See every edit on every molecule
Long-read sequencing reads your edited locus end to end, on single DNA molecules. You get the exact edit carried by each allele, the frequency of rare events, and publication-ready evidence — from one isolated clone to a pool screened at 15,000 reads.
Don't miss what matters
A gene-editing experiment rarely produces one clean event. It produces a population of molecules: perfect edits, partial edits, indels, large deletions between two cut sites, and unintended insertions.
One molecule, one answer
Each long read covers the whole amplicon or construct, so every substitution and indel it carries is read on the same DNA molecule. A complete edit signature is observed directly, never inferred by combining separate fragments.
Large and complex events stay visible
Inter-cut deletions, insertions of several hundred bases, inversions and plasmid backbone integration span further than a short read. On a long read they appear as what they are, instead of silently dropping out of the alignment.
Alleles read separately
For clones, reads are clustered into their phased alleles before classification. You see whether an edit is homozygous, heterozygous, or whether one allele carries a large deletion the other does not.
Choose your sequencing depth
Choose your sequencing depth according to how rare you expect your events to be: the rarer the event, the more reads you need to detect it. All levels run on the same Oxford Nanopore R10.4.1 chemistry with SUP basecalling and CRISPR proprietary BioIT pipeline.
CRISPR Amplicon
1 isolated clone · up to 1,000 reads
Genotype a picked clone: which allele carries which edit, and is it homozygous or heterozygous.
- Clone validation before expansion
- Allele-level phasing
Amplicon pool DEEP
Pool · up to 5,000 reads
Measure editing efficiency on a bulk population and compare guides or conditions.
- Guide RNA benchmarking
- Bulk editing efficiency
Amplicon pool Ultra Deep
Pool · up to 10,000 reads
Resolve minority populations and low-efficiency editing strategies with tighter confidence.
- Low-efficiency HDR knock-in
- Minority populations resolved
Amplicon pool Extra Deep
Pool · up to 15,000 reads
Hunt rare events: prime editing in a pool, screening libraries, first-pass off-target questions.
- Rare-event detection
- Pooled screening strategies
Beyond the target site: whole-genome assessment
An amplicon tells you what happened where you looked. A whole genome tells you what happened everywhere else — the question a reviewer, a partner or a regulator will ask about an edited cell line.
On-target integration
Did the donor fragment integrate at the intended locus?
RESOLVED BY
Long reads spanning both junctions, so each side of the insertion is read on one molecule.
Insert architecture
Is the donor present as a single copy, or as a concatemer or a partial insert?
RESOLVED BY
Single reads spanning the full insertion, end to end, including repeated copies.
Genome-wide integrity
Did the donor integrate anywhere else, and did the gRNA create off-target lesions?
RESOLVED BY
Unbiased whole-genome long-read coverage — not restricted to computationally predicted sites.
Nanopore Whole Genome Long Read SMARTSeq
- DNA QC and library preparation — Qubit and TapeStation, then a ligation-based protocol with DNA end repair and adapter ligation
- Insert size optimisation — fragments targeted at ~10 kb to maximise output
- One dedicated PromethION flow cell per sample — no barcoding compromise
- Long-read sequencing — 70 to 80 Gb of data per sample on average
- Post-sequencing — POD5 raw signal converted to FASTQ with SUP basecalling for maximal base accuracy
Data volumes depend on the quality and quantity of the starting material. Turnaround: 20 working days. Available for human and mouse samples.
Starting material
| Requirement | Value |
|---|---|
| Recommended input | 3 µg of high molecular weight genomic DNA per sample |
| Absolute minimum | 1.5 µg per sample |
| Recommended extraction | Puregene kit (Qiagen) |
| Why that kit | It preserves long native molecules, which is what gives you long reads |
| Purity | Minimal protein, RNA, salt and organic contaminants |
What you receive
Raw signal & reads
- POD5 raw nanopore signal
- FASTQ from SUP basecalling
Alignments & assembly
- BAM and BAI against the reference genome and against your reference carrying the donor construct
- FASTA de novo assembly of the integration locus, with both junction sequences resolved
Tables & report
- TSV and XLSX: base-by-base quality and coverage, variant calls, off-target status per site
- HTML report with the annotated integration and the consolidated results
One pipeline, four scientific questions
Tell us what you are looking for; the analysis is configured for that question, not for a generic average.
✂ NHEJ — knock-out efficiency
Every insertion and deletion is extracted from the alignment and classified independently in a ±30 bp window around each annotated cut site. With several guides on the same amplicon, each cut site gets its own editing rate, so you can tell which guide actually worked.
- Editing rate per cut site, among reads that truly span it
- Indels inside vs outside the window counted separately
- Inter-cut (large) deletions reported as their own class
🧩 HDR — knock-in validation
Reads are compared against both your wild-type and your knock-in reference. The expected motif is checked base by base at the mutated position, and a read only counts as a confident knock-in when identity and coverage thresholds are met across the whole locus.
- Exact knock-in vs partial / low-quality knock-in
- HDR percentage among confidently classified reads
- Reads sorted into one FASTQ file per category
🎯 Prime Editing — rare events in a pool
A prime editing product is not one substitution: it is a complete haplotype of substitutions that must all be present on the same molecule. Reads are mapped to the wild-type reference, the base and its quality score are extracted at each target position, and the local haplotype is rebuilt read by read.
- Complete edit vs partial incorporation, never merged
- Indels within ±10 bp of the target kept as a separate class
- Wild-type controls processed identically, as a background reference
🔎 Transgene & integration sites
For knock-in lines and transgenic models, junction reads that align both to your construct and to the host genome pinpoint the integration locus, and show whether the whole construct or only part of it was integrated.
- Candidate insertion site with supporting junction reads
- Partial vs complete construct integration
- First-pass view of additional integration events
- Full genome-wide assessment available (see below)
Prime Editing: a signature, not a single base
A prime editing product is defined by several substitutions that must all appear on the same DNA molecule. Count them separately and you overestimate your efficiency.
For each read, the base and its quality score are extracted at every target position, and the local haplotype is rebuilt. Only a read carrying the full signature counts as a complete edit.
| Class | What it means for you |
|---|---|
| Complete edit | All expected substitutions on one molecule: your product |
| Partial edit | Part of the signature only; a possible intermediate, reported separately |
| Indel at target | Insertion or deletion within ±10 bp of the target: a different event |
| Other haplotype | An unexpected combination, to be compared with your controls |
| Wild type | Unedited at every target position |
Reading a rare-event frequency honestly
The prime editing frequency is reported as complete edits divided by informative reads — reads that passed mapping, identity, coverage and base-quality thresholds. Reads that could not support a call are excluded from that denominator and listed in their own categories.
- Your wild-type controls run through the exact same pipeline, giving you the background level of the method itself
- Frequencies are reported descriptively; no statistical test is applied automatically, so you stay in control of the comparison
- Every read keeps its identifier, so any molecule behind a percentage can be traced back in the FASTQ and viewed in IGV
How a read becomes a result
Every read is judged against explicit, published thresholds. Reads that cannot support a confident call are excluded from the denominator rather than quietly counted.
Analysis workflow
- Basecalling — Dorado SUP (super-accuracy) on R10.4.1 chemistry
- Quality report — NanoPlot: read length, quality, N50, cumulative yield
- Filtering — minimum read quality and a length window matched to your amplicon, removing short fragments and concatemers
- Alignment — minimap2 in map-ont preset, secondary alignments disabled, sorted and indexed with samtools
- Event extraction — indels read from the CIGAR string, target bases and their Phred scores read position by position
- Classification & frequency — one class per read, then frequencies over a stated denominator
| Quality criterion | Threshold | What it protects against |
|---|---|---|
| Mapping quality | MAPQ ≥ 30 | Reads placed at the wrong locus (~1/1000 error) |
| Alignment identity | ≥ 97% | Divergent reads creating false haplotypes |
| Base quality at target | Phred ≥ 15 | Calling an edit on a noisy base |
| Coverage of target | All positions covered | Partial reads counted as wild type |
| Alignment type | Primary only | Chimeric reads counted twice |
Reads that fail a criterion are reported in their own category (low MAPQ, low identity, ambiguous, not informative) so that nothing disappears silently from the analysis.
What your report looks like
Every project is delivered with an interactive HTML report. The figures below are built from synthetic data created for this page — no customer result is shown.
Editing rate per cut site
Three guides on the same amplicon. Guides A and B cut efficiently; guide C does not. Each rate is calculated only among reads that span that specific cut site.
Edit classes, allele by allele
Reads of a single clone split into two phased alleles, then classified. Allele B carries more insertions than allele A — a difference an averaged result would hide.
Every read, visible in IGV
You receive sorted and indexed BAM files with the reference and the target windows, so you can inspect any event yourself at base-pair resolution — and show it in a figure or to a reviewer.
| Quality check | Observed | Expected | Status |
|---|---|---|---|
| All required output files present | 0 missing | 0 missing | PASS |
| Reference length matches the submitted sequence | 496 bp | 496 bp | PASS |
| Read counts match between FASTQ and tables | 742 / 742 | identical | PASS |
| Classification frequencies sum to 100% | 100.0% | 100% | PASS |
| Event coordinates fall inside the reference | 0 outside | 0 outside | PASS |
Synthetic example of the automated QC table. Each report ends with these checks: the numbers in the figures, the tables and the FASTQ files must agree before the report is released.
Everything you receive
Delivered to your personal, secure Cloud Space — ready to analyse further, to archive, and to publish.
The interpretation
- HTML report: method, thresholds, figures, results and QC in one readable document
- Consolidated Excel: one sheet per level — summary, haplotype counts, and the classification of every single read
- Column dictionary: every column defined, so a colleague or a reviewer can follow
The evidence
- BAM + BAI sorted and indexed, ready for IGV
- Reference FASTA and its index, plus a BED file of the target windows
- FASTQ sorted by category: the reads behind each number, separately downloadable
The raw material
- Raw FASTQ from SUP basecalling, merged per sample
- Filtered FASTQ actually used by the analysis
- NanoPlot QC report: length and quality distributions, N50, yield
- Materials & Methods document, ready to adapt for a manuscript
How to order
Everything happens from your lab. No dry ice, no courier paperwork, no customs form.
Order online
Place your SMARTOrder in your SMART Account: number of samples, depth, and your SampleSheet with the reference sequences and gRNA.
Label your samples
Stick a Long Read SMARTSeq label on each tube and protect them in a SMARTBox, with your name and order number on the box.
Drop or hand over
Leave the box in your Fridge Box on site, or hand it to our team during the daily pick-up.
We sequence & analyse
Library preparation, sequencing and the CRISPR BioIT pipeline run in your local SMARTLab.
Get your results
Report, tables, BAM and FASTQ files land in your personal, secure Cloud Space.
What to send us
- Your DNA: purified PCR amplicon, or purified circular construct, meeting our concentration and volume requirements
- Your reference sequence in FASTA: wild type, and the expected edited sequence for knock-in or prime editing
- Your gRNA sequence(s), so cut sites and edit windows are placed automatically
- Your controls: an unedited sample processed in the same run is the best way to read a rare-event frequency with confidence
Not sure which depth you need?
Tell us the frequency you expect and the question you need to answer. We will recommend the depth, the controls and the amplicon design — before you spend anything.
Frequent questions
How long does it take?
Your results are delivered within 72 hours of sample collection — library preparation, sequencing and the full bioinformatics analysis included.
How deep should I sequence?
The deeper you sequence, the rarer the event you can observe and quantify. Tell us the frequency you expect and we will recommend a depth, the controls to include and the amplicon design before you order. Every report states the exact denominator used, so you always know what a percentage is calculated on.
Can you analyse several guides on one amplicon?
Yes. Each annotated cut site is analysed in its own window and gets its own editing rate, calculated among the reads that actually span that site.
When should I move from an amplicon to a whole genome?
An amplicon answers what happened at the locus you targeted. Move to whole-genome long-read sequencing when you need to prove where a donor integrated, whether it is a single copy, and whether anything happened elsewhere in the genome — typically before publishing a cell line, transferring it to a partner, or entering a regulated workflow. This service runs on a dedicated flow cell per sample, is available for human and mouse samples, is priced on quote and is delivered in 20 working days.
Do I need bioinformatics skills to read the results?
No. The HTML report is written to be read directly, and the Excel file gives the same numbers for your own figures. The BAM files are there if you want to inspect the alignments yourself in IGV, with a step-by-step guide included.
Can I publish these results?
Yes. Each delivery includes a Materials & Methods document with the software versions and every threshold used, and an automated QC table that must pass before the report is released.
What if my samples do not pass quality control?
We tell you before going further, with the NanoPlot report showing why, and we advise on how to prepare the next batch.
Secure the success of your gene editing experiment
Verify the complete sequence of your construct, quantify your edits allele by allele, and keep the evidence behind every number.
