---
license: mit
datasets:
- chaitjo/gRNAde_datasets
tags:
- rna
- biology
- rna-design
- biomolecule-design
- 3d-design
- inverse-folding
- inverse-design
---

# gRNAde Model Checkpoints

[![BioRxiv](https://img.shields.io/badge/bioRxiv-2025.11.29-b31b1b.svg)](https://www.biorxiv.org/content/10.1101/2025.11.29.691298)
[![License](https://img.shields.io/badge/License-MIT-blue.svg)](https://github.com/chaitjo/geometric-rna-design/blob/main/LICENSE)
[![GitHub](https://img.shields.io/badge/GitHub-Repository-black)](https://github.com/chaitjo/geometric-rna-design)

This repository contains pre-trained model checkpoints for **gRNAde**, a generative AI framework for RNA inverse design.

![gRNAde pipeline](https://raw.githubusercontent.com/chaitjo/geometric-rna-design/refs/heads/main/gRNAde_pipeline.jpg)

## 📦 Model Checkpoints

```
.
├── gRNAde_drop3d@0.75_maxlen@500.h5    # Main gRNAde model checkpoint
├── rhofold/                            # RhoFold checkpoint (optional)
├── ribonanzanet/                       # RibonanzaNet checkpoint
└── ribonanzanet_sec_struct/            # RibonanzaNet secondary structure checkpoint
```

### gRNAde Model
- **File**: `gRNAde_drop3d@0.75_maxlen@500.h5`
- **Description**: Core gRNAde model trained on RNA structures from PDB (≤4Å resolution, RNASolo database, Oct 2023 cutoff)
- **Architecture**: Geometric Graph Neural Network conditioned on 3D backbone coordinates and secondary structures
- **Training**: 75% 3D coordinate dropout, maximum sequence length 500 nucleotides
- **Use case**: Generating RNA sequences for target 3D structures and secondary structures

### RibonanzaNet
- **Directory**: `ribonanzanet/`
- **Description**: RNA structure foundation model for predicting per-nucleotide SHAPE reactivity profiles
- **Use case**: High-throughput computational screening of designed sequences
- **Reference**: Trained on Ribonanza dataset with diverse natural and synthetic RNAs

### RibonanzaNet Secondary Structure
- **Directory**: `ribonanzanet_sec_struct/`
- **Description**: RibonanzaNet variant for predicting pseudoknotted secondary structures
- **Use case**: Alternative structural screening metric and OpenKnot score calculation

### RhoFold (Optional)
- **Directory**: `rhofold/`
- **Description**: RNA sequence to 3D structure prediction tool
- **Use case**: Predicting 3D structures of designed sequences (not used by default in the pipeline)

## 🚀 Quick Start

After setting up the [gRNAde codebase](https://github.com/chaitjo/geometric-rna-design), checkpoints can be downloaded manually and placed in the appropriate directory, or using HuggingFace CLI (recommended):

```bash
# Ensure you are in the base directory
cd ~/geometric-rna-design

# Install HuggingFace CLI (https://huggingface.co/docs/huggingface_hub/main/en/guides/cli)
curl -LsSf https://hf.co/cli/install.sh | bash
# alternate: pip install -U "huggingface_hub", or brew install huggingface-cli
hf auth login

# Download all checkpoints to checkpoints/ directory
huggingface-cli download chaitjo/gRNAde --local-dir checkpoints/
```

Alternatively, download specific files:

```bash
# Download only the main gRNAde checkpoint
huggingface-cli download chaitjo/gRNAde gRNAde_drop3d@0.75_maxlen@500.h5 --local-dir checkpoints/

# Download only RibonanzaNet checkpoints (required for design pipeline)
huggingface-cli download chaitjo/gRNAde ribonanzanet/ --local-dir checkpoints/
huggingface-cli download chaitjo/gRNAde ribonanzanet_sec_struct/ --local-dir checkpoints/
```

## Citations

```
@article{joshi2025generative,
  title={Generative inverse design of RNA structure and function with g{RNA}de},
  author={Joshi, Chaitanya K and Gianni, Edoardo and Kwok, Samantha LY and Mathis, Simon V and Lio, Pietro and Holliger, Philipp},
  journal={bioRxiv},
  year={2025},
  publisher={Cold Spring Harbor Laboratory}
}

@inproceedings{joshi2025grnade,
  title={g{RNA}de: Geometric Deep Learning for 3D RNA inverse design},
  author={Joshi, Chaitanya K and Jamasb, Arian R and Vi{\~n}as, Ramon and Harris, Charles and Mathis, Simon V and Morehead, Alex and Anand, Rishabh and Li{\`o}, Pietro},
  booktitle={International Conference on Learning Representations (ICLR)},
  year={2025},
}

@incollection{joshi2024grnade,
  title={g{RNA}de: A Geometric Deep Learning pipeline for 3D RNA inverse design},
  author={Joshi, Chaitanya K and Li{\`o}, Pietro},
  booktitle={RNA Design: Methods and Protocols},
  pages={121--135},
  year={2024},
  publisher={Springer}
}
```