diff --git a/README.md b/README.md index 26dd4ea..9c7f50d 100644 --- a/README.md +++ b/README.md @@ -1,2 +1,87 @@ # exp-overview an overview and reference for the experimental projects + +## CellMap Experimental Projects + +This repository provides an overview of the current experimental setups and projects within the CellMap initiative at Janelia Research Campus. + +## Current Experimental Projects + +| Project | Repository | Status | Dataset(s) | Organelles/Targets | ML Framework | Description | Key Tools | +|---------|------------|--------|------------|-------------------|--------------|-------------|-----------| +| **C. elegans Expression** | [expr_c-elegans](https://github.com/janelia-cellmap/expr_c-elegans) | 🔶 Archived | jrc_c-elegans-op50-1
jrc_c-elegans-bw-1
jrc_c-elegans-comma-1 | Mitochondria [mito]
Lipid droplets [ld] | DaCapo + UNet | C. elegans organelle segmentation experiments | DaCapo, process-blockwise | + +## Core CellMap Infrastructure + +| Component | Repository | Purpose | Key Features | +|-----------|------------|---------|--------------| +| **Data Loading** | [cellmap-data](https://github.com/janelia-cellmap/cellmap-data) | PyTorch data loading for ML training | Multi-class segmentation, spatial transforms, efficient streaming | +| **Model Repository** | [cellmap-models](https://github.com/janelia-cellmap/cellmap-models) | Pre-trained segmentation models | COSEM models, whole-cell organelle segmentation | +| **Segmentation Challenge** | [cellmap-segmentation-challenge](https://github.com/janelia-cellmap/cellmap-segmentation-challenge) | Competition framework and toolbox | Data access, training, prediction, evaluation | +| **Data Analysis** | [cellmap-analyze](https://github.com/janelia-cellmap/cellmap-analyze) | Analysis utilities | Post-processing and metrics | +| **Nucleus Segmentation** | [cellmap-nuclei](https://github.com/janelia-cellmap/cellmap-nuclei) | Nucleus segmentation for FIB-SEM | Specialized nucleus detection | +| **Visualization** | [cellmap-flow](https://github.com/janelia-cellmap/cellmap-flow) | Results visualization | Neuroglancer integration | + +## Experimental Setup Details + +### C. elegans Expression Project +- **Datasets**: + - `jrc_c-elegans-op50-1` and `jrc_c-elegans-bw-1` (similar characteristics, combined for training) + - `jrc_c-elegans-comma-1` (different characteristics, limited crops available) +- **Target Organelles**: + - Mitochondria using full pipeline + - Lipid droplets using lightweight U-Net (large, round, easy to identify) +- **ML Framework**: [DaCapo](https://github.com/janelia-cellmap/dacapo) for training +- **Inference**: [process-blockwise](https://github.com/janelia-cellmap/process-blockwise) +- **Status**: Archived (completed experimental phase) + +## Data Format and Storage +- **Primary Format**: OME-NGFF/Zarr with multiscale support +- **Storage**: Local filesystem and cloud (S3/GCS) via TensorStore +- **Visualization**: Neuroglancer for 3D volume inspection + +## Common Tools and Workflows + +### Training Pipeline +1. **Data Preparation**: cellmap-data for efficient loading and augmentation +2. **Model Training**: DaCapo framework with configurable architectures +3. **Inference**: process-blockwise for large-volume prediction +4. **Visualization**: cellmap-flow with Neuroglancer integration + +### Model Architectures +- **U-Net variants**: 2D and 3D configurations +- **Multi-class segmentation**: Support for mutually exclusive classes +- **Pre-trained models**: Available via cellmap-models + +## Getting Started + +To work with CellMap experimental data: + +1. **Install core packages**: + ```bash + pip install cellmap-data cellmap-models + ``` + +2. **Download challenge data**: + ```bash + pip install cellmap-segmentation-challenge + csc fetch-data + ``` + +3. **Explore existing experiments**: + - Review the [expr_c-elegans](https://github.com/janelia-cellmap/expr_c-elegans) repository for C. elegans workflows + - Check [cellmap-segmentation-challenge](https://github.com/janelia-cellmap/cellmap-segmentation-challenge) for training examples + +## Contributing + +To add a new experimental project: +1. Create a repository following the `expr_*` naming convention +2. Include a comprehensive README with dataset and organelle information +3. Document the ML frameworks and tools used +4. Update this overview table with your project details + +## Resources + +- [CellMap Project Page](https://www.janelia.org/project-team/cellmap) +- [CellMap Segmentation Challenge](https://cellmapchallenge.janelia.org/) +- [DaCapo Documentation](https://github.com/janelia-cellmap/dacapo)