In Silico Off-Target Prediction for ASO/siRNA

In Silico Off-Target Prediction for ASO/siRNA

1. Background

Antisense oligonucleotides (ASOs) and small interfering RNAs (siRNAs), as next-generation precision nucleic acid therapeutics, achieve targeted regulation of gene expression through sequence-specific recognition of RNA. They have emerged as pivotal modalities for the treatment of rare diseases, genetic disorders, neurodegenerative diseases, and metabolic conditions. ASOs primarily exert their effects via RNase H-mediated mRNA degradation or steric hindrance, while siRNAs rely on RNA-induced silencing complex (RISC)-mediated cleavage of target mRNA, offering superior target selectivity. Compared with conventional small-molecule drugs and antibody therapeutics, nucleic acid medicines offer distinct advantages including broad target coverage, shortened design cycles, and customizable sequences—enabling drug development against targets previously considered intractable. They have thus secured a critical strategic position in global biopharmaceutical pipelines.

With several ASO and siRNA therapeutics receiving regulatory approvals worldwide—spanning indications such as spinal muscular atrophy (SMA), hereditary transthyretin-mediated amyloidosis (hATTR), acute hepatic porphyria, and hypercholesterolemia—the clinical value and commercial viability of nucleic acid drugs have been thoroughly validated. Simultaneously, the global nucleic acid drug pipeline continues to expand, with a growing number of candidate molecules advancing through preclinical and clinical development stages, signaling a broad market outlook. This sector has become one of the most promising growth areas in contemporary biopharmaceutical development.

Nevertheless, the success rate of nucleic acid drug development is critically dependent on precision sequence design and comprehensive safety evaluation. Safety risks associated with ASOs/siRNAs arise primarily from two dimensions: (1) Sequence-dependent (hybridization-dependent) off-target effects—due to high sequence similarity, a drug may form unintended hybridization complexes with non-target genes, causing aberrant expression of off-target transcripts and constituting a major source of preclinical toxicity signals; (2) Sequence-independent effects—such as complement activation and coagulation interference (class effects) associated with ASOs, and immune stimulation triggered by siRNA activation of innate immune receptors including TLR7/TLR8. In recent years, ICH, major global regulatory agencies (FDA, EMA, PMDA, CDE), and the Oligonucleotide Safety Working Group (OSWG) have published specific technical guidance documents, establishing clear and systematic requirements for the safety evaluation of nucleic acid drugs.

Figure 1. Overview of Key Guidance Documents from ICH/FDA/EMA/PMDA/CDE

In silico off-target prediction, as a critical component of early drug discovery, enables systematic identification of potential off-target sites prior to in vitro experimentation, providing predictive reference for subsequent experimental validation and significantly reducing both development costs and timelines. In November 2024, the FDA published the Nonclinical Safety Assessment of Oligonucleotide-Based Therapeutics guidance, explicitly requiring that prior to IND submission, candidate sequences and their metabolites undergo comprehensive hybridization-dependent off-target assessments covering the transcriptome, nuclear genome, and mitochondrial genome. These in silico predictions must be cross-analyzed with RNA-seq transcriptomic experimental validation. Systematic and standardized safety evaluation has become a fundamental prerequisite for advancing oligonucleotide therapeutics through the IND filing process.

Figure 2. FDA Requirements for Sequence-Dependent Hybridization Off-Target Assessment of Oligonucleotide Therapeutics

(The guidance explicitly mandates systematic evaluation of potential ONT off-target sites using a combined in silico and in vitro approach, including RNA-seq)

ZhuHai GeneRulor, drawing on extensive experience in safety assessment services, has established a regulatory-compliant ASO/siRNA in silico off-target prediction platform by integrating cutting-edge bioinformatics algorithms with authoritative database resources—providing scientific and comprehensive data support for nucleic acid drug safety evaluation.

2. Principles of In Silico Off-Target Prediction

In silico off-target prediction is grounded in Watson-Crick base-pairing principles, employing sequence alignment algorithms to systematically search for nucleic acid fragments with high similarity to the query ASO/siRNA sequence across the entire genome and transcriptome. Even with a limited number of base mismatches, such sequence similarity can lead to non-specific binding and interference with normal function of unintended genes.

The prediction workflow adopts a hierarchical assessment strategy:

·Whole-sequence scanning: Comprehensive search of the reference genome's transcriptome, nuclear genome, and mitochondrial genome using the full drug sequence (e.g., a 20 bp ASO) and its potential in vivo metabolites (e.g., truncated sequences).

·Mismatch-tolerant alignment analysis: Setting rational mismatch thresholds according to drug type to systematically identify all potential binding sites meeting the defined similarity criteria.

·Risk-stratified annotation: Integration of tissue expression data, gene function databases, disease association data, and essential gene information for multidimensional safety assessment of identified off-target sites.

Through these analyses, high-, medium-, and low-risk off-target sites are comprehensively identified, providing predictive references for in vitro experimental validation. In silico prediction and RNA-seq transcriptomic validation can be conducted independently or concurrently; cross-analysis of both approaches enhances the completeness of the off-target assessment.

Figure 3. Schematic Illustration of ASO/siRNA Off-Target Effects

3. Technical Features and Advantages

（1）Comprehensive Assessment Meeting Regulatory Requirements

Strictly adhering to FDA guidance, the assessment scope covers the transcriptome, nuclear genome, and mitochondrial genome, with analysis encompassing both the full drug sequence and its potential in vivo metabolites. Differentiated prediction strategies for different drug types (ASO/siRNA) ensure comprehensiveness and accuracy in off-target risk assessment.

（2）Precise Risk Stratification to Identify High-Risk Targets

Based on sequence homology analysis, combined with target tissue expression data, essential gene databases, and disease association databases, a three-tier risk stratification system (high/medium/low risk) is established to precisely identify off-target sites requiring priority attention, providing clear direction for research and development decision-making.

（3）Rapid Turnaround Seamlessly Integrated with Experimental Workflows

The in silico prediction cycle is short (10 business days), significantly reducing early-stage development costs. Prediction results can be cross-analyzed with RNA-seq transcriptomic validation experiments to form a complete off-target assessment framework, while simultaneously supporting the regulatory data requirements for IND submissions.

Figure 4. Off-Target Prediction Analysis Workflow: Sequence Input→Whole-Genome Scanning→Off-Target Site Identification→Functional Annotation→Risk Stratification

4. Application Scenarios and Service Advantages

4.1 Application Scenarios

·Early candidate sequence screening: During the drug design phase, multiple candidate ASO/siRNA sequences are evaluated for off-target risk to prioritize sequences with lower off-target risk for downstream development.

·IND submission support: Providing regulatory-compliant in silico off-target prediction reports as an essential component of the nonclinical safety evaluation package.

·Experimental validation reference: Generating a list of predicted off-target sites for RNA-seq transcriptomic validation experiments; cross-analysis of in silico predictions and experimental results enhances assessment credibility.

·Sequence optimization guidance: Providing scientific basis for drug sequence optimization and redesign based on sequence-feature analysis of identified off-target sites.

·Retrospective failure analysis: Retrospectively investigating potential off-target risk sources for drug candidates that have exhibited safety signals during preclinical or clinical development.

4.2 Service Advantages

·Regulatory compliance: Strictly adhering to the FDA Nonclinical Safety Assessment of Oligonucleotide-Based Therapeutics guidance (2024) and relevant ICH guidelines; assessment scope and methodology meet regulatory requirements.

·Rapid and efficient: Short in silico prediction cycle (10 business days) significantly accelerates development timelines and reduces early development costs.

·Professional and reliable: Backed by extensive experience in nucleic acid drug development, integrating multiple authoritative bioinformatics databases to ensure accuracy and comprehensiveness of analytical results.

·Customized services: Supporting human, mouse, and other common model organisms; personalized analysis schemes available based on client targets, indications, and development stage.

·Complementary validation: In silico prediction and RNA-seq transcriptomic validation can be conducted independently or concurrently; cross-analysis of both forms a complete off-target assessment framework.

5. Representative Report Highlights

The off-target prediction analysis reports delivered by ZhuHai GeneRulor are comprehensive in content and clearly structured, encompassing detailed off-target site annotation, risk stratification assessment, functional enrichment analysis, and other core components. The following presents selected key results from a representative report (using an HBV-targeting ASO project as an example).

（1）Off-target site prediction results: Based on whole-genome sequence alignment analysis, potential off-target sites of the ASO sequence and its metabolites are systematically identified. Prediction results include detailed annotations for each off-target site, including genomic coordinates, gene/transcript information, mismatch patterns, and strand orientation.

Figure 5. Example of Off-Target Site Prediction Results

（2）Risk stratification of off-target genes: Tissue expression data, essential gene databases, and disease association databases are integrated to conduct multidimensional safety assessments of the identified off-target genes, establishing a three-tier risk stratification framework (high/medium/low risk) to precisely pinpoint sites requiring priority attention.

Figure 6. Essentiality Assessment (LOEUF Score) of High-Risk Off-Target Genes

（3）Tissue expression relevance analysis: Based on authoritative tissue expression databases, the expression profiles of off-target genes in the target tissue of drug action (liver) are systematically evaluated. Off-target genes expressed in hepatic tissue are selected as priority subjects for subsequent safety assessment.

Figure 7. List of Off-Target Genes Expressed in the Target Tissue (Liver)

（4）Functional enrichment analysis: GO functional enrichment and KEGG pathway enrichment analyses are performed on high- and medium-risk off-target genes to systematically characterize their core biological functions and key signaling pathways, providing a scientific basis for mechanistic interpretation and risk evaluation of off-target effects.

Figure 8. GO Functional Enrichment Analysis of High-Risk Off-Target Genes

Figure 9. GO and KEGG Enrichment Results Visualization for Medium-Risk Genes

6. Service Scope and Deliverables

Deliverables:

·Complete off-target site annotation dataset (Excel format)

·High-risk off-target gene list with functional annotation

·Off-target prediction analysis report (PDF format)

·Functional enrichment analysis results and visualization charts

7. Sample Submission Requirements and Project Timeline

8. References

[1] U.S. Food and Drug Administration. Nonclinical Safety Assessment of Oligonucleotide-Based Therapeutics Guidance for Industry. Draft Guidance. November (2024).

[2] Yoshida, T. et al. Evaluation of off-target effects of gapmer antisense oligonucleotides using human cells. Genes Cells 24, 827–835 (2019).

[3] Andersson, P. et al. Assessing Hybridization-Dependent Off-Target Risk for Therapeutic Oligonucleotides: Updated Industry Recommendations. Nucleic Acid Ther. (2024).

[4] Goyenvalle, A. et al. Considerations in the preclinical assessment of the safety of antisense oligonucleotides. Nucleic Acid Ther. 33, 1–16 (2023).

[5] Kamola, P. J. et al. In silico and in vitro evaluation of exonic and intronic off-target effects form a critical element of therapeutic ASO gapmer optimization. Nucleic Acids Res. 43, 8638–8650 (2015).

[6] Lindow, M. et al. Assessing unintended hybridization-induced biological effects of oligonucleotides. Nat. Biotechnol. 30, 920–923 (2012).