---
license: cc-by-nc-sa-4.0
---

# Pathomics

An extension of the HEST dataset with cell-level segmentation, embeddings, and additional WSI samples.

## Relationship with HEST
This dataset is designed as a **companion layer** to HEST-1k. 
For samples originating from HEST, you need to download the base 
data from the original HEST repository. Pathomics only stores 
the additional analyses (cell segmentation, embeddings, etc.).

## Quick Start

To use this dataset, you need to gain access to both the **pathomics** and **HEST** repositories. Please follow these steps:

### 1. Request Access
* **pathomics Access**: Click the **"Request Access"** button at the top of this page.
* **HEST Data Access**: Navigate to the [HEST Repository](https://huggingface.co/datasets/MahmoodLab/hest) and click **"Request Access"** (Access is granted automatically). 
  > 💡 *Note: Once granted, you can manually inspect files via the "Files and versions" tab.*

### 2. Configure Hugging Face Token
To load the data programmatically, you need a **Write** access token:
1. Go to [Hugging Face Settings > Tokens](https://huggingface.co/settings/tokens).
2. Click **New token**, name it (e.g., `hest_access`), and set the role to **Write**.
3. **Copy** the token.

### 3. Environment Setup & Login
Install the necessary library and authenticate your session:

```bash
pip install huggingface-hub
```

```python
from huggingface_hub import login

# Replace 'YOUR_HF_TOKEN' with the token you just created
login(token="YOUR_HF_TOKEN")
```

## Download script:

```python
from huggingface_hub import snapshot_download
import pandas as pd

import glob
import os

def download_pathomics(ids=None, pathomics_dir='pathomics_data', hest_dir='pathomics_data', 
                       download_hest_base=True):
    """
    Download pathomics data, optionally downloading the corresponding HEST basic data at the same time.
    """
    # 1. Download pathomics
    if ids is None:
        patterns = '*'
    else:
        # patterns = [f"*{sid}[_./]**" for sid in ids] + ['*.csv', '*.md', 'scripts/*']
        patterns = [f"*{sid}[_./]**" for sid in ids] + ['*.csv']
    
    snapshot_download(
        repo_id='Boyoungc/pathomics',
        allow_patterns=patterns,
        repo_type='dataset',
        local_dir=pathomics_dir,
    )
    
    # 2. Optional: Automatically download the corresponding HEST basic data.
    if download_hest_base:

        csv_path = sorted([f for f in os.listdir(pathomics_dir) 
                          if f.startswith('PATHOMICS_v')])[-1]
        meta = pd.read_csv(os.path.join(pathomics_dir, csv_path))
        
        if ids:
            meta = meta[meta['id'].isin(ids)]
        hest_ids = meta[meta['source'] == 'hest']['hest_id'].dropna().tolist()
        
        if hest_ids:
            hest_patterns = [f"*{hid}[_.]**" for hid in hest_ids]
            snapshot_download(
                repo_id='MahmoodLab/hest',
                allow_patterns=hest_patterns,
                repo_type='dataset',
                local_dir=hest_dir,
            )
            print(f"Downloaded {len(hest_ids)} HEST base samples")

download_pathomics(ids=['NCBI689', 'MEND62'])
```


# usage
download the data by specific IDs, please refer to the PATHOMICS_v.csv file
```python
download_pathomics(ids=['NCBI689', 'MEND62'])
```

or download all the data:
```python
download_pathomics()
```