Can NGS Replace Conventional Viral Safety Testing for Cell and Gene Therapy Products?

Meta Intent: A practical, risk-based guide to determining which conventional viral safety assays next-generation sequencing (NGS) may replace, supplement, or leave unchanged across cell and gene therapy manufacturing workflows.

Next-generation sequencing (NGS), also called high-throughput sequencing (HTS), detects viral nucleic acids across a broad sequence space without requiring a separate assay for every candidate virus. In CGT manufacturing, it can support risk-based investigation of cell banks, viral seeds, raw materials, and harvests—but only when the workflow is validated for a defined matrix and decision.

The question is not whether NGS is "better" than every conventional assay. The useful question is narrower: what exact evidence does a legacy assay provide, and can a validated NGS workflow provide evidence that is fit for the same intended use? A broad nucleic-acid screen, an infectivity assay, a replication-competent virus assay, and a viral-clearance study do not answer the same scientific question. Treating them as interchangeable is the fastest route to a weak replacement rationale.

For many adventitious-virus applications, NGS can now move beyond a purely supplemental role. It has the strongest case for replacing broad in vivo assays and consolidating selected virus-specific molecular assays. It may also supplement, or in some circumstances replace, in vitro cell-culture assays when the end-to-end workflow is validated in the relevant matrix. It does not, however, automatically establish that a detected sequence came from an infectious virus, prove that a vector cannot generate replication-competent virus, or demonstrate that a manufacturing process clears or inactivates virus.

This guide provides a replacement map for making that distinction before a method-development program begins.

The short answer: NGS can replace selected assays, not the whole viral safety strategy

A conventional viral safety program is a layered argument. It begins with prevention: qualified raw materials, controlled manufacturing conditions, and characterized cell substrates. It then adds testing at appropriate points, together with evidence that the process can remove or inactivate relevant viruses when such clearance is feasible. No single negative result carries the entire argument.

NGS changes the detection layer. Non-targeted workflows can search broadly for known, divergent, and unexpected viral sequences, including signals outside the original risk list. That is a meaningful advantage over a fixed set of primers, permissive cell lines, or animal species. In a program that needs broad screening of cell banks, virus seeds, or harvest material, a well-designed Viral Metagenomic Sequencing strategy can therefore be more informative than adding further narrow assays one by one.

But NGS measures sequence evidence. A conventional cell-culture assay relies on virus replication and biological activity in a selected indicator system. A clearance study tests a scaled-down process step. An RCR, RCL, or RCA assay is designed around a specific replication risk associated with a vector system. Replacing one of those assays requires more than showing that NGS generates many reads or detects a control spike at one concentration.

The appropriate decision has three possible outcomes:

• Replace when the existing test and the NGS assay address the same intended-use question, and replacement-grade validation supports that claim.
• Supplement when NGS contributes useful breadth but cannot independently answer a biological or functional endpoint.
• Retain when the conventional assay measures infectivity, replication competence, or process clearance that sequence detection does not establish.

Figure 1: The CGT Viral Safety Assurance Stack. A layered graphic showing prevention controls, broad adventitious-virus detection, infectivity assessment, replication-competent virus assessment, viral-clearance evidence, and lot disposition. NGS is highlighted across the broad-detection layer rather than across every layer.

Start by separating the four evidence questions

The phrase "viral safety testing" is convenient, but it conceals four different evidence questions. A replacement plan should name the question first and choose the assay second.

Are viral nucleic acid sequences present?

This is the natural home of NGS. Non-targeted analysis can search for nucleic acids from a broad range of known and unexpected viruses, while targeted analysis can increase attention to a defined risk set. The output is a sequence signal with a specified limit of detection, breadth of coverage, and reporting rule. In this setting, the relevant controls include extraction controls, negative process controls, reference virus material, library controls, and a defined bioinformatics pipeline.

The practical distinction is important. A Viral Genome Sequencing project may focus on characterizing a known virus or vector genome. Adventitious-virus NGS instead begins with the possibility that the relevant virus is unknown, divergent, or present in a difficult background. The workflows can share sequencing components, but their validation claims should not be borrowed from one another.

Is the signal associated with an infectious virus?

NGS can show that viral nucleic acid is present. It cannot, by sequence detection alone, demonstrate whether a virus is replication competent, whether intact infectious particles are present, or whether the signal is a noninfectious remnant. This is especially relevant for endogenous retroviral sequences, vector-related material, helper-virus sequences, residual nucleic acid, and low-level reagent background.

A positive result should therefore trigger an investigation rather than an immediate conclusion about product contamination. Depending on the signal and the risk assessment, follow-up may include repeat extraction, targeted PCR or digital PCR, targeted sequencing, culture on a relevant permissive system, or an assay designed to test infectivity. The purpose is to move from a sequence observation to a biologically interpretable conclusion.

Can the vector generate replication-competent virus?

Replication-competent retrovirus (RCR), replication-competent lentivirus (RCL), and replication-competent adenovirus (RCA) are not simply broad adventitious-agent signals. They are vector-system-specific risks that may emerge through recombination, residual helper components, or other manufacturing events. The critical evidence is tied to the ability to replicate under appropriate conditions, not merely the presence or absence of a sequence match.

For this reason, a negative broad NGS result should not be presented as an automatic replacement for an RCR, RCL, or RCA testing strategy. NGS can strengthen a wider investigation and may help characterize unexpected sequence findings. The replication-competent-virus decision, however, needs its own intended use, controls, and supporting evidence.

Can the process remove or inactivate virus?

Viral clearance is a process-capability question. It asks whether a manufacturing step reduces a relevant or model virus under defined conditions, often using a scaled-down representation of the process. A negative sequencing result in a purified sample does not establish the reduction factor of a purification or inactivation step. Conversely, a strong clearance package does not remove the need to manage contamination prevention and testing upstream.

Figure 2: Four Evidence Questions, Four Different Endpoints. A four-panel diagram contrasting sequence detection, infectivity, replication competence, and viral clearance. Each panel shows its endpoint, suitable evidence type, and the boundary of what a negative NGS result can support.

Where replacement is strongest—and where it stops

The table below is not a universal regulatory template. It is a planning device for deciding what must be justified in a product-specific replacement package.

For CGT, this is a scientific planning framework rather than a direct statement of release-test requirements. ICH Q5A(R2) excludes cell therapy products from its formal scope. Its NGS principles can inform a product-specific risk assessment and validation strategy, but they should not be transferred unchanged to a CGT program.

The most defensible claim is usually specific: "NGS is proposed to replace the in vivo broad adventitious-virus assay at the indicated manufacturing stage." That claim is far stronger than "NGS replaces viral safety testing." It tells reviewers which evidence is changing, what remains in place, and how the new method will be evaluated.

Figure 3: Replace, Supplement, or Retain. A decision matrix with seven assay objectives arranged along a horizontal replacement continuum. The design uses three evidence zones—replace, supplement, and retain—rather than a simple yes/no comparison.

Define intended use before designing the wet-lab workflow

An NGS method cannot be validated in the abstract. The intended use determines the sample, preprocessing, reference materials, performance criteria, and follow-up logic.

First, specify the product modality. An AAV preparation, a lentiviral vector, an adenoviral vector, and an ex vivo modified cell product produce very different sequence backgrounds. A high-titer vector can consume reads, while host-cell nucleic acids can obscure low-abundance viruses in an unprocessed harvest. A method that performs well on purified vector material is not automatically suitable for a cell-rich or protein-rich harvest.

Second, specify the manufacturing stage. A master cell bank, working cell bank, virus seed, unprocessed bulk harvest, purified bulk, and final product do not share the same matrix or risk profile. Cell-bank testing may prioritize endogenous and adventitious viral sequences associated with the substrate and raw-material history. A harvest assay may need to tolerate high host-cell background. A purified viral-vector sample may need to distinguish the intended vector from possible contaminating viruses with related sequences.

Third, specify the decision that the result supports. Is the method being used for cell-substrate characterization, routine lot testing, process-change comparability, an investigation of an unexpected signal, or replacement of a legacy assay? If the decision is not written before validation begins, the acceptance criteria often become a list of convenient measurements rather than evidence for a defined use.

Cell identity is an upstream but separate control. A Cell Line Identification assessment can help confirm that a production-cell substrate has not been substituted or mixed up, but it does not replace a viral adventitious-agent test. Keeping those claims separate makes the overall control strategy easier to defend.

Figure 4: Product-and-Stage Sampling Map. A manufacturing map with parallel AAV, lentiviral-vector, and cell-product lanes. Each lane marks cell bank, seed, harvest, purified material, and final product, with different matrix risks shown as concise callouts.

Build an end-to-end NGS workflow, not a sequencing-only assay

Sequencing is only one module. The performance of the complete workflow depends on what happens before and after the instrument run.

Choose the analyte deliberately

A broad workflow may need to capture both DNA and RNA viruses. The choice between total nucleic acid, separate DNA and RNA fractions, transcript-focused analysis, and particle-associated nucleic acid analysis should be justified by the sample and risk question. A workflow that is optimized for RNA can underperform for certain DNA-virus inputs, while a DNA-focused workflow may miss RNA-virus evidence without a complementary preparation step.

Where RNA-bearing signals are part of the intended coverage, an RNA-Seq design can be relevant to the library strategy. The service link is not a promise that a generic transcriptome workflow is already validated as a viral-safety assay. Instead, it highlights the technical connection: RNA extraction, reverse transcription, library construction, and sequence-quality controls require matrix-specific development when the goal is broad virus detection.

Treat enrichment as a trade-off, not a free sensitivity gain

Filtration, nuclease treatment, particle concentration, host depletion, ribosomal-RNA depletion, and targeted enrichment can improve the signal-to-background ratio. Each can also introduce bias. Nuclease treatment may reduce free nucleic acid while favoring protected genomes. A concentration step may recover some viral particles better than others. Probe-based enrichment can increase sensitivity for a known family yet weaken the claim of agnostic detection.

The correct question is not "Which preprocessing step gives the most reads?" It is "Which preprocessing strategy preserves the viruses the assay is expected to detect in this sample matrix?" A reference panel should include viruses with different genome types, particle structures, envelope properties, and resistance profiles so that the workflow is not tuned to one easy-to-detect model.

Make bioinformatics part of the validated method

An NGS assay has no single endpoint until the analysis pipeline turns reads into a reportable signal. Host subtraction, quality trimming, de novo assembly, reference alignment, taxonomic assignment, and thresholds for coverage or sequence identity all influence the final result. The viral database, software version, read-filtering rules, and positive-call criteria should therefore be documented and controlled as analytical-method components.

For confirmation after a reportable signal, Targeted Region Sequencing or Amplicon Sequencing Services can support focused follow-up of a defined sequence region. These are confirmation tools within a broader investigation, not a substitute for the breadth of the original non-targeted screen.

Figure 5: End-to-End NGS Control Architecture. A semi-isometric laboratory-to-data pipeline: sample receipt, preprocessing, extraction, library preparation, sequencing, host subtraction, viral analysis, and report review. Each module has a compact QC gate and control icon; the final branch leads either to negative release evidence or positive-signal investigation.

What replacement-grade validation needs to demonstrate

Replacement validation should mirror the intended use, not a generic sequencing checklist. The most informative package is end-to-end: the controls enter before the workflow steps that could lose or distort viral material.

Sensitivity and limit of detection

The limit of detection should be defined using materials that resemble the viral diversity the assay claims to cover. A single well-behaved DNA virus cannot stand in for small single-stranded DNA viruses, double-stranded RNA viruses, enveloped RNA viruses, large DNA viruses, or poorly extracted viral particles. Spike material should enter before extraction or enrichment whenever the goal is to test the whole method.

The reported limit also needs a unit and matrix context. Genome copies per milliliter, copies per extraction, and copies relative to host or vector background are not interchangeable. Low-level detection is affected by sampled volume and random distribution of particles. A method may show strong sensitivity in a clean preparation and a weaker, yet still acceptable, sensitivity in an unprocessed harvest. That is not a failure if the intended-use claim is written honestly and the method is validated for the relevant stage.

Breadth and specificity

Breadth is the ability to recover diverse viral signals; specificity is the ability to avoid calling irrelevant, contaminating, or misassigned sequences as reportable viruses. Both depend on sample handling and computational interpretation. A broad database is necessary, but it also increases the chance of ambiguous matches. Predefined rules for read length, alignment quality, genome distribution, negative-control context, and orthogonal follow-up are therefore as important as the read count itself.

Matrix verification and robustness

Matrix verification is essential when an NGS method changes from a cell bank to a harvest, from one vector platform to another, or from a low-host to high-host background. It should challenge the method with the intended sample characteristics, including vector abundance, host-cell nucleic acids, media components, and possible interfering materials. Robustness studies should also examine operator, run, lot, and time-related variation.

Bioinformatics lifecycle control

Viral databases evolve. Software is patched. Filtering and taxonomy rules improve. Those changes can strengthen an assay, but they can also alter the result for the same raw data. A replacement-grade method needs defined change-control triggers, archived pipeline versions, documented database releases, and a re-evaluation policy. Without that lifecycle plan, a method can slowly change without a clear link to its original validation.

Figure 6: Validation Evidence Wheel. A circular technical graphic centered on the intended-use statement, with six linked evidence segments: representative reference viruses, sensitivity, breadth, specificity, matrix verification, and pipeline lifecycle control. Small outer callouts show sample volume, negative controls, and predefined positive-call rules.

Positive NGS signals need a triage plan before routine use

The most costly uncertainty is not a clean negative run. It is a low-level unexpected signal that appears in a product sample, a control, or both. The investigation should be planned before routine testing starts.

Begin with the analytical context. Was the signal also present in an extraction blank, library blank, environmental control, or another sample in the same run? Is the read distribution confined to a short region, or does it span multiple regions of a viral genome? Does the signal align better with the intended vector, a helper-virus element, an endogenous sequence, or a known reagent contaminant? A single read may be meaningful in a well-controlled system, but only in the context of predefined criteria and controls.

Next, repeat the observation independently where possible. A new extraction from retained material, a repeat library, or an orthogonal targeted test can help distinguish a sporadic technical event from a reproducible sequence signal. If the identified virus or related virus is relevant to the risk assessment, the investigation should proceed to biological interpretation. This may require an infectivity approach, additional characterization, manufacturing-environment review, or a focused assessment of raw materials and cell banks.

The important operational principle is that a positive NGS result is not automatically a product-failure result, but it is never merely a bioinformatics inconvenience. It should move through a documented escalation path with defined technical, biological, and quality decisions.

Figure 7: Positive-Signal Triage. A branching decision graphic that begins with a reportable sequence hit and separates control-associated noise, vector-related sequence, endogenous sequence, and plausible adventitious-virus signal. The final branch maps to repeat extraction, targeted confirmation, infectivity follow-up, and manufacturing investigation.

CGT-specific replacement decisions: AAV, lentiviral vectors, and cell products

AAV manufacturing

AAV processes illustrate why high-titer intended virus can complicate broad detection. The vector itself may dominate the sequence library, while residual host-cell nucleic acids and helper-virus-related material create additional analytical background. The NGS workflow must show that its preprocessing and analysis rules preserve sensitivity for an external low-abundance virus in that background.

Vector characterization also needs its own boundary. AAV Genome Sequencing is relevant for evaluating vector genome features such as integrity or heterogeneity. It should not be described as an adventitious-virus screen unless the workflow, database, controls, and intended use have been developed for that broader purpose. Similarly, AAV integration-site analysis addresses a different genomic question from contaminant detection.

For AAV production, the replacement discussion should identify whether NGS is being used at the virus seed, unprocessed harvest, purified bulk, or another defined stage. The same platform may be useful across stages, but the validation package cannot assume that a low-complexity purified sample represents a high-background harvest.

Lentiviral and retroviral vector workflows

Broad NGS can be valuable for identifying unexpected viral sequences in production cells, raw materials, and harvests. Yet lentiviral and retroviral systems also require dedicated consideration of replication competence. An adventitious-virus NGS workflow and an RCL/RCR strategy may share samples or supporting information, but they do not have identical endpoints.

Lentiviral/Retroviral Integration Sites Analysis can provide information about where vector sequences integrate in a research sample. It is useful to mention because it is often discussed alongside vector safety, but it should not be used as a proxy for adventitious-virus testing or replication-competent-virus detection. The safest content architecture keeps those analytical claims separate and explains how each fits into the broader product-characterization plan.

Cell-based products

Cell-based products introduce a different practical constraint: the product matrix may be cell-rich, limited in volume, and sensitive to lengthy testing timelines. Risk assessment must consider donor-derived or raw-material-associated exposure, culture duration, cell-bank history where applicable, and the feasibility of sampling enough material to make a negative result meaningful.

For this category, it is particularly important not to overextend guidance written for viral vectors or cell-line-derived biologics. The useful approach is to apply the same analytical questions—what is the sample, what virus classes must be detectable, what sequence background is expected, and what decision will the result support—then build a product-specific validation plan. NGS may provide a strong broad-detection component, but its place in the final safety package remains risk based.

Move from supplementation to replacement through a controlled transition

The most efficient transition is staged. Start with a written gap assessment for the legacy assay: sample type, risk coverage, endpoint, sensitivity expectation, major limitations, and associated decision. Then write the proposed NGS intended use in one sentence. If that sentence is vague, the replacement program is not ready.

Next, develop and qualify the end-to-end workflow in the target matrix. Do not restrict the study to instrument performance or a sequencing vendor's general specifications. The meaningful data are recovery across relevant virus types, sensitivity in the relevant background, negative-control behavior, signal-triage rules, and reproducibility of the complete process.

A bridging strategy should be justified rather than copied mechanically. In some cases, a direct one-to-one comparison with an older biological assay does not answer the same question because the methods have different breadth and endpoints. What matters is whether the data establish that the NGS workflow is suitable for the intended use and whether remaining risks are managed by other layers of the safety program.

Finally, plan the operational lifecycle. Specify how reference materials will be used, how the viral database and pipeline will be versioned, what changes trigger requalification or validation, how retained samples will be handled, and who owns the escalation decision for unexpected hits. That makes the method usable beyond a proof-of-concept study.

When NGS is not ready to replace the existing test

NGS should remain supplementary when the intended use is undefined, the test matrix has not been verified, the validation panel is too narrow, or the positive-signal workflow is unresolved. It should also remain supplementary when the legacy assay provides functional evidence—such as infectivity, replication competence, or process clearance—that the sequencing result cannot establish.

Other warning signs include an uncurated or undocumented reference database, uncontrolled changes to the bioinformatics pipeline, an absence of meaningful negative controls, and sensitivity claims based solely on a clean matrix. In these cases, more sequencing depth is not the remedy. The remedy is a better-defined analytical question and a validation design that tests the entire workflow.

Conclusion

NGS is changing how CGT developers approach broad adventitious-virus detection. Its main strength is not that it makes every older assay obsolete. Its strength is that it can generate a broader, sequence-based view of known and unexpected viruses when the workflow is designed and validated for a specific sample and decision.

The practical replacement strategy is therefore modular. Replace broad detection assays where NGS supplies fit-for-purpose evidence. Supplement assays when NGS adds breadth but does not answer the biological endpoint. Retain dedicated studies for infectivity, replication-competent virus, and viral clearance when those endpoints remain necessary. That framework turns NGS from a technology claim into a defensible analytical plan.

FAQ

Research Use Only: This content is intended for scientific and analytical-development discussion and does not establish product release requirements.

References:

1. International Council for Harmonisation. ICH Q5A(R2): Viral Safety Evaluation of Biotechnology Products Derived from Cell Lines of Human or Animal Origin. 2023.
2. U.S. Food and Drug Administration. FDA Guidance on Replication-Competent Retrovirus Testing. 2020.
3. Chin P-J, Tsou J-H, Armstrong A, et al. Evaluation of high-throughput sequencing for replacing the conventional adventitious virus detection assays used for biologics. npj Vaccines. 2026;11:28. DOI: 10.1038/s41541-025-01351-2. CC BY 4.0.
4. Chin P-J, Lambert C, Beurdelay P, et al. Virus detection by short read high throughput sequencing in a high virus low cellular background. npj Vaccines. 2025;10:61. DOI: 10.1038/s41541-025-01104-1. CC BY 4.0.
5. Ng SH, Braxton C, Eloit M, et al. Current Perspectives on High-Throughput Sequencing for Adventitious Virus Detection: Upstream Sample Processing and Library Preparation. Viruses. 2018;10(10):566. DOI: 10.3390/v10100566. CC BY 4.0.

Related Services

• Viral Metagenomic Sequencing
• Viral Genome Sequencing
• RNA-Seq
• Targeted Region Sequencing
• Amplicon Sequencing Services
• AAV Genome Sequencing
• Lentiviral/Retroviral Integration Sites Analysis
• Cell Line Identification