---
license: mit
tags:
  - cryo-et
  - tomography
  - segmentation
  - biology
  - microscopy
  - sam
  - medsam
pretty_name: AgentProject XuLab — CryoET Large Files
size_categories:
  - 1B<n<10B
---

# AgentProject XuLab — Large Files

This dataset hosts large binary files for the [AgentProject_XuLab](https://github.com/vatsal-ts/AgentProject_XuLab) repository, which is part of the XuLab research project on AI-driven Cryo-Electron Tomography (CryoET) analysis.

These files are stored here due to GitHub's 100MB file size limit.

## Contents

| File | Size | Description |
|---|---|---|
| `sam_vit_b_01ec64.pth` | 358 MB | SAM (Segment Anything Model) ViT-B checkpoint |
| `medsam_vit_b.pth` | 358 MB | MedSAM ViT-B checkpoint (medical image segmentation) |
| `tomograms/local.mrc` | 1.6 GB | CryoET tomogram — full volume (NPC1-deficient HEK293T cells) |
| `tomograms/mitochondrion_20036.mrc` | 385 MB | CryoET tomogram — mitochondrion annotation volume |

## About the Data

The tomogram files are from a CryoET dataset of **NPC1-deficient human HEK293T cells** with annotations for multiple cellular features including lysosomes, mitochondria, and membranes.

The `.pth` files are pretrained model weights used by the CryoETAgent pipeline for organelle segmentation:
- **SAM** ([Kirillov et al., 2023](https://arxiv.org/abs/2304.02643)) — general-purpose segmentation
- **MedSAM** ([Ma et al., 2024](https://www.nature.com/articles/s41467-024-44824-z)) — SAM fine-tuned for medical/biological imaging

## Usage

Clone the GitHub repo and run the download script:

```bash
git clone https://github.com/vatsal-ts/AgentProject_XuLab
cd AgentProject_XuLab
pip install huggingface_hub
python download_large_files.py
```

This will place all files in their correct locations automatically.

## Related

- GitHub repo: [vatsal-ts/AgentProject_XuLab](https://github.com/vatsal-ts/AgentProject_XuLab)
- CryoETAgent: AI agent for automated CryoET organelle analysis