Research Use Only (RUO): This article discusses experimental planning and bioinformatics interpretation for viral vector integration site analysis (ISA). It does not provide clinical recommendations and does not claim that sequencing alone demonstrates insertional risk.

Key takeaways

• AAV and lentiviral ISA answer different "default" questions because the biology differs: AAV is usually episomal (integration is low-signal), while lentiviral vectors are integration-dependent (integration is expected and often used for clonal tracking).
• Sample strategy is the real differentiator. What you collect (tissue vs cultured cells, enriched cell subsets, timepoints, matched negatives) often determines whether "no integration detected" is informative.
• Detection is technical; interpretation is biological. A vector–host junction call can be technically solid yet biologically ambiguous (e.g., single, low-support AAV junction in a tissue with high episomal background).
• Clonal interpretation is not symmetric. Recurrent sites and increasing abundance have clearer meaning for lentiviral-modified cell populations than for AAV-treated tissues, where repeated detection can also reflect assay artifacts, sampling, or local DNA break hotspots.
• Provider evaluation should focus on controls, evidence thresholds, and deliverables (raw data + reproducible pipeline + QC reporting), not on generic "sensitivity" claims.

1. Introduction: Why one integration site analysis design does not fit every vector

"Integration site analysis" often gets treated as a single service category: enrich vector–host junctions, sequence them, map to a reference genome, and report where the vector inserted. In practice, that mental model fits integrating vectors far better than it fits AAV.

AAV and lentiviral vectors can both appear in the same program portfolio, and both may trigger integration-related questions. But their default molecular states differ so dramatically that a one-size-fits-all ISA design can fail in two ways:

1. AAV programs can waste effort chasing biology that isn't expected (most vector DNA is episomal), while under-investing in controls needed to prevent false positives caused by episomal background.
2. Lentiviral programs can under-specify longitudinal and clonal requirements, producing a report that lists integration sites but cannot support the project's actual decision: "Is any clone expanding over time, and what does that mean in this experimental context?"

Good ISA planning starts by separating two layers:

• Technical detection problem: How do we recover and sequence authentic vector–host junctions from the sample we have?
• Biological interpretation problem: Once we have junctions, what do they imply about persistence, selection, clonal behavior, or genome context—and what do they not imply?

A useful comparison, therefore, should be framed around samples, sensitivity tradeoffs, and clonal interpretation, not around general vector pros/cons.

2. Biological differences that affect study design

Predominantly episomal AAV biology and possible integration events

Recombinant AAV (rAAV) is designed to deliver a transgene that typically persists as episomal DNA, often as circular monomers and concatemers, rather than as a provirus integrated into host chromosomes. A primate tissue study published in Journal of Virology reported rAAV vector genomes persisting predominantly as episomal circles in vivo (AAV vector genomes persist as episomal chromatin in primate muscle (2008)).

Integration can occur, but it is usually treated as a low-frequency, low-signal event relative to the episomal background. Mechanistically, AAV integration is generally thought to be linked to host DNA repair at pre-existing DNA breaks, and junctions often show features consistent with non-homologous end-joining or microhomology. Early large-scale junction analyses helped establish this experimental reality (Large-Scale Analysis of AAV Vector Integration (2005)).

The study-design implication is straightforward:

• If you put AAV-treated samples into a generic "retroviral ISA" workflow, the episomal background can dominate library molecules, and you can end up with either (a) poor recovery of true integration junctions or (b) a non-trivial false-positive burden from artifactual junctions.

Modern discussions emphasize that integration estimates and interpretation are tightly coupled to methodology and controls, and that "detectable integration" is not synonymous with a safety conclusion. A useful synthesis is provided by Molecular Therapy's integrated perspective on AAV integration (Evaluating the state of the science for AAV integration (2022)).

Integration-dependent lentiviral vector biology

Lentiviral vectors (LVs) deliver genetic payloads primarily through chromosomal integration. In many research and translational workflows—especially ex vivo modification of cells—stable expression is expected to depend on integration and cell division.

That difference changes the ISA goalpost:

• For lentiviral-modified cell products or cell populations, integration site analysis is not merely "where did it integrate?" It is often a clonal tracking tool: each unique integration event can function as a heritable marker for a cell lineage.
• Longitudinal sampling becomes central because the key interpretation questions are typically about clone persistence and expansion, not about whether integration exists at all.

Even with an integrating system, however, ISA still has interpretability limits: PCR bias, restriction site bias, and sampling bottlenecks can distort apparent clonal abundance. A methods-focused review in Blood emphasizes that technical choices can easily lead to false interpretation of clonal composition (The evolution of viral integration site analysis (2020)).

Figure 1. Side-by-side schematic of AAV, which predominantly persists as episomal DNA with rare chromosomal integration, and lentiviral vectors, which form integration-dependent proviruses, highlighting vector–host junction detection.

3. Research questions commonly asked for AAV projects

AAV ISA requests often originate from one of three realities: (1) the program needs to characterize rare integration, (2) prior data raised a question about persistence or hotspot behavior, or (3) a stakeholder needs a methodologically defensible "look" at integration under defined conditions.

Common AAV ISA research questions include:

1. Are there detectable AAV integration sites in this model and tissue under this dosing and timepoint design?

   • This is not a yes/no about biology; it's a question about detectability under a defined assay.

2. Do integration junctions show evidence of truncations, rearrangements, or partial cassettes that affect detection?

   • Especially relevant when ITR-anchored strategies are used.

3. Are there recurrent integration loci across animals or samples?

   • Interpreting recurrence requires careful consideration of genomic break hotspots and methodological artifacts.

4. What is the distribution of integration relative to genomic features (genes, CpG islands, repeats, etc.)?

   • Often framed descriptively rather than inferentially.

5. What controls and evidence thresholds are required to distinguish true junctions from episomal-derived artifacts?

6. Do different sample processing methods (fresh frozen vs preserved, extraction kits, fragmentation) alter the "background junction" landscape?

A practical takeaway: AAV ISA is frequently as much a method validation and artifact-control exercise as it is a biological discovery effort.

4. Research questions commonly asked for lentiviral projects

Because integration is expected for LVs, lentiviral ISA questions are usually framed around population structure, clonality, and genomic context.

Common lentiviral ISA research questions include (often motivated by how lentiviral integration sites behave as heritable clonal markers in expanding or persisting cell populations):

1. What is the integration site landscape (integrome) for this vector in this target cell type?
2. Are any integration sites recurrent across samples or enriched near specific genomic features?
3. Is the engineered cell population polyclonal or oligoclonal, and how stable is that profile over time or passages?
4. Are specific clones increasing in relative abundance across longitudinal samples?
5. How do vector design choices (e.g., enhancer/promoter configuration) and process choices (transduction conditions, culture duration) influence the detectable profile?
6. Do different lineages or sorted cell subsets show different clonal contributions?

Lentiviral ISA can also be used to validate that downstream assays (e.g., gene expression readouts) are not being driven by a single dominant clone—again, with the caveat that sequencing evidence alone doesn't establish biological causality.

5. Sample planning differences

Sample design is where AAV vs lentiviral ISA most sharply diverge. A strong ISA plan makes explicit (a) what sample types are in scope, (b) what each sample can and cannot answer, and (c) what controls are required for interpretability.

Vector material

AAV:

• Vector material is not an "ISA sample," but it can be critical context:

  • to understand ITR integrity, genome heterogeneity, and potential backbone/fragment contaminants that could later appear in junction-like reads.
  • to design capture probes or primers that tolerate truncation and rearrangements.

Lentiviral:

• Vector plasmid maps and the exact LTR and internal cassette sequences matter because junction enrichment often anchors on known vector sequence.
• For clonal analysis, you may also need a clear plan for how vector copy number (VCN) measurements will relate to ISA interpretation (e.g., expected multiplicity of integrations per cell).

Cultured cells

AAV:

• Cultured cells are useful when you want controlled conditions and high DNA availability, but beware that dividing cells can dilute episomes and change the episome/integration ratio over time.
• If you're comparing promoters, capsids, or doses, cultured models can help isolate variables—but results may not translate directly to complex tissues.

Lentiviral:

• Cultured cells are often the primary substrate for LV ISA.

• Key design questions include whether you need:

  • bulk cell pellets (population-level ISA)
  • sorted subsets (lineage-specific clonal tracking)
  • single-cell approaches (only when the question truly requires cell-level linkage)

Tissue samples

AAV:

• Tissue is common for in vivo AAV programs and is where episomal background can be highest.
• Tissue heterogeneity (cell types, proliferation state) matters because integration may be more detectable in some compartments than others.
• Practical planning: consider whether to enrich or sort relevant cell populations (when feasible) to avoid averaging signals across unrelated cell types.

Lentiviral:

• For in vivo LV studies, tissues can be sampled, but interpretability depends on whether the biological unit you care about is a mixed population or a defined cell subset.
• When clonal tracking is the goal, sample purity and consistent processing across timepoints can matter as much as depth.

Longitudinal samples

AAV:

• Longitudinal sampling is often about persistence and detection stability rather than classic clonal expansion.
• Timepoints should be chosen to reflect when episomal forms stabilize and when integration junctions—if present—would be most recoverable given assay constraints.
• Interpret "appearance/disappearance" cautiously: it may reflect sampling or assay stochasticity rather than biology.

Lentiviral:

• Longitudinal sampling is a central design axis.

• The question is typically not "does integration exist?" but "does the clonal composition change?"

• Decide upfront whether the study is:

  • surveillance-style (broad timepoints, coarse resolution) or
  • mechanism-style (more frequent sampling, subset sorting, possibly orthogonal assays).

Reference or control samples

Controls should be planned as first-class samples, not add-ons.

AAV controls often need to address episomal artifacts:

• negative controls that match tissue type and handling (to define background chimeras)
• process controls that detect cross-contamination and index hopping
• controls that help distinguish vector–vector concatemers from vector–host junctions

Lentiviral controls often need to address quantitative and clonal bias:

• spike-ins or reference materials (where available) to check whether the pipeline can recover expected sites
• technical replicates to estimate reproducibility of clone-abundance estimates
• controls for restriction enzyme or PCR bias (e.g., alternative fragmentation strategy) depending on method

6. Detection strategy considerations

ISA is not one method. It's a family of workflows that differ mainly in how they enrich vector–host junctions and how they trade sensitivity against bias and input requirements.

A helpful high-level taxonomy of integration site identification methods and their bias sources is provided in a recent methods landscape review (Methodological landscape in integration site identification (2025)).

Vector-host junction enrichment

At the center of any ISA design is junction enrichment: increasing the fraction of sequencing molecules that contain a true vector-host junction.

• For AAV, enrichment must succeed despite abundant episomal molecules that may contain vector-only fragments or concatemers.
• For lentiviral vectors, enrichment is typically more straightforward because integration junctions are expected to be prevalent in genomic DNA from modified cells, but bias still enters through fragmentation and amplification.

Targeted amplification

Targeted amplification strategies (e.g., LM-PCR, LAM-PCR family methods) use primers in vector sequence plus adaptors/linkers.

Key considerations:

• Restriction enzyme dependence can bias recovery toward integration sites near cut sites. Protocol discussions around LAM-PCR emphasize this retrieval bias conceptually (JoVE LAM-PCR protocol paper (2014)).
• Restriction-enzyme-free or "non-restrictive" approaches aim to reduce that bias, but can introduce other efficiency tradeoffs (restriction enzyme–free LAM-PCR approaches (2012)).

AAV-specific caution: if your primer anchoring assumes intact ITRs or intact vector ends, truncations and rearrangements can create blind spots.

Capture-based methods

Capture-based approaches (target enrichment using probes) can be appealing when:

• you want to reduce reliance on particular restriction sites
• you want to recover junctions even when vector ends are imperfect
• you need a workflow that better tolerates structural heterogeneity

But capture approaches have their own tradeoffs:

• higher DNA input requirements in some designs
• variable capture efficiency
• the need to design probes that match what is actually present in the sample (including rearrangements)

Whole-genome approaches

Whole-genome sequencing (WGS) is often described as "unbiased," but for ISA it may become a depth problem: the vast majority of reads are host genome, and junction molecules can be extremely sparse—especially for AAV.

This is why many programs evaluate WGS as one component in a toolbox rather than as a default. Comparative studies highlight tradeoffs between targeted enrichment approaches and whole-genome approaches for viral integration detection (a 2023 comparison of targeted enrichment vs WGS (Molecular Therapy Methods)).

Figure 2. Workflow comparison of AAV and lentiviral integration site analysis, from sample collection through junction enrichment, sequencing, mapping, quality control, and clonal interpretation.

7. Sensitivity and low-frequency integration events

ISA sensitivity is not a single number. It's the product of:

• biological prevalence (how many cells truly contain integrations)
• DNA input (how many genomes are actually interrogated)
• junction recovery efficiency (enrichment method, fragmentation, ligation)
• sequencing depth and library complexity
• bioinformatics filtering (how aggressively you remove artifacts)

This is why fixed sensitivity claims are usually misleading unless the provider specifies:

• sample type and input amount
• vector and target cell/tissue
• enrichment method and read length
• evidence thresholds for calling junctions
• whether support counts reflect unique molecules or PCR-amplified duplicates

AAV: low-frequency integration is a design challenge, not a "deeper sequencing" problem

If episomal AAV genomes outnumber integrated junction molecules by orders of magnitude, "just sequence deeper" often increases costs faster than it increases confidence, because you may be sequencing more episomal-derived molecules and more noise.

In this context, the design lever is usually enrichment + controls, not depth alone. The Molecular Therapy integrated perspective discussed earlier emphasizes methodological challenges and careful interpretation for AAV integration.

Lentiviral: low-frequency clones are still meaningful, but abundance estimates need calibration

For lentiviral-modified cell populations, ISA is often used to detect and track clones across time. A low-abundance clone may matter for some questions (e.g., population diversity), but abundance estimates can be distorted by PCR and sampling.

Practically, "sensitivity" for lentiviral ISA should be discussed as:

• "what fraction of unique sites do we recover given this input?"
• "how stable are abundance rankings across technical replicates?"
• "what evidence do we provide that abundance reflects biology rather than amplification?"

8. Supporting-read and mapping evidence

To evaluate an ISA report (or to write acceptance criteria for a CRO), it helps to define what counts as sufficient technical evidence for an integration site call.

Most pipelines rely on a combination of:

• split/soft-clipped reads that span the junction breakpoint
• discordant read pairs where one mate maps to vector and the other to host
• mapping quality, repeat-region handling, and duplicate removal

False positives are common in low-complexity or repeat regions, and integration callers vary substantially in how they handle these cases. A study describing BATVI highlights that repeat regions can drive many false integration predictions if not handled carefully (BATVI paper (2017)).

Conversely, methods designed for high-precision breakpoint detection emphasize strategies to keep false discovery low even in repeats, as described for VIRUSBreakend (VIRUSBreakend (Bioinformatics, 2021)).

Practical evidence checklist for provider reports

When assessing a provider or internal pipeline, ask whether the report includes:

• explicit evidence definitions (what counts as a junction)
• read-level support metrics (and whether duplicates are removed)
• mapping quality thresholds and rationale
• handling of multi-mapping reads and repeats
• representative IGV-style screenshots or equivalent evidence for top sites
• clear criteria for "low-confidence" vs "high-confidence" calls

This matters most for AAV, where the false positive cost is high, but it also matters for lentiviral studies where apparent "dominant clones" can be amplification artifacts.

9. Differences in clonal interpretation

Clonality is where teams most often over-interpret ISA. The core risk is confusing technical recurrence (the same junction being detected multiple times) with biological dominance (a clone expanding).

Repeated integration sites

Lentiviral: repeated detection of the same integration site across replicates or timepoints is often consistent with a persistent clone (especially in ex vivo modified cell populations), but you still need to control for amplification bias.

AAV: repeated detection can be harder to interpret:

• It may reflect a real, rare integrant sampled multiple times.
• It may reflect local genomic break hotspots.
• Or it may reflect workflow artifacts when episomal background generates chimeric molecules.

This is why AAV ISA projects often need stricter process controls and conservative interpretation language.

Clonal abundance

Most ISA workflows produce relative abundance proxies (read counts, molecule counts, or normalized measures). But abundance is not a direct measurement of cell counts unless you have:

• a way to control PCR duplicate inflation
• sufficient library complexity
• consistent DNA input across samples
• an explicit model of how counts map to genomes/cells

Even in integrating systems, abundance can vary due to technical effects. A methods-focused review in Blood discusses how ISA technical choices can lead to false interpretation of clonal composition.

PCR amplification bias

PCR bias can distort:

• which sites are recovered at all
• the apparent dominance ordering of clones
• the stability of abundance estimates across runs

Methods differ in where they place amplification steps and how much they rely on PCR. Method selection should therefore be aligned to your core decision:

• If you primarily need presence/absence of junctions, high-sensitivity PCR-based approaches may be appropriate.
• If you need quantitative clonal ranking, you may need more emphasis on bias reduction, replicates, and orthogonal checks.

A recent methods landscape review provides a useful overview of where these biases arise across methods.

Longitudinal changes

Lentiviral: Increasing abundance of a specific site across timepoints is often interpreted as a clonal expansion signal. Whether that expansion is expected (e.g., selection under culture conditions) or concerning (e.g., unexpected dominance) depends on context, vector design, cell type, and additional orthogonal data.

AAV: Changes across timepoints can be dominated by:

• shifting episomal copy numbers
• tissue sampling variability
• library complexity differences

Therefore, AAV longitudinal ISA is often better framed as:

• "Do we repeatedly detect junctions with consistent evidence under consistent processing?"
• "Do junctions concentrate in any loci across animals?"

…rather than as direct clone tracking.

Figure 3. Interpretation concept comparing isolated low-frequency junction calls with recurrent or expanding clonal signals across longitudinal samples.

10. AAV vs lentiviral integration site analysis comparison table

Dimension	AAV integration site analysis	Lentiviral integration site analysis
Default molecular state	Predominantly episomal; integration is low-signal	Integration-dependent; integration is expected
Primary planning risk	Episomal background → artifacts or poor junction recovery	Abundance distortion from bias; sampling bottlenecks
Most common core objective	Detect/characterize rare AAV integration sites under defined conditions	Map integrome + track clonal composition over time
"No integration detected" means…	Often "not detected under this assay/input/controls"	More informative for method failure or sample issues if integration is expected
Sample priorities	Tissue context + matched negatives; consider cell-type enrichment; careful handling	Defined cell populations; subset sorting for lineage questions; consistent longitudinal collection
Enrichment strategy emphasis	Strong emphasis on junction enrichment that minimizes episomal interference	Emphasis on reproducible recovery and quantitative consistency
Whole-genome approaches	Often limited by rarity of junction molecules; may require targeted enrichment support	Can be used but often less efficient than junction-targeted approaches
Evidence thresholds	Typically stricter to avoid false positives	Must balance false positives vs missing low-abundance clones
Clonal interpretation	Conservative; repeated sites may not equal expanding clone	Central; repeated sites + increasing abundance more interpretable (with bias controls)
Key deliverable expectation	Junction list with evidence + artifact controls + limitations	Junction list + clonal abundance metrics + longitudinal tracking views + QC

11. Controls and quality checkpoints

A provider-ready ISA design should define controls at three layers: molecular biology, sequencing, and bioinformatics.

Molecular biology controls

• Negative controls processed alongside study samples (same matrix if possible)
• No-template controls to detect reagent contamination
• Process controls that detect cross-sample contamination

AAV-specific emphasis:

• controls that help quantify and limit episomal-derived artifacts (define what "background junction-like reads" look like)

Lentiviral-specific emphasis:

• controls or replicates that reveal PCR/restriction bias and abundance instability

Sequencing QC checkpoints

• library complexity/duplication metrics
• index cross-talk checks (especially when multiplexing many samples)
• consistent read length and depth planning aligned to the enrichment method

Bioinformatics QC checkpoints

• explicit filtering steps for duplicates and low mapping quality
• repeat-region handling strategy
• "high-confidence vs low-confidence" call classification

For repeat-driven false positives, BATVI provides a cautionary example of how integration calling can be confounded without careful handling.

12. How to choose a study design based on the research question

A practical way to choose an ISA design is to start with the question and work backwards to the minimal evidence needed.

Step 1: Classify your primary question

Pick the closest category:

1. Detection question: "Do we detect any integration events?"
2. Distribution question: "Where in the genome are integrations occurring?"
3. Quantification question: "How frequent are integrations / how does abundance compare?"
4. Clonal dynamics question: "Are specific clones persistent or expanding longitudinally?"
5. Structure question: "Are integrations truncated/rearranged/complex?"

Step 2: Map question → preferred sample + method emphasis

• If you need clonal dynamics, prioritize defined cell populations, longitudinal timepoints, technical replicates, and bias-aware abundance reporting.
• If you need rare AAV event detection, prioritize junction enrichment design + matched negatives; accept that interpretability depends on assay constraints.
• If you need structure, consider approaches that can resolve complex molecules (e.g., long-read support or methods that tolerate rearrangements).

Step 3: Define acceptance criteria in advance

Examples of decision-grade acceptance criteria:

• Minimum evidence thresholds for calling a junction
• Required controls to quantify background
• Required deliverables (raw reads, mapping summary, QC, and a reproducible pipeline description)

13. Information to prepare before project assessment

To make provider assessment efficient—and to prevent a study that "runs" but can't answer the question—prepare:

Vector and construct information

• Vector type (AAV vs LV), serotype/pseudotype, and a full sequence map of the delivered genome
• Known features that affect enrichment (ITRs, LTRs, internal repeats)

Sample inventory

• matrices (cells/tissue types), preservation method, expected DNA yield
• number of samples and timepoints
• any planned cell enrichment/sorting

Desired outputs

• integration site list format (coordinates, genome build)
• evidence fields needed (supporting reads, mapping quality, flags)
• clonality outputs (relative abundance metrics, longitudinal plots)

Controls and constraints

• what negative controls are available and how they match sample handling
• multiplexing plan constraints and contamination concerns

Bioinformatics transparency requirements

• availability of FASTQ
• pipeline description (tools, reference genome, filtering logic)
• reproducibility expectations

14. Interpretation limitations

A decision-grade ISA report should include limitations in plain scientific language.

Key limitations to state explicitly:

• Detection ≠ biological consequence. A technically credible vector–host junction does not, by itself, establish functional impact.
• Non-detection is conditional. "No integration detected" means "not detected under this assay, input, and filtering regime."
• Bias is unavoidable; it must be managed. Restriction site bias, PCR bias, and mapping ambiguity can shape both detection and abundance estimates.
• Repeat regions are a persistent confounder. Integration calling in low-mappability regions can inflate false positives without careful handling.
• AAV-specific: episomal background and rearrangements complicate inference. Artifacts and structural heterogeneity can change what is detectable.

For AAV integration specifically, state-of-science discussions emphasize methodological challenges and conservative interpretation.

15. Conclusion

AAV vs lentiviral integration site analysis is less about which vector is "better" and more about matching biology → samples → enrichment strategy → evidence thresholds → interpretation rules.

• In AAV projects, integration detection is typically a low-signal problem dominated by episomal background and structural heterogeneity, so success depends on artifact-aware design and conservative interpretation.
• In lentiviral projects, integration is expected and ISA is often used to understand clonal structure and dynamics, so success depends on longitudinal sampling, bias-aware abundance reporting, and clear QC gates.

If you're evaluating an ISA provider, prioritize transparency and decision utility: controls, evidence thresholds, reproducibility, and deliverables—over generic sensitivity claims.

Next step (RUO): If you want a feasibility assessment, prepare your vector map, sample list (matrices + timepoints), and the exact decisions you want the ISA data to support. You can then compare workflows (targeted amplification vs capture vs whole-genome + long-read support) against those decisions.

Learn more about CD Genomics' related RUO services: AAV Integration Site Analysis, Lentiviral/Retroviral Integration Site Sequencing and Integration Site Analysis (ISA) Service.

16. FAQ

1) Do AAV projects always need integration site analysis?

Not always. Many AAV studies focus on vector genome integrity, expression, or biodistribution rather than junction mapping. ISA is most helpful when your question involves detecting or characterizing rare AAV integration sites under defined conditions, comparing designs or process changes, or building an evidence-backed narrative around what is and isn't detectable with your assay.

2) Why can't I use the same ISA workflow for AAV and lentiviral vectors?

Because the biological "starting state" differs. For lentiviral vectors, integration is expected and junction molecules are a natural part of genomic DNA from modified cells. For AAV, episomal vector DNA can be abundant and can overwhelm the assay or generate artifacts unless enrichment and controls are designed specifically to manage that background.

3) What does "vector–host junction" evidence typically look like?

Most pipelines look for reads that span the breakpoint (split/soft-clipped reads) plus supporting read pairs where one mate maps to vector and the other to host. High mapping quality, duplicate handling, and repeat-region filtering are key to reducing false positives, especially when the signal is rare.

4) Can I interpret read counts as the frequency of integration?

Only cautiously. Read counts are shaped by DNA input, junction recovery efficiency, PCR amplification, and filtering. For clonal integration analysis, abundance proxies are most useful when you have replicates and clearly defined rules for deduplication and normalization, and when the method's biases are acknowledged.

5) How should I interpret repeated detection of the same integration site?

In lentiviral-modified cell populations, repeated detection across timepoints often supports persistence of a clone, and increasing abundance can suggest clonal expansion—subject to bias controls. In AAV studies, repeated detection can also reflect artifacts or hotspot behavior, so recurrence should be interpreted conservatively and in the context of matched controls and evidence strength.

6) When should I consider capture-based methods instead of PCR-based enrichment?

Capture-based approaches can be useful when you want to reduce reliance on restriction sites or when vector ends may be truncated or rearranged, which can break primer anchoring. The tradeoff is typically higher input requirements and variable capture efficiency, so the choice should be driven by your sample constraints and the need for quantitative stability.

7) Does WGS "solve" bias in integration site analysis?

WGS can reduce certain enrichment biases, but it introduces a different limitation: junction molecules may be extremely rare relative to total host DNA, especially for AAV. Many programs use WGS as a complement to targeted enrichment rather than as a default, because depth requirements can become impractical without a strong biological signal.

8) What should I ask an ISA provider to include in the final report?

Ask for: (1) a junction list with genome build and coordinates, (2) evidence fields (supporting reads, mapping quality, deduplication approach), (3) QC metrics (library complexity, duplication, contamination checks), (4) clear criteria for high- vs low-confidence calls, and (5) a reproducible bioinformatics summary plus access to raw data if your governance model requires it.