CONTENTS
Bioinfo Tools
Antibody Discovery
Learn
Contact
EN
中文
ASO User Guide
2026-07-30 | ASO development team
ASO
ASO Designer is a cutting-edge, AI-powered tool tailored for the precision design of Antisense Oligonucleotides (ASOs). The tool leverages state-of-the-art deep learning architectures to assist researchers in identifying optimal sequences for Exon Skipping, Exon Inclusion, and mRNA Degradation.
Tool Introduction and Background
ASO Designer is a cutting-edge, AI-powered tool tailored for the precision design of Antisense Oligonucleotides (ASOs). The tool leverages state-of-the-art deep learning architectures to assist researchers in identifying optimal sequences for Exon Skipping, Exon Inclusion, and mRNA Degradation.
ASO Designer integrates two specialized engine modules to ensure high-accuracy predictions:
- OligoAI [1] (for mRNA Degradation): Built upon RiNALMo [2] (Ribonucleic Acid Language Model), a 33-layer Transformer-based "giga-scale" model. It predicts inhibition efficiency (%) by encoding the ASO sequence alongside its pre-mRNA context, accounting for chemical modifications and experimental conditions.
- ASO-ExonPred (for Splicing Modulation): Specifically designed for splice-switching [3] ASOs. It utilizes a CNN-Transformer hybrid architecture (inspired by SpliceTransformer [4]) trained on genomic-scale data. It integrates multiple biophysical features, including RNA secondary structure via RNAstructure [5], hybridization thermodynamics via OligoWalk [6], and off-target risk assessment via BLAT [7].
Detailed Operating Steps
1. Choose Your Mechanism
At the top of the page, select the strategy based on your therapeutic goal:
- Exon Skipping: Facilitates the "skipping" of a specific target exon during splicing.
- Exon Inclusion: Promotes the retention of a target exon in the final mRNA transcript.
- mRNA Degradation: Targets mRNA for breakdown using mechanisms like RNase H-dependent gapmers.
2. Set Up Your Parameters
Enter your data in the "Parameters" panel on the left side of the screen.
A. For Exon Skipping and Exon Inclusion
- Gene Name: Enter the official human gene symbol. The system will provide suggestions as you type.
- Transcript: Select a specific transcript (RefSeq NM_ or Ensembl ENST_ [8]). Note: Different transcripts may have different exon numbering and coordinates.
- Target Exon:
- Exon Numbers: Choose the exon you wish to target from the dropdown.
- Exon Coordinate: Manually input the exact start/end coordinates if needed.
- ASO Length: Select the desired length (standard range is 12–30nt; default is 20nt).
- Mutation (Optional): Enter mutations using HGVS nomenclature (e.g., c.1234G>A). This allows the AI to predict efficiency within a specific variant background.
B. For mRNA Degradation
- Sugar Modifications: Enter the pattern for sugar-ring modifications (e.g., 5xMOE, 10xDNA, 5xMOE). Supported types: MOE, DNA, cEt.
- Backbone Modifications: Define the linkage chemistry (e.g., 19xPS). Supported types: PS (Phosphorothioate), PO (Phosphodiester).
- Transfection Method & Dosage: Select the delivery method (e.g., Electroporation, Gymnosis, Lipofection) and set the dosage (default is 4000nM).
3. Run the Design
- Click the【Design】button to start the AI calculation. It usually takes between 10 and 30 seconds to get your results.
- You can click【Reset】at any time to clear all fields.
4. Review Results
The review process depends on the mechanism you chose:
A. For Exon Skipping and Exon Inclusion
- Table: View the top 10 recommended sequences and their Splicing Scores.
- Scatter Plot: Check the visual map for performance. Aim for high-scoring dots and avoid red dots marked as "Exon Off-Target" risks.
B. For mRNA Degradation:
- Pick a Transcript: After the calculation finishes, check the box next to one or more transcripts in the structure map to load their specific data.
- Review Indicators: Once a transcript is selected, a detailed list will appear below showing the Genomic Coordinate, Region (Exonic/Intronic), and the AI Score.
- Filter: Higher scores indicate better predicted degradation efficiency.
5. Export data
Click the [Download] button to save the full candidate list as a CSV file, including detailed metrics like gc_content, tm_value, and prediction_score.
The Science Behind the Scoring
To help you select the best candidates, ASO Designer provides multi-dimensional scoring:
Splicing Modulation Scoring (ASO-ExonPred)
The primary metric is the Weighted Percentile Score (0–100). A higher score indicates a higher probability of successful splicing modulation. The score is a composite of:
- Splice Model Score: Predicted effect based on the CNN-Transformer model [4].
- Thermodynamics: Binding affinity calculated via OligoWalk [6].
- Off-target Penalty: Sequences with high off-target risks (based on BLAT [7] alignment) are penalized.
- Sequence Constraints: Penalties are applied for extreme GC content or unfavorable Tm values.
Degradation Efficiency (OligoAI [1])
The Prediction Score (0–100) represents the estimated Inhibition %. This model is trained on the ASO Atlas [9] dataset-comprising approximately 188,521 experimental data points-to ensure high correlation with real-world laboratory results.
Key Reminders & Compliance
- Human Species Only: This tool is currently optimized specifically for human genes.
- Open Source Disclosure: ASO Designer utilizes and acknowledges the following open-source models and tools: OligoAI [1], RiNALMo [2], RNAstructure [5], OligoWalk [6], and BLAT [7].
- Experimental Validation: While our AI provides high-confidence rankings, these scores are statistical estimates. We strongly recommend experimental validation before advancing to clinical stages.
References
- [1] B. Hill et al., *Accurately modelling RNase H-mediated antisense oligonucleotide efficacy*, bioRxiv (2025). https://doi.org/10.1101/2025.10.29.685292; https://www.biorxiv.org/content/10.1101/2025.10.29.685292v1
- [2] R. J. Penić, M. Šikić, M. Sušanj, *RiNALMo: General-purpose RNA language models can generalize well on structure prediction tasks*, arXiv:2403.00043 (2024). https://arxiv.org/abs/2403.00043
- [3] Aartsma-Rus, A. (2012). Overview on Aon Design. Methods in Molecular Biology.
- [4] You, N., et al. (2024). SpliceTransformer predicts tissue-specific splicing linked to human diseases. Nature Communications.
- [5] Reuter, J. S., & Mathews, D. H. (2010). RNAstructure: software for RNA secondary structure prediction and analysis. BMC Bioinformatics.
- [6] Lu, Z. J., & Mathews, D. H. (2008). OligoWalk: an online siRNA design tool utilizing hybridization thermodynamics. Nucleic Acids Research.
- [7] Kent, W. J. (2002). BLAT-the BLAST-like alignment tool. Genome Research.
- [8] Cunningham, F., et al. (2024). Ensembl 2024. Nucleic Acids Research. https://doi.org/10.1093/nar/gkad1049
- [9] Hill, B. et al. (2025). ASO Atlas: a comprehensive dataset of RNase H antisense oligonucleotides. bioRxiv.