With INDUCE-seq® you’re not just measuring outcomes. 

A lot of methods look at what’s left behind after editing; repair outcomes like indels and translocations. ±ő±·¶Ů±«°ä·ˇ-˛ő±đ±ç® looks directly at DNA breaks as they happen, providing a snapshot in time.  Timing matters a lot more than people expect, if sampled too late, the breaks have been repaired, likewise too early, and they haven’t been induced yet. 

The question isn’t just what you are measuring, it’s when you are measuring it. 

There isn’t a “standard” setup  

One of the most common things we see is reuse of the same experimental setups across different systems. 

The right design depends on a few things all moving together: 

• How efficient the editing is  

• How it’s being delivered 

• What editing system is being used  

• What cells are being worked in  

Change any one of those, and the timing can shift. 

For example: RNPs behave very differently to plasmids, Cas9 isn’t Cas12a and iPSCs don’t behave like primary cells. Therefore, copying a setup from a previous experiment, even one that worked well can quietly set up the experiment for failure. 

Timepoints are usually the make-or-break decision 

±ő±·¶Ů±«°ä·ˇ-˛ő±đ±ç® captures what’s happening at that exact moment, meaning that experiments need to land in the window where breaks are actively being formed. The easiest way to find that window is not to simply guess but to check. 

By running a quick time course and measuring indel levels over time will highlight when editing ramps up and when it levels off. That ramp-up phase is where the break signal is. It’s a simple step, but it saves a lot of frustration later. 

Controls aren’t just “good practice”, they’re what make the data usable 

±ő±·¶Ů±«°ä·ˇ-˛ő±đ±ç® will pick up all types of DNA breaks in the cell. Without proper controls, it can be guess-work to identify which are editing induced. At the very least the following controls are required: 

• A delivery-matched negative control at each timepoint  

• deally an untreated baseline as well  

Beyond that, positive and assay controls help to sanity-check that everything’s behaving as expected. It might feel like extra work, but it’s what turns a dataset into something that is easier to interpret. 

Replicates tell you more than you think 

It’s easy to treat replicates as a box to tick, with ±ő±·¶Ů±«°ä·ˇ-˛ő±đ±ç®, they’re quite revealing. Because the workflow is PCR-free and quantitative, consistency across replicates is a clear signal of whether things are working properly. 

• Sample replicates → are things stable within the setup?  

• Biological replicates → are the breaks consistent across experiments, or just specific to that one run?  

Including appropriate replicates isn’t optional. FDA guidance states that multiple biological replicates are expected for off-target nomination and confirmation for regulatory submission.  

You can’t optimise everything, so decide what matters 

Every experiment has trade-offs. More timepoints means a better understanding of editing kinetics. More replicates result in stronger confidence. More conditions allow for a broader comparison. However, all three can’t be maximized at once. The experimental design must be deliberate. 

If guides are being screened, keep it simple, pick a good timepoint and prioritize throughput. 

If moving towards lead selection or IND work, there’s a need to go deeper, so, more timepoints with better replication and in more relevant cells.  

Different stages have a need for different priorities. 

Most problems we see aren’t technical, they’re design-related 

When experiments don’t work, it’s rarely because the assay failed, it’s usually: 

• Low levels of editing 

• Wrong timepoint  

• Missing controls  

• Insufficient number of replicates 

• A mismatch between the design and the biology  

If these are well considered everything else becomes a lot easier. 

Final thought 

There isn’t a perfect template for designing these types of experiments, however, there is a common thread in the ones that work well. They’re thought through upfront and consider the biology seriously. They don’t try to shortcut the design stage and the data tends to speak for itself.  

To read about ±ő±·¶Ů±«°ä·ˇ-˛ő±đ±ç® experimental design considerations in more detail, please check out our technical note  

Permalink

]]>Why Experimental Setup Determines Off-target Data Quality Gene Editing±ő±·¶Ů±«°ä·ˇ-˛ő±đ±ç®DNAJamie HarmesTue, 10 Mar 2026 10:53:28 +0000/blog/induce-seq-a-new-standard-for-genome-wide-dna-break-characterisation-in-gene-editing69429e7198b87e513d8a40ea:698f03926f313a0b9d7ba0a9:69aff17189e77a22f5d820afGene editing safety is ultimately a question of DNA break behaviour. Understanding where breaks occur, how often, under what conditions and understanding if those events can be measured reproducibly in the cell types that matter. As gene editing technologies move from research settings toward clinical translation, answering these questions with precision is now foundational.

Despite the rapid evolution of CRISPR-Cas systems, base editors, and prime editors, the tools used to characterize off-target activity have not always kept pace. Many widely adopted methods rely on indirect readouts, PCR-amplified libraries, or fragmented multi-assay workflows. Others measure the final genomic outcome long after editing has occurred, rather than capturing the break event itself.

The consequence is familiar to many teams: incomplete visibility, ambiguous interpretation, and difficulty standardising data across discovery, optimisation, and IND-enabling studies. ±ő±·¶Ů±«°ä·ˇ-˛ő±đ±ç® has been developed to address this gap directly. 

Why direct DNA break characterization matters

Nuclease-based gene editing systems are designed to introduce targeted DNA damage at specific genomic loci. However, no editing system operates with perfect specificity. In addition to the intended on-target modification, unintended off-target activity can occur across the genome.

For nuclease-based systems, these events often manifest as double-strand breaks (DSBs), which represent a significant genotoxic risk. Off-target DNA breaks can lead to large-scale genomic rearrangements, activation of oncogenes, disruption of tumour suppressor genes, and other adverse outcomes.

Importantly, even newer approaches such as base editing and prime editing, often positioned as avoiding DSBs have been shown to induce DNA breaks and other forms of genotoxicity under certain conditions.

As regulatory expectations evolve, empirical, genome-wide evidence of editing activity is increasingly required. Regulatory agencies including the FDA and EMA now expect unbiased, genome-wide characterisation in clinically relevant cell types, rather than reliance solely on in silico or in vitro approaches.

There is also a clear shift toward earlier assessment during discovery, enabling teams to eliminate suboptimal candidates before significant time and cost are invested.

Despite this, many programmes still lack a scalable, sensitive, cell-based method capable of directly measuring both on-target and off-target DNA break activity early enough to influence decision-making. 

A different approach: capturing breaks at their point of occurrence

±ő±·¶Ů±«°ä·ˇ-˛ő±đ±ç® is a scalable, genome-wide, in cellulo platform for the direct detection and quantification of DNA breaks.

Rather than extracting genomic DNA first and labelling break ends later, ±ő±·¶Ů±«°ä·ˇ-˛ő±đ±ç® performs in situ break labelling within fixed and permeabilised cells. This preserves the genomic context of break events as they existed inside the cell and avoids distortions introduced by post-extraction manipulation and PCR amplification.

Each break end is directly labelled with sequencing adapters, enabling sequencing to initiate from the break itself. Because the workflow is PCR-free, each sequencing read corresponds to a single captured break event, providing a quantitative and unbiased representation of DNA break frequency.

This PCR-free design is central to quantitative confidence, particularly when measuring low-frequency off-target events. 

From break labelling to sequencing

Following in situ labelling, genomic DNA is extracted and mechanically fragmented to generate sequencing-compatible fragments. A partially functional sequencing adapter is ligated to fragmented ends, creating a selective library architecture.

Only fragments that carry both the break-labelled adapter and the complementary sequencing adapter form functional constructs capable of binding to the sequencing flow cell. Unlabelled genomic fragments are rendered non-functional.

This selective library design enriches specifically for break-labelled fragments, dramatically improving sensitivity while reducing the sequencing depth required to detect rare events.

The output is genome-wide, single-nucleotide resolution mapping of DNA breaks across the entire genome.  

Integrated bioinformatics for decision-ready outputs

Sequencing data generated by ±ő±·¶Ů±«°ä·ˇ-˛ő±đ±ç® are processed through an integrated bioinformatics platform purpose-built for genome-wide break mapping.

Reads are mapped to the reference genome, break sites are resolved at base-level precision, and candidate on- and off-target sites are identified through a dual analytical framework combining frequency-based and homology-based analysis.

This approach enables:

• Quantitative assessment of break frequency (enabled by PCR-free design)

• Cross-referencing with predicted cleavage sites

• Nomination and ranking of candidate off-target sites based on evidence of nuclease-induced activity

Each candidate site is assigned a probability score reflecting the likelihood of true induction versus background noise.

Outputs are delivered through structured, interpretable reports, including nomination tables, break site plots, mismatch plots, and supporting datasets suitable for both discovery optimisation and regulatory documentation.

The emphasis is not simply on detection, but on clarity, prioritisation, and decision-making.

What ±ő±·¶Ů±«°ä·ˇ-˛ő±đ±ç® delivers

At its core, ±ő±·¶Ů±«°ä·ˇ-˛ő±đ±ç® provides:

• Genome-wide, single-nucleotide resolution mapping of DNA breaks

• Simultaneous characterisation of on-target and off-target activity within a single workflow

• Measurement of both induced and endogenous background DNA breaks

• Compatibility with major gene editing systems, including CRISPR, TALENs, Zinc-Fingers, as well as base and prime editing

• Broad applicability across primary cells, stem cells, T cells, iPSCs, and immortalised cell lines

• A standardised, in-house workflow capable of delivering results within days

Running the platform in-house ensures full control of data, auditability, and programme confidentiality—an increasingly important consideration as programmes move toward regulatory submission.

Applications across the gene editing pipeline

One of the strengths of ±ő±·¶Ů±«°ä·ˇ-˛ő±đ±ç® is its flexibility across development stages. 

Discovery
±ő±·¶Ů±«°ä·ˇ-˛ő±đ±ç® enables rapid, side-by-side comparison of guide RNAs, nuclease variants, and editing conditions, generating full on- and off-target profiles to support early candidate selection and programme de-risking.

Lead characterisation and optimisation
By sampling multiple timepoints post-editing, teams can capture break formation and repair dynamics, providing insight into editing kinetics, nuclease behaviour, and cell-type specific responses. These data inform optimisation strategies, delivery approaches, and nuclease engineering decisions.

Translational and IND-enabling studies
Regulatory expectations increasingly require unbiased, genome-wide data generated in clinically relevant systems using well-characterised methods. ±ő±·¶Ů±«°ä·ˇ-˛ő±đ±ç® provides reproducible, standardised outputs suitable for inclusion in IND data packages, with clear structure and traceability.

Across each stage, the workflow remains consistent, reducing variability and enabling continuity from discovery through to clinical translation.

Raising the standard for break analysis

As gene editing technologies become more powerful, the standard for genomic safety assessment rises alongside them.

Empirical, genome-wide evidence of editing activity is no longer a late-stage requirement—it is an expectation throughout development.

±ő±·¶Ů±«°ä·ˇ-˛ő±đ±ç® addresses this need by directly capturing DNA breaks at their point of formation, within intact cells, without PCR amplification. The result is quantitative precision, single-nucleotide resolution, and an integrated analytical framework that translates complex sequencing data into clear, decision-ready outputs.

For gene editing teams seeking to de-risk programmes, accelerate iteration cycles, and build robust datasets aligned with modern regulatory expectations, ±ő±·¶Ů±«°ä·ˇ-˛ő±đ±ç® represents a shift from indirect inference to direct measurement.

And in genome editing, direct measurement is what ultimately builds confidence.

Permalink

]]>±ő±·¶Ů±«°ä·ˇ-˛ő±đ±ç®: A New Standard for Genome-Wide DNA Break Characterization in Gene EditingGene EditingBioinformatics±ő±·¶Ů±«°ä·ˇ-˛ő±đ±ç®Jamie HarmesTue, 28 May 2019 09:27:55 +0000/blog/solving-the-off-target-analysis-bottleneck-decision-focused-bioinformatics-for-gene-editing69429e7198b87e513d8a40ea:698f03926f313a0b9d7ba0a9:698f03926f313a0b9d7ba0b0Gene editing programs no longer struggle to generate data. They struggle to interpret it. 

Genome-wide off-target mapping technologies have advanced rapidly in recent years. It’s now routine to generate hundreds to thousands of putative off-target sites from a single experiment. Detection sensitivity has improved. Sequencing costs have fallen. Throughput has increased. Yet the critical question remains surprisingly difficult to answer: Which of these sites matter? 

As editing programs move from early discovery toward IND-enabling studies, the pressure shifts from identifying events to discriminating between them. The challenge is no longer technical detection. It’s decision clarity. 

The hidden fragmentation in off-target analysis 

Across the industry, wet-lab technologies and bioinformatics workflows are often developed in parallel rather than in partnership. A laboratory assay may be robust and reproducible, but the downstream analysis pipeline frequently relies on adapted academic tools, custom scripts, or loosely maintained research software. 

This separation creates friction. Analytical assumptions may not fully reflect the chemistry of the assay. Updates to one side of the workflow are not always mirrored on the other. As programs scale, these small disconnects compound. 

In early research environments this may be manageable. In translational or regulated settings, it becomes a risk. 

Bioinformatics tools used for off-target assessment often originate in academic groups where innovation is prioritised over long-term maintenance. They can be powerful in expert hands, but they are rarely built for cross-functional biotech teams working under timeline pressure. Documentation may be light. Compute requirements may be heavy. Reproducibility between operators may depend on specialist knowledge. 

None of this is inherently flawed. But it is not optimised for industrial development. 

When outputs don’t drive action 

Most pipelines focus on identifying and reporting putative off-target sites. They generate extensive tables of genomic positions, event` counts, and statistical values. For data scientists, this level of detail is necessary. For programme leaders, however, it can obscure the central question: what should we prioritise next? 

A list of thousands of detected sites does not equate to a prioritised off-target profile. Without structured ranking, replicate-aware filtering, and treated-versus-control normalisation, interpretation becomes manual and iterative. Weeks can be spent moving from raw output to a defensible shortlist of candidate sites. 

At scale, this slows experimental cycles and introduces subjective interpretation. The bottleneck is not sequencing depth. It is analytical discrimination. 

Moving from detection to discrimination 

As gene editing technologies mature, analytical expectations must mature with them. A robust off-target workflow should not simply catalogue genome-wide breaks. It should distinguish likely editor-induced events from endogenous background noise and low-confidence signals, using quantitative and statistical frameworks that are transparent and reproducible. This requires bioinformatics that is deliberately designed around the assay generating the data. 

±ő±·¶Ů±«°ä·ˇ-˛ő±đ±ç® Analysis: tightly coupled assay and analysis 

±ő±·¶Ů±«°ä·ˇ-˛ő±đ±ç® Analysis was developed alongside the ±ő±·¶Ů±«°ä·ˇ-˛ő±đ±ç® wet lab assay with that principle in mind. Rather than adapting a generic sequencing pipeline, the analytical framework was designed specifically for PCR-free double-strand break mapping at genome scale. 

The workflow begins with rigorous read processing. FASTQ files undergo quality assessment and trimming before alignment to the selected reference genome. Break positions are resolved at base-level precision and merged across replicates to generate proportional genome-wide break counts. The output is not simply mapped reads, but a quantitative representation of break frequency across the genome. 

Each detected break site is then annotated in biological context. Intersection with genes and repeat regions is assessed, reproducibility across replicates is evaluated, and proximity to guide-like sequences is considered where relevant. This contextual layer allows interpretation to move beyond position alone. 

Crucially, break sites detected in treated samples are compared directly with matched controls. By generating normalised treated-to-control ratios at identical genomic positions, endogenous background breaks can be separated from treatment-associated signals. This step materially improves signal discrimination and reduces false prioritisation. 

From there, quantitative and statistical modelling is applied to nominate a subset of high-confidence induced break sites from the thousands detected. Rather than presenting users with an undifferentiated catalogue, the platform produces a prioritised and defensible shortlist suitable for downstream validation or regulatory assessment. 

The emphasis is not simply on finding breaks, but on ranking them in a way that supports confident decision-making. 

Designed for accessibility without sacrificing depth 

One of the persistent tensions in bioinformatics is accessibility versus analytical sophistication. Powerful pipelines often require command-line execution, parameter tuning, and cluster management. This places analysis in the hands of a small number of specialists and can create dependency bottlenecks within growing teams. 

±ő±·¶Ů±«°ä·ˇ-˛ő±đ±ç®  Analysis addresses this by integrating compute and interface within a single platform. Analyses are launched through a browser-based graphical interface, and cloud resources are provisioned on demand. From FASTQ upload to interactive report, processing typically completes in under two hours. 

This removes the need for local infrastructure, pipeline maintenance, or specialist compute configuration. At the same time, detailed tabular outputs remain available for data scientists who require deeper interrogation. 

The goal is not to simplify the science. It is to remove unnecessary operational friction. 

Shortening the path from experiment to decision 

As gene editing programs advance toward clinical translation, timelines tighten and expectations rise. Off-target data must be robust, reproducible, and clearly interpretable. Regulatory discussions demand defensible prioritisation rather than raw detection counts. 

By tightly integrating assay chemistry with a purpose-built analytical engine, ±ő±·¶Ů±«°ä·ˇ-˛ő±đ±ç®  Analysis shortens interpretation timelines from weeks to days. More importantly, it reduces ambiguity. Teams can move from genome-wide detection to structured nomination without relying on fragmented toolchains or manual filtering cycles. 

In an environment where gene editing platforms continue to evolve, analytical clarity is no longer optional. It is foundational. 

Detection will continue to improve. Sensitivity will increase. Throughput will expand. 

But without integrated, decision-focused bioinformatics, more data does not mean better decisions. 

And in translational gene editing, decisions are what matter. 

Permalink

]]>Solving the Off-Target Analysis Bottleneck: Decision-Focused Bioinformatics for Gene Editing