Interactive application for multimodal spatial omics processing and metabolic barcode generation.
MetaBar integrates multiplexed immunofluorescence (IF) and mass spectrometry imaging (MSI) data into a unified point-and-click analysis pipeline. No programming experience required.
Developed by the Coskun Lab at Georgia Institute of Technology.
Windows only. Requires ~20 GB free disk space and an internet connection on first launch.
Download the latest installer from the Releases page:
| File | Description |
|---|---|
MetaBar_Setup.exe |
Installer (run this) |
MetaBar_Setup-1.bin |
Installer data part 1 |
MetaBar_Setup.exe.
- Download both files from the Releases page into the same folder.
- Double-click
MetaBar_Setup.exeand follow the wizard. - On first launch, the app will automatically download and install its Python dependencies (10–20 minutes, internet required). This only happens once.
- The app opens in your browser at
http://localhost:8501.
For detailed instructions, see the User Manual.
Two datasets are available on Zenodo to explore MetaBar with real data.
| File | Dataset | Size | Contents |
|---|---|---|---|
demo_clozapine_dose.zip |
Clozapine dose–response (mouse brain) | 7.2 GB | IF TIFF, MALDI TIFF, channel names |
demo_triomic.zip |
Trimodal AD vs WT (mouse brain) | — | IF TIFF, MALDI TIFF, channel names |
demo_human_colorectal_cancer.zip |
Human colorectal cancer | 65.6 MB | IF TIFF, MALDI TIFF, cell mask, phenotype annotations, channel names |
How to use: Load the IF and MSI TIFFs from the sidebar, then follow the Analysis Pipeline (Cell Segmentation → Clustering → GNN). The colorectal cancer sample also includes a pre-made cell_mask.tif and phenotypes.csv you can load directly in the Custom Data tab to skip segmentation.
| File | Dataset | Size | Contents |
|---|---|---|---|
results_clozapine_dose.zip |
Clozapine dose–response | 3.1 GB | Segmentation, projection, clustering, positivity, GNN, cross-sample comparison |
results_triomic.zip |
Trimodal AD vs WT | — | Segmentation, projection, clustering, positivity, GNN, cross-sample comparison |
results_human_colorectal_cancer.zip |
Human colorectal cancer | 464.2 MB | Segmentation, projection, annotations, GNN, cross-sample comparison |
How to use:
- Napari viewer — extract the zip, open the app, go to Interactive Viewer (napari), set the results folder path, and click Launch napari.
- Cross-sample comparison — go to Cross-Sample Comparative Analysis, add each sample subfolder as a separate entry, and run comparison.
Each zip contains one representative sample plus the cross-sample comparison outputs for that dataset.
MetaBar provides a complete workflow:
| Step | Description |
|---|---|
| Data Loading | IF (TIFF) and MSI (TIFF or imzML/IBD) |
| Preprocessing | Rotate, flip, interactive crop |
| Registration | MSI → IF alignment via Fiji SIFT |
| Cell Segmentation | Nuclear segmentation using Mesmer (DeepCell) |
| Nuclei Expansion | Cytoplasmic region approximation |
| MBP Mask | Myelinated tissue region detection |
| MSI Projection | LR → HR Gaussian-weighted interpolation |
| Superpixel Segmentation | SLIC-based tissue parcellation |
| Clustering | PCA + UMAP + Leiden + k-means |
| Positivity Thresholding | Per-channel GMM-based binary labelling |
| GNN Explainability | GraphSAGE/GCN/GATv2 feature importance |
| Comparative Analysis | Cross-sample GNN importance comparison |
| Custom Data | Upload your own masks and cell annotations |
| Interactive Viewer | napari-based spatial exploration |
| Requirement | Minimum |
|---|---|
| OS | Windows 10 or 11 (64-bit) |
| RAM | 16 GB (32 GB recommended) |
| Disk | 20 GB free |
| GPU | NVIDIA GPU recommended (CPU fallback automatic) |
| Internet | Required on first launch only |
If you use MetaBar in your research, please cite:
Ozturk et al. (2026). MetaBar: Interactive application for multimodal spatial omics processing and metabolic barcode generation. Nature Communications.
For questions or bug reports, please open an issue.
Coskun Lab · Department of Biomedical Engineering · Georgia Institute of Technology