# 🧬 AbLang2 Sequence Restorer - Hugging Face Spaces

This is a Gradio web application that provides the AbLang2 sequence restoration utility through Hugging Face Spaces.

## 🎯 What it does

The AbLang2 Sequence Restorer allows you to:
- **Restore masked residues** (*) in antibody sequences
- **Work with paired sequences** (heavy and light chains)
- **Handle single chains** (heavy or light chain only)
- **Use alignment** for variable missing lengths

## 🚀 How to use

1. **Enter sequences**: Provide heavy chain, light chain, or both sequences
2. **Mask residues**: Use `*` to indicate residues you want to restore
3. **Choose alignment**: Enable "Use Alignment" for variable missing lengths
4. **Get results**: Click "Restore Sequences" to get the restored antibody sequences

## 📝 Example Usage

### Example 1: Both chains with masked residues
- **Heavy Chain**: `EVQ***SGGEVKKPGASVKVSCRASGYTFRNYGLTWVRQAPGQGLEWMGWISAYNGNTNYAQKFQGRVTLTTDTSTSTAYMELRSLRSDDTAVYFCAR**PGHGAAFMDVWGTGTTVTVSS`
- **Light Chain**: `DIQLTQSPLSLPVTLGQPASISCRSS*SLEASDTNIYLSWFQQRPGQSPRRLIYKI*NRDSGVPDRFSGSGSGTHFTLRISRVEADDVAVYYCMQGTHWPPAFGQGTKVDIK`

### Example 2: Heavy chain only
- **Heavy Chain**: `EVQLVESGGGLVQPGGSLRLSCAASGFTFSSYAMGWVRQAPGKGLEWVSAISGSGGSTYYADSVKGRFTISRDNSKNTLYLQMNSLRAEDTAVYYCARDY**GMDVWGQGTTVTVSS`
- **Light Chain**: (leave empty)

## 🔧 Technical Details

- **Model**: AbLang2 from Hugging Face Hub (`hemantn/ablang2`)
- **Framework**: Gradio for the web interface
- **Backend**: PyTorch with Transformers library
- **Processing**: Automatic GPU acceleration when available

## 📚 Related Resources

- **Original AbLang2**: [https://github.com/TobiasHeOl/AbLang2](https://github.com/TobiasHeOl/AbLang2)
- **Model Repository**: [https://huggingface.co/hemantn/ablang2](https://huggingface.co/hemantn/ablang2)
- **Full Documentation**: See the main README.md for comprehensive usage examples

## 🤝 Citation

If you use this tool in your research, please cite the original AbLang2 paper:

```
@article{Olsen2024,
  title={Addressing the antibody germline bias and its effect on language models for improved antibody design},
  author={Tobias H. Olsen, Iain H. Moal and Charlotte M. Deane},
  journal={bioRxiv},
  doi={https://doi.org/10.1101/2024.02.02.578678},
  year={2024}
}
```