissues,
-
Cell Segmentation
- Run segmentation with Mesmer using
scripts/run_main.shand do cell profiling to generate a cell x antibody matrix.
- Run segmentation with Mesmer using
-
Downstream Analysis
- Perform pixel-level quality control (QC), cell-level QC, clustering, maker test using
scripts/run_analysis.sh.
- Perform pixel-level quality control (QC), cell-level QC, clustering, maker test using
-
Patch Visualization
- Aegle can segment images patch by patch. To visualize patch locations on the original image, use the standalong web app
aegle_patch_viewer.
- Aegle can segment images patch by patch. To visualize patch locations on the original image, use the standalong web app
-
Segmentation Evaluation
- Assess and visualize segmentation quality using the Napari Mask Density plugin, which provides intuitive overlays and density maps for inspecting segmentation performance.
Aegle is under active development. The pipeline is not yet ready for production use, and we are continuously improving its features and performance.
If you have any questions or feedback, please feel free to reach out to us.
Comprehensive documentation for Aegle is currently under development and will be hosted at: 👉 https://kimpenn.github.io/aegle/
The Aegle pipeline builds upon and is inspired by the following projects. We greatly appreciate their contributions to the field. More detailed disclaimers and attributions will be included in the source code as development progresses.