Pseudo-single-cell spatial proteomics of human pancreatic islet tissue. This repository contains the complete analysis pipeline and 2,215 protein abundance maps at cellular resolution.
Browse the interactive protein maps
This project maps protein expression in human pancreatic islets by combining:
- CellPose 2.0 deep learning cell segmentation
- FIJI/ImageJ cell boundary refinement and pixel-to-cell mapping
- Immunofluorescence (INS, GCG, DAPI) for cell type classification (alpha, beta, acinar)
- Nanopots mass spectrometry for protein quantification
- Azimuth single-cell RNA-seq reference for cell-type deconvolution
| Stage | Script | Description |
|---|---|---|
| 0 | 0a_1-3D_spatial_proteomics.R |
3D spatial analysis with Giotto |
| 1 | 1b_1-ROI_mapping_and_cell_type_assignment.R |
Cell segmentation and type assignment |
| 2 | 2b_1-ROI_to_pixel.R |
ROI-to-pixel coordinate mapping |
| 3 | 3a_1-ROI_and_pixel_to_MS.R |
Mass spectrometry data integration |
| 4 | 4a_1-RNA-Seq.R |
RNA-seq deconvolution |
| 5 | 5a_1-cell_type_adjusted_protein_maps.R |
Final protein map generation |
2,215 protein abundance maps in output/RD5-final_protein_maps/final_protein_maps/, each showing cell-type-adjusted relative intensity (navy blue = low, yellow = high) across segmented cell polygons.
The companion website at skelly001.github.io/3D_Islet_Mapping features:
- Interactive slider comparisons between cell type maps and protein expression
- Searchable gallery of all 2,215 protein maps
- Pipeline visualization and methods summary