Design effective and precise ASO sequences with our advanced computational platform
The software currently relies on standard bioinformatics and synthetic biology data formats, FASTA, GFF, and GTF, for genome and transcript data. It is compatible with existing RNA design and modeling tools such as ViennaRNA and RIsearch. One of our tool's output options is the SBOL - Synthetic Biology Open Language format.
Our SBOL implementation structures each designed ASO as a ComponentDefinition with associated sequence information, chemical modifications, and metadata. Each ASO is assigned a unique identifier and includes the target gene name, sequence elements, and modification patterns. This standardized format enables seamless integration with SBOL-compatible design tools and facilitates the sharing of ASO designs across different research platforms and laboratories.
The ASO Designer pipeline was validated through both wet-lab experiments and computational benchmarking.
In the wet lab, the software’s predictions were tested in HEK293 and A549 human cell lines, targeting GFP and MALAT1 transcripts. ASOs designed by the model were chemically synthesized and demonstrated superior knockdown efficiency compared to industry reference sequences, as confirmed by RT-qPCR and FACS assays. Additional biological validation included synthetic-lethality ASO tests against PRMT5, RIOK1, and MAT2A, as well as yeast-based ASO uptake and activity assays verified via flow cytometry. These results confirm that the computational predictions translate into biologically active and effective ASOs.
In addition, the model was trained and validated using large-scale inhibition datasets (~40,000 ASO-target measurements). Performance was assessed through train/test splits and correlation analyses, showing high agreement between predicted and observed inhibition levels. This ensures that the software is both experimentally validated and statistically robust.
ASO Designer can be applied to any gene in humans, mice and yeast and in later versions to and E. coli, making it broadly useful for gene knockdown research, RNA therapeutics, and synthetic lethality studies. The open-source design allows other teams to reuse it for custom targets, benchmark ASO efficacy, or integrate it into larger therapeutic pipelines. We estimate that there are more than 25,000 potential users of our software in the world, and this audience is expected to grow significantly in the next 5 years. Further details on target audience size are provided in the Audience Size Analysis section below.
Moreover, we are currently preparing an extensive research paper documenting our methodology, validation studies, and comparative analyses, which will provide the scientific community with detailed insights into our algorithmic approach and experimental validations. This publication will not only establish the theoretical foundation for our tool but also offer standardized protocols and benchmarks that other researchers can adopt and build upon. By sharing our findings through peer review, we aim to accelerate broader adoption of computational ASO design and contribute to the growing body of knowledge in antisense therapeutics.
The software is designed for straightforward integration within existing bioinformatics workflows. It follows common Python packaging conventions, making it easy to install and import as a module or command-line tool. Users can install it directly from PyPI using 'pip install asodesigner'.
The package relies exclusively on standard data formats (FASTA, GFF, GTF) and well-established Python libraries such as NumPy, Pandas, BioPython, and gffutils, ensuring compatibility with most RNA analysis pipelines. External tools, including ViennaRNA (for secondary-structure prediction) and RIsearch (for hybridization modeling), are seamlessly integrated, and all dependencies are pinned for reproducible builds.
ASO Designer was developed with accessibility and clarity in mind. The web interface is intentionally minimal, requiring only the selection of an organism and target gene as input. Optional parameters such as Top-k results and a Detailed Analysis checkbox make it suitable for both entry-level users seeking quick predictions and advanced users requiring deeper interpretability.
A comprehensive User Guide walks users through every stage of the workflow, from data input and parameter selection to interpreting output plots and downloading designed ASOs. The FAQ section addresses common questions about supported organisms, runtime, and result interpretation. A short set of images showcases the complete workflow from gene selection to ASO result. Together, these resources make the platform self-explanatory, easy to navigate, and practical for both newcomers to ASO design and experienced computational biologists integrating it into larger pipelines.
The codebase is well-structured and modular, featuring a setup script, Conda environment file, and pytest validation suite. Documentation includes setup instructions and experiment links. The repository is open-source and designed for community contributions, making it easy for future teams to extend.
ASO Designer is an open-source platform that helps researchers design and rank antisense oligonucleotides (ASOs) for targeted RNA knockdown.
It combines sequence analysis, structural modeling, cellular context, and machine learning to predict which ASOs are most likely to succeed in living cells.
Unlike simple GC% or melting temperature calculators, ASO Designer integrates:
The system uses a learning-to-rank model trained on experimental inhibition data to prioritize the most promising sequences.
Want to design your own ASO sequences? Try our powerful computational tool designed specifically for precision targeting.
Design effective and precise ASO sequences with our advanced computational platform
Gene Silencing and downregulation is a fundamental approach in research, medicine, and specifically synthetic biology. By selectively reducing the expression of specific genes, scientists can uncover their biological roles, validate therapeutic targets, and engineer new cellular behaviors. Unlike genome-editing tools such as CRISPR, which introduce permanent changes to the DNA sequence, silencing and downregulation provide a transient and reversible modulation of gene activity. This temporal control allows researchers to investigate gene function without altering the genome itself, offering a safer and more flexible strategy for both experimental and therapeutic applications. In research, silencing enables functional genomics studies, exploration of antibiotic resistance mechanisms, metabolic engineering of synthetic pathways, and other synthetic biology applications where fine-tuning gene expression is essential. In medicine, targeted silencing of disease-associated genes has led to major breakthroughs in treating genetic disorders, cancers, and viral infections. Antisense oligonucleotides (ASOs) are among the most versatile and clinically advanced tools for gene silencing, with several FDA-approved therapies already demonstrating their precision and efficacy [1].
Designing effective antisense oligonucleotides (ASOs) is still largely a trial-and-error process. Many sequences that look promising in silico fail in cells because they ignore cell-specific context, RNA structure and accessibility, RNase H1 cleavage windows, transcript abundance and half-life, and expression-weighted off-targets. Existing tools often optimize one or two factors (e.g., GC% or ΔG) in isolation, making it hard for wet-lab teams to prioritize candidates with a high probability of in-cell success and to reproduce results across labs and datasets.
ASO Designer was built to close this gap. It unifies thermodynamics and hybridization modeling with cell-line aware features and a rank model trained on large experimental inhibition datasets, producing ranked, interpretable ASOs rather than single opaque scores. By using standard formats (FASTA/GFF/GTF), shipping an open-source package and a simple web UI, and logging run configurations for reproducibility, ASO Designer reduces experimental cycles and cost, enables transparent decision-making, and makes advanced ASO design accessible to both iGEM teams and research groups integrating RNA therapeutics or synthetic-lethality screens into their pipelines.
Our estimates for the addressable research community for the ASO Designer software tool comprise roughly 12,930–26,100 potential users worldwide. This estimate encompasses ~630–945 academic institutions. According to the World Higher Education Database (WHED) [2], there are around 21,000 universities globally, of which approximately 30% host life-science programs and 10–15% conduct ASO-related work. According to GlobalData [3], around 285 industry organizations develop or test antisense therapeutics. With an average of 4–6 active users per academic lab, and ~4 labs per institute, and 10–12 per company, the expected user base spans around 13,000 researchers on the conservative end to ≈26,000 on the generous end.
Importantly, the tool could also benefit a significant number of iGEM teams working in synthetic biology, genetic regulation, and RNA-based design. Each year, over 400 iGEM teams participate worldwide, many exploring gene control, RNA therapeutics, or modular chassis engineering, areas where ASO Designer provides a ready-to-use, open-source solution for rational oligonucleotide design and target validation. Similar to the success of CRISPR-Cas systems in revolutionizing genetic research, our tool harnesses ASO technology to provide an equally powerful yet complementary approach for investigating gene knockdown effects. By streamlining the process of targeted gene silencing, this platform becomes an essential resource that expands the synthetic biology toolkit, offering researchers a versatile and accessible method for exploring gene function and regulatory networks.
Because ASO Designer is open-source, English-based, and browser-accessible, it can serve nearly all of this niche from launch without localization barriers. At the beta stage, realistic engagement is expected from 1–3% of the reachable pool, between 100-1,000 early adopters, sufficient to collect statistically meaningful feedback and benchmark usability.
The TAUSO pipeline integrates biological knowledge, structural modeling, and machine learning to transform a target gene sequence into a set of experimentally viable antisense oligonucleotide (ASO) candidates. Starting from user inputs such as organism and target gene, the system automatically retrieves genomic references, normalizes exon–intron annotations, and generates multiple ASO candidates in parallel chemistry modes. Each candidate is analyzed through a comprehensive feature computation pipeline encompassing thermodynamic stability, RNA accessibility, codon usage bias, and contextual environment. These features feed into a learning-to-rank model, trained on real inhibition data, which scores each sequence by predicted efficacy. Subsequent off-target assessment, intelligent shortlisting, and exportable reports ensure that every user receives high-confidence ASOs ready for experimental testing or further computational refinement.
Although several computational tools for antisense oligonucleotide (ASO) design have been developed in recent years, none of them fully capture the biological reality of ASO activity. Most existing platforms focus on isolated aspects of the design process, such as only optimizing chemical modifications, relying on minimal features like accessibility or hybridization, or providing annotations without empirical scoring.
Our platform was created to address this gap.
We addressed these limitations by improving the accuracy of accessibility and hybridization features, incorporating novel descriptors from literature, and basing our model on empirical data to ensure realistic biological relevance. We developed a comprehensive, machine learning-driven, and biologically aware framework that expands beyond conventional design logic by incorporating new experimental and structural datasets unavailable in existing tools, including:
By combining these complementary biological layers within a machine learning model trained on empirical data, Oncoligo transforms ASO design from a rule-based task into a predictive, biologically validated discovery process.
The table below illustrates how Oncoligo improves upon and outperforms each existing ASO design tool. While all current platforms contribute valuable functionalities, they remain limited in scope, empirical depth, or generalizability. Oncoligo overcomes these limitations by integrating experimental data, enhanced folding and hybridization modeling, and ML-based prioritization into one unified and reproducible design pipeline-delivering more accurate, biologically meaningful, and scalable results.
| Model | Strength | Limitation | Our Added Value |
|---|---|---|---|
| ASOptimizer | Deep learning, optimizes modifications [4] | Optimizes given sequence only | Integrated sequence finding with multiple features |
| PFRED | Rich annotations (exon, SNPs, conservation) [5] | No prioritization, manual analysis | Empirical automated ranking pipeline |
| QIAGEN LNA GapmeR Designer | LNA optimization [6] | Narrow scope, rule-based | Generalizable & empirical multi-step framework |
| LNCASO | long non-coding RNAs-specific design [7] | Not generalizable, data undisclosed | Broad applicability & reproducible |
Table 1: This table summarizes the main strengths and limitations of current computational platforms for antisense oligonucleotide design and highlights how Oncoligo surpasses each of them. While existing tools focus on specific aspects - such as modification optimization or annotation visualization - Oncoligo integrates machine learning, empirical data, and structural modeling into a unified, biologically informed framework that delivers superior generalizability and predictive accuracy.
Altogether, Oncoligo provides a superior and more reliable approach to ASO design - bridging the gap left by current tools and setting a new standard for predictive accuracy.
You can learn more about the computational foundations of our framework in the dedicated Model Page, where we describe the model’s architecture, feature integration, and predictive principles in detail.
Select the organism and reference build you’re working with. This choice determines the transcriptome, isoforms, and expression tables used later for off-target analysis and coverage calculations, so it must match your experiment (for example, Human - GRCh38; Yeast - S288C). After picking exactly one organism, continue to the next step.
Provide the gene you want us to analyze. You can either choose the Gene by its Canonical Name, upload a FASTA/GenBank file, or paste the sequence directly into the text box. What you enter here defines the search space for candidate ASOs; longer inputs explore more sites and may take a little longer to process.
To get a more customized result sheet, the Top-k results is a numeric input field (default = 10) specifying how many optimized ASOs will be returned. Get detailed analysis: checkbox enabling an extended output with off-target detail analysis.
Confirm where we should send your results. Enter your name and a valid email address, then click Start Processing. You will receive a confirmation email that the processing has begun. Your job is queued on the server, and when it finishes, you’ll receive the result page by mail.
ASO Designer is a user-friendly tool for designing antisense oligonucleotides (ASOs) to silence specific genes. It uses advanced algorithms to suggest sequences that are efficient, specific, and practical for research use.
The ASO Designer currently supports human, mouse and yeast gene knockdown. E. coli is going to be available in the next software update. If you’re working with another species, feel free to email us at igem@tauex.tau.ac.il to request support in future updates.
Most jobs are completed within a few minutes to about 30 minutes. Runtime depends on gene length, organism, structural complexity, and server availability.
We rank candidates using multiple criteria, including binding energy (target pairing strength),
RNA folding/accessibility, GC content (stability balance), off-target predictions,
and additional model features. This combination helps surface ASOs that are both effective and practical.
For further reading visit our Model page - Methods section.
No computational tool can guarantee success. ASO Designer provides high-quality predictions to save time and resources, but all candidates should be validated experimentally.
You can upload your own FASTA sequence, and ASO Designer will design ASOs directly from that input, no need to rely solely on built-in databases.
Yes. ASO Designer works with any transcript sequence, including coding and non-coding RNAs (e.g., lncRNAs).
No. The interface is designed to be intuitive for researchers, clinicians, and students. You only need the gene or sequence of interest, no coding required.
Yes. ASO Designer is freely available for academic and research use. Please cite the tool in any publications or presentations that use it.
We value every user. If you have questions or run into issues, please reach out at igem@tauex.tau.ac.il. Your feedback helps us improve and supports smoother research workflows.
Inputs include standard genomics and context files such as FASTA (or GenBank) and GFF/GTF annotations. Outputs include a ranked PDF report together with feature-rich tables for downstream analysis. Interfaces include a simple web UI for point-and-click runs and a PyPI package for local or CI workflows (pip install asodesigner).
Citation
ASO Designer: an open-source platform for antisense oligonucleotide design. iGEM Team TAU 2025.
Expand organism support
Add mouse and E. coli as first-class, validated builds, then extend to additional model organisms.
Provide strain/build selection and versioned transcriptomes with matching expression tables.
Collaborate with researchers
Partner with iGEM teams and external labs for co-designed case studies, usability testing,
and shared benchmark datasets. Establish a feedback loop (issues, surveys, office hours) to prioritize features that unblock real experiments.
