Poster
Variant-Aware Off-Target Analysis for Informed Guide RNA Selection
Abstract
Background: The specificity of CRISPR-based gene editing is critical for therapeutic success and safety. Off-target activity may cause unintended modifications, potentially disrupting normal gene function or introducing harmful genetic alterations. Careful consideration of off-target effects during guide RNA selection and development can significantly reduce these risks. It is also important to account for genetic diversity within human populations when evaluating off-target effects to effectively assess and mitigate risks in target patient populations. In this study, we combined in-silico and biochemical methods to implement a variant-aware approach for selecting guide RNA (gRNA) for editing the PCSK9 gene while considering population-scale genetic diversity. We then utilized the ONE-seq assay to obtain in-depth biochemical data, to thoroughly characterize off-target risk.
Methods: As described below, we generated data for guides targeting the PCSK9 gene using the CRISPR-Cas9 nuclease system while considering common variants in the major human superpopulations.
- Guide Profiler™ - This in silico computational tool was used to assess off-target risk profiles for eight candidate gRNAs against the human reference genome (Hg38) and 3,502 haplotype-phased genomes. The guides were ranked based on off-target burden, functional annotation and proximity to coding and regulatory regions. To evaluate the tool's utility in ranking off-target risks, PCSK9-1, PCSK9-3 and PCSK9-4 were selected for further screening.
- Guide Select™ - A multiplexed, variant-aware biochemical assay was performed for these three guides, covering all off-target sites with an edit distance of 4 from the guide sequence. The assay identified PCSK9-1 as the most specific, based on cleavage efficiency and off-target activity across diverse genetic backgrounds.
- ONE-seq™ - A high-sensitivity, variant-aware biochemical assay was performed for a comprehensive off-target risk assessment of PCSK9-1, mapping cleavage sites genome-wide and prioritizing high-risk loci through biochemical validation and deep sequencing.
Results: Guide PCSK9-1 was ranked highest by the Guide Profiler among all guides analyzed. The biochemical screen using Guide Select provided results consistent with the in silico analysis, identifying the fewest off-targets for PCSK9-1, confirming it as the optimal candidate. ONE-seq provided deep risk characterization, confirming minimal cleavage at high-risk loci and suitability for therapeutic applications.
Conclusion: This variant-aware, multi-step approach is a practical strategy for gRNA selection, systematically reducing off-target risk at each stage. By integrating computational and biochemical assays with population-scale variant analysis, we improve confidence in therapeutic genome editing and regulatory readiness.
Computational screening and off-target risk stratification of PCSK9 sgRNAs using Guide Profiler
A) Guide Select workflow
B) sgRNA sequences and IDs
| Target GENCODE region | Sponsor guide ID | Sequence |
|---|---|---|
| PCSK9 | PCSK9-1 | CCCGCACCTTGGCGCAGCGGtgg |
| PCSK9 | PCSK9-2 | GGTGGCTCACCAGCTCCAGCagg |
| PCSK9 | PCSK9-3 | GCTTACCTGTCTGTGGMGCggg |
| PCSK9 | PCSK9-4 | TGCTTACCTGTCTGTGGMGcgg |
| PCSK9 | PCSK9-5 | TTGGAAAGACGGAGGCAGCCtgg |
| PCSK9 | PCSK9-6 | GAAP/3ACGGAGGCAGCCTGGtgg |
| PCSK9 | PCSK9-7 | TCCCAGGCCTGGAGTTTATTcgg |
| PCSK9 | PCSK9-8 | AGCACCTACCTCGGGAGCTGagg |
C) Search parameters for library enumeration
| Search parameters | Enumeration |
|---|---|
| Maximum total differences from target sequence | 6 |
| Maximum mismatches | 6 |
| Maximum PAM mismatches | 1 |
| Maximum proto mismatches | 6 |
| Maximum gap bases | 2 |
| Maximum number of inserted bases | 2 |
| Maximum number of deleted bases | 2 |
| Maximum number of gaps between proto and PAM | 1 |
D) Guides in order of low to high off-target risk
Figure 1: A) Example of Guide Profiler workflow for in silico guide RNA screening. B) Sequences and IDs of sgRNAs targeting PCSK9 evaluated using Guide Profiler. C) Search parameters for off-target enumeration. D) Guides ordered by off-target risk (low to high) are shown here. This table shows a high-level summary of findings for all guides in the “All Loci” and “Loci ≤ 3 Diffs” sections. Columns in this section are colored by best to worst values (cream to burgundy). "Risk Profiler" functions as a guide for interpreting relative off-target risk of guides. Columns in this section are colored by best to worst values/ranks (dark green to white).
Population-scale variant analysis enhances off-target profiling of PCSK9 guides
A) Counts of haplotype-phased genomes via Guide Profiler
B) Number of off-target loci for all guides
C) Counts of potential off-targets by number of differences from PCSK9-1 guide sequence
D) Superpopulation Allele Frequency (AF) for off-targets with PCSK9-4 guide
E) Variant analysis at the on-target loci for PCSK9-2
| Superpopulation | Population | 1-55044044-C-G(-) |
|---|---|---|
| AFR | SAN(San) | 0.16700 |
| SAS | ITU(Telugu) | 0.00485 |
Figure 2: A) The map provides a visual representation of the counts of haplotype-phased genomes included in the Guide Profiler analysis. The size of each circle is proportional to the number of individuals in a subpopulation that is mapped to a superpopulation by color. AFR (African), EUR (European), SAS (South Asian), MID (Middle Eastern), EAS (East Asian), AMR (Mixed American) and OCE (Oceanian). B) Bar plot shows counts of off-target loci for each guide present in the reference and individual genomes (variants). C) Heat map showing the counts of potential off-targets by number of differences from guide sequence. As the number of differences from the guide increases (as indicated by the total row in the heat map), the number of potential off-targets increases. D) Superpopulation Allele Frequency (AF) information is shown for off-targets present in individual genomes (and not in hg38 genome) with fewer than three differences from the guide sequence. E) Variant analysis at the on-target loci revealed a high-frequency variant (AF = 16.7%) for the PCSK9-2 guide.
Biochemical validation of top-ranked PCSK9 sgRNAs from Guide Profiler using multiplexed Guide Select assay
A) Guide Select workflow
B) Candidate off-targets binned by cleavage frequency
| Guide |
Higher cleavage frequency (ONE-seq Screen score ≥ 0.01) |
Lower cleavage frequency (ONE-seq Screen score ≥ 0.001 & < 0.01) |
Total off-target loci screened |
|---|---|---|---|
| PCSK9-1 | 8 | 22 | 1910 |
| PCSK9-3 | 11 | 42 | 2014 |
| PCSK9-4 | 15 | 47 | 2718 |
C) Guide Select nominated off-targets by tier
D) Candidate off-target loci from tier 1 and tier 2a
E) Gene, transcript and regulatory region annotation
Figure 3: A) Example of Guide Select workflow for multiplexed, biochemical guide RNA screening. B) Count of candidate off-targets binned by cleavage frequency. A small number of sites screened in the Guide Select assay are cleaved at appreciable frequencies. C) Guide Select nominated off-targets are shown by tier for guides of interest. The on-target is not presented in this figure. D) Candidate off-target loci from tier 1 and tier 2a (high cleavage frequency) on the same chromosome as the on-target locus for all guides are shown here. Note, due to scaling, arrows indicating off-target sites very close in location may not be individually discernable. E) Counts of tier 1, 2a and 2b putative off-target loci that are linked to annotations from the Gencode and Encode databases.
Comprehensive off-target nomination and risk tiering of PCSK9-1 with ONE-seq
A) ONE-seq workflow
B) ONE-seq nominated off-targets by tier for PCSK9-1
C) Description of ERC tiers assigned to candidate off-target loci
| ERC tier | Description | ONE-seq score | Concern score | # Loci |
|---|---|---|---|---|
| 1 |
|
≥0.01 | ≥8 | 2 |
| 2a |
|
≥0.01 | <8 | 6 |
| 2b |
|
<0.01 | ≥8 | 5 |
| 3 |
|
<0.01 | <8 | 30 |
D) & E) Contribution of individual genomes to nominated off-target loci
Figure 4: A) Example of ONE-seq workflow for comprehensive off-target nomination and risk assessment. B) ONE-seq nominated off-targets are shown by tier for PCSK9-1. No on-target variation is observed. C) Description of ERC tiers assigned to candidate off-target loci. Higher ONE-seq scores indicate observation of higher cleavage frequencies in the ONE-seq assay. Annotation concern scores represent summary of genomic, functional and known disease associations of loci. D) & E) Contribution of individual genomes to the list of nominated off-target loci.
Summary
- SeQure DX™ offers a portfolio of assays to enable variant-aware guide selection and risk assessment.
- Guide Profiler was used to perform in silico, variant-aware screening of eight PCSK9-targeting sgRNAs,
ranking candidates by predicted off-target burden and functional annotation. - Population-scale genome analysis revealed guide-specific off-target sites and variants at the on-target locus not captured in the reference genome alone.
- Top-ranked guides were evaluated using the multiplexed Guide Select assay to biochemically assess specificity across diverse genetic backgrounds.
- ONE-seq provided comprehensive, genome-wide biochemical validation of the lead guide, confirming minimal off-target activity and supporting its suitability for therapeutic development.
