Figure 1. Transgene integration characterization separates four evidence questions that should be planned and reported independently.

A positive PCR result confirms that at least part of a transgene is present, but it does not establish where the construct integrated, how many copies are present, how those copies are arranged, or whether the intended insert remains intact. This guide separates those four questions and shows how sequencing, quantitative assays, and orthogonal validation can be combined for stable cell lines, BAC constructs, and transgenic animal models.

This content concerns research-use genomic characterization. CD Genomics does not provide clinical diagnosis or treatment recommendations through these services.

Key Takeaways

• Treat location, copy number, orientation, and integrity as separate measurements. One assay rarely answers all four with equal confidence.
• Require evidence across both host–transgene junctions. A coordinate without junction sequence or read-level support is incomplete.
• Interpret copy number at the right level. Total copies per genome, copies per locus, and copies per allele are not interchangeable.
• Use long-range evidence for complex arrays. Concatemers, inversions, backbone fragments, and local host rearrangements can remain hidden when analysis depends only on short amplicons.
• Define the final evidence package before sequencing. The project design should specify reference sequences, controls, expected outputs, and rules for orthogonal confirmation.

Four Questions Define the Evidence

Transgene characterization often begins with a single request: "Can you confirm the integration?" That wording is too broad for study planning. Confirmation may refer to the presence of a transgene-specific sequence, a mapped host–transgene junction, a stable copy-number estimate, or a reconstructed allele that spans the complete inserted sequence. These are different endpoints.

The first planning step is to convert the request into four explicit questions. Researchers who need a coordinated approach can review the Transgene Integration Site Analysis service, but the assay choice should still follow the evidence required by the study.

Figure 2. A claim-first decision map connects each characterization question to the evidence and deliverables needed to answer it.

Map the Integration Locus

An insertion-site call should connect known transgene sequence to uniquely placed host sequence. Split reads that cross a host–transgene boundary provide direct evidence. Discordant read pairs or local coverage changes can support the call, but they usually do not define the breakpoint with the same precision.

Two boundaries, not one coordinate

For a simple insertion, the left and right junctions anchor the complete event to the host genome. One detected boundary may still be useful, especially in a repeat-rich region, but it does not prove that the opposite side has the expected structure. The undetected boundary may contain a deletion, inversion, host-derived insertion, or rearrangement that prevents straightforward amplification or mapping.

Reports should state the reference genome and version, chromosome or contig, breakpoint coordinates, strand, supporting read identifiers, alignment quality, and whether each boundary was independently confirmed. The guide to validating low-frequency integration sites describes how junction support, mapping context, controls, and duplicate handling affect confidence. Although that guide focuses on AAV, the evidence principles apply more broadly.

Multiple loci and mixed populations

A clonal line and a pooled population pose different problems. In a clone, multiple junctions may indicate several integration loci or one complex rearranged allele. In a pool, the same set of junctions may arise from distinct subclones. Read abundance cannot be treated as a direct cellular frequency without considering library enrichment, amplification, mapping bias, and DNA input.

This distinction matters for engineered cell populations and for founder animals that may be mosaic. If the biological question concerns stable inheritance or a defined production clone, follow-up testing should use a material stage that represents that question. For integrating viral vectors, the lentiviral integration sites analysis service is relevant when site distribution and comparative integration patterns are the primary endpoints.

Measure Copy Number Carefully

Copy number sounds like a single number, but its denominator and biological level must be stated. A report may describe copies per diploid genome, copies per cell, copies per transgenic allele, or an average across a heterogeneous sample. Those values answer different questions.

Total copies versus locus copies

Quantitative PCR and digital PCR can estimate total abundance of a selected transgene region relative to a reference locus. Sequencing depth can provide an independent estimate by comparing coverage over the transgene with coverage over suitable host regions. Neither approach automatically assigns each copy to a specific integration locus.

If a line carries two integration loci, total copy number can be correct while the distribution between loci remains unresolved. Conversely, a junction-based method can identify two loci without showing how many tandem copies lie within each locus. Whole Genome Sequencing can add genome-wide context and depth information, while targeted quantitative assays can provide higher measurement precision for predefined sequences. The two data types are complementary.

Assay design changes the answer

Primer or probe placement determines which sequence is counted. An assay targeting the promoter will miss a promoter-deleted copy. An assay inside a repeated coding region may count partial fragments that retain that segment. A single-copy host reference also requires attention to species, ploidy, sex chromosomes, and known copy-number variation.

Before accepting a copy-number result, check four items:

• Which exact region of the construct is measured?
• Is the value normalized per haploid genome, diploid genome, cell, or allele?
• Does the sample contain one clone, a mixed pool, or potentially mosaic tissue?
• Was the estimate compared with an independent method or a known-copy control?

Copy number is most informative when it is interpreted beside the structural map. Agreement between digital quantification, read depth, and the number of reconstructed repeat units strengthens the result. Disagreement is not merely a QC failure; it can reveal truncation, duplicated assay targets, sample mixture, or incomplete assembly.

Resolve Orientation and Concatemers

Random integration frequently creates more than a single forward copy. The inserted allele may contain head-to-tail repeats, head-to-head inversions, partial fragments, duplicated vector elements, or unexpected sequence from the plasmid backbone. Local host sequence may also be deleted, duplicated, or rearranged during integration [1–3].

Short reads can identify host–transgene junctions and estimate coverage, but repeated construct sequence makes long-range ordering difficult. A read that spans only one internal junction cannot establish the complete array. Long reads and long-fragment enrichment can connect several components on the same molecule, which makes them useful for resolving order and orientation [4]. The short-read versus long-read AAV sequencing guide provides a related platform comparison; the present decision remains driven by insert size, repeat complexity, DNA integrity, and the required resolution.

Figure 3. Long-range evidence helps distinguish simple inserts from tandem, inverted, partial, and rearranged integration structures.

Prove Full-Length Insert Integrity

Integrity means continuity and identity across the intended construct, not simply detection of its endpoints. A full assessment asks whether every expected component is present in the correct order and whether any unexpected component is inserted between them.

Start with the exact reference

The analysis needs the final construct sequence used in the experiment, not a generic vector map or an earlier plasmid version. The reference should include the intended cargo, regulatory elements, linkers, homology arms when applicable, and the backbone sequence that should be screened as an unwanted integration. If the study used a BAC, the complete BAC reference and known assembly gaps should be supplied.

Once the correct reference is available, the data can be checked for:

• coverage across every expected base or defined interval;
• single-nucleotide changes and small insertions or deletions;
• truncations at construct boundaries or internal elements;
• duplicated, inverted, or rearranged components;
• backbone, helper-vector, adapter, or host-derived sequence;
• changes in the host genome immediately surrounding the insertion.

Targeted Region Sequencing can support focused confirmation when the locus and expected sequence are already known. For large or repetitive inserts, targeted short fragments may need long-read or genome-wide escalation because no collection of isolated amplicons proves that distant segments reside on the same molecule.

Orthogonal confirmation closes gaps

Sequencing evidence should guide, rather than replace, targeted confirmation. Junction PCR followed by Sanger sequencing can confirm a specific boundary. Digital PCR can test copy-number hypotheses. Southern blotting or cytogenetic methods may still add value when an independent physical measurement is needed. The method should be chosen for the unresolved claim, not included by habit.

The strongest final statement is bounded: which portions are directly observed, which are supported by multiple methods, and which remain unresolved. "Full-length intact" should be reserved for an allele whose required sequence continuity has been demonstrated across the complete intended insert.

Match Methods to Model Systems

The same four questions apply across model systems, but sample structure and expected allele complexity change the design. A stable cell clone can be evaluated against its parental line. A transgenic founder may be mosaic. A BAC can be larger than the reads or captured fragments used to study it. An AAV-assisted knock-in can include donor and vector-derived sequences that were not part of the intended allele [1,3]. Plant and insect studies also show that paired-end WGS, targeted enrichment, dedicated integration callers, and long-range reconstruction can contribute complementary evidence when insert structure or local repeats complicate interpretation [5–8].

For projects that combine targeted editing with broader genome evaluation, the Genome Editing and Engineering Solutions page provides related service context. When the construct or integration process is AAV-specific, the AAV Sequencing service can address vector genome structure before or alongside host-genome integration analysis.

Build an Evidence-Complete Project

A sequencing platform should not be selected before the project team defines the claim it needs to support. Start with the expected integration mechanism, construct size, host genome, sample clonality, and whether the insertion locus is known. Then identify what would count as sufficient evidence for each endpoint.

Inputs worth preparing

The following materials reduce avoidable ambiguity:

• the exact vector or donor sequence in FASTA format, including backbone and helper elements that should be screened;
• the host species, strain, reference build, sex, and ploidy where relevant;
• a sample lineage table connecting parental material, edited pools, clones, founders, and later generations;
• positive, negative, parental, and wild-type controls appropriate to the model;
• prior PCR products, Sanger traces, copy-number results, and expected junction sequences;
• high-molecular-weight genomic DNA when long-range structure is a required endpoint;
• a written list of expected deliverables and criteria for calling a result complete or unresolved.

DNA integrity is especially important for long-read analysis. If molecules are shorter than the repeated or inserted structure, sequencing cannot physically bridge the region even when total yield is high. For fixed, degraded, or limited-input material, the project may need a staged strategy: junction discovery first, targeted confirmation second, and structural escalation only for samples that meet the input requirements.

Design errors that create ambiguity

Several problems are preventable at kickoff. An incomplete vector reference can misclassify backbone-derived reads, a missing parental control complicates variant interpretation, and pooled cells cannot support a clone-level allele map. One internal qPCR target can also make partial copies look intact.

The AAV integration analysis resource illustrates how integration-site questions differ from broader vector characterization. Even outside AAV studies, the useful lesson is to separate discovery, quantification, structural reconstruction, and confirmation rather than asking one assay to do all four.

Read the Final Data Package

The final report should allow an independent reviewer to trace every conclusion back to evidence. A list of genomic coordinates is not enough when the stated goal includes copy number, orientation, or full-length integrity.

Figure 4. An evidence-complete data package makes each conclusion traceable to read-level, quantitative, and orthogonal support.

Genome-wide findings should be reported with their detection limits. Researchers dealing with broader rearrangements can consult the WGS structural-variant detection guide. A useful closing statement should report the locus, copy-number basis, supported structure, insert continuity, and remaining uncertainty—not merely "integration confirmed."

FAQ

For research use only. Not for use in diagnostic procedures or individual treatment decisions.