An Enhanced Dual & Dynamic U-Mamba Model for Long-Range RNA Secondary Structure Prediction
UFold-X is a GPU-accelerated RNA secondary structure prediction server. For sequences longer than 5000 nt, please use the command-line version available on GitHub.
Usage:
1. Please upload a FASTA sequence file or paste sequences in the input box. You can click the "Show example data" button below to see an example.
2. Enter your output file name (no suffix needed) and click the "Submit" button.
3. Once the job is finished, you can visualize and download your results from the Download panel.
Note: The computation may take some time based on the size of your data. If you have any questions, please feel free to contact us.
UFold-X: An Enhanced Dual & Dynamic U-Mamba Model for Long-Range RNA Secondary Structure Prediction
My Prediction Tasks
View your prediction history by entering your email address below.
Your Tasks
Tip: Click on a completed task to view or download the results.
Note: Tasks are automatically deleted after 3 days.
Task Details
Sequence Count:
Model Type:
UFold-X User Guide
Input Format
UFold-X accepts RNA sequences in standard FASTA format. Each sequence should have a header line starting with ">" followed by the sequence:
>sequence_name_1 |AUGCGAUCGAUAGCUAGCUAGUCGAUCGUAGCUAGCUAG |>sequence_name_2 |GCUAGCUAGCUAGCUAGCUAGCUAGCUAGCUAGCUAG
Multiple sequences can be submitted in a single job (maximum 100 sequences per job).
Supported Nucleotides
Allowed characters: A, U, C, G. Character T will be automatically converted to U.
Important: Ambiguous nucleotides such as N are NOT supported and will be automatically rejected with a clear error message.
Sequence Length
Maximum length per sequence: 5000 nt. Minimum length: 10 nt.
The appropriate model is automatically selected based on your input sequence length:
- ≤ 600 nt: UNet model
- 600–1800 nt: Model 1
- ≤ 1800 nt (mixed): Model 2
- 1800–5000 nt: Model 3
Output Formats
Two output formats are available for download:
- Dot-bracket (*.db) : Plain text with dot (.) for unpaired and parentheses () for paired nucleotides.
- CT format (*.ct) : Connectivity table format with position, nucleotide, and pairing information.
Task Tracking with Email (Optional)
You can optionally enter your email address when submitting a prediction job to track your tasks:
- Enter your email : Fill in the "Email (Optional)" field on the submission page.
- View tasks : Go to "My Tasks" page and enter your email to see all your prediction tasks.
- Load results : Click "Load Result" on any completed task to view or download results.
- Offline prediction : With email tracking, you can close the browser and return later to check results.
Note: All tasks (with or without email) are automatically deleted after 3 days.
Troubleshooting
Server busy: If you receive a GPU busy message, please wait a few minutes and try again.
Disconnection: Long sequences may take several minutes to process. Please be patient and do not reload the page.
Invalid input: Ensure your sequence contains only A, U, C, G and follows FASTA format. N is not supported.
Long sequences: For sequences longer than 5000 nt, please use the command-line version .
Quick Start Guide
New to UFold-X? Start our interactive guide to learn how to use the server.
Show Guide AgainContact Us
Contact us if you have any questions about the usage of this website.
Contact information
Laiyi Fu
Email: fulaiyi@gmail.com
Github Page: Github web
Cite:
Please consider to cite our paper if you use UFold-X in your research.
Paper title: UFold-X: An Enhanced Dual & Dynamic U-Mamba Model for Long-Range RNA Secondary Structure Prediction
Corresponding author e-mail: hequan.sun@xjtu.edu.cn & danyang.wu@xjtu.edu.cn
Acknowledgement
We would like to thank VARNA tools for visualizing the RNA secondary structure.