DRfold2: Ab initio RNA structure prediction with composite language model and denoised end-to-end learning
DRfold2 is a deep learning method for RNA structure prediction. At its core, DRfold2 utilizes the RNA Composite Language Model (RCLM), which provides enhanced full likelihood approximation capabilities to effectively capture co-evolutionary signals from unsupervised sequence data.
- Advanced RNA Composite Language Model (RCLM)
- End-to-end structure and geometry prediction
- Optimization protocol
- Python (tested on 3.10.4, 3.11.4, and 3.11.7)
- PyTorch (tested on 1.11, 2.01, and 2.21)
- NumPy
- SciPy
- BioPython
- OpenMM (required for structure refinement) - Installation Guide
-
Clone and navigate to the DRfold2 directory:
git clone https://github.com/leeyang/DRfold2 cd DRfold2 -
Run the installation script:
bash install.sh
This will download model weights ~1.3GB and install Arena.
For single model prediction:
python DRfold_infer.py [input fasta file] [output_dir]For multiple model prediction (up to 5 models):
python DRfold_infer.py [input fasta file] [output_dir] 1[input fasta file]: Target sequence in FASTA format[output_dir]: Directory for saving intermediate and final results- Final predictions will be saved as
[output dir]/relax/model_*.pdb
To further refine a predicted structure:
python script/refine.py [input pdb] [output pdb]python DRfold_infer.py test/seq.fasta test/8fza_A/ 1The final results can be found in test/8fza_A/relax/.
Note: For long RNA sequences, you may want to clear intermediate results from test/8fza_A/ to save space.
DRfold2 has been extensively tested on non-redundant test sets with various redundancy cut-offs, consistently demonstrating superior performance in:
- 3D structure prediction
- 2D base pair modeling
- Co-evolutionary feature learning from unsupervised data
Please report any issues or bugs on our GitHub Issues page.
If you use DRfold2 in your research, please cite:
@article{li2025drfold2,
title={Ab initio RNA structure prediction with composite language model and denoised end-to-end learning},
author={Yang Li, Chenjie Feng, Xi Zhang, Yang Zhang.},
journal={},
year={2025}
}Copyright (c) 2025 Yang Li
Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.