Long-Read Sequencing for Gene Therapy and AAV Characterization

Gene-therapy vector characterization becomes difficult when the question is not simply whether a transgene sequence is present, but whether the complete vector genome is packaged as designed, which partial or rearranged forms coexist in the preparation, whether plasmid or host-cell sequences are present, and how vector DNA is organized after delivery. These are fundamentally long-range sequence questions.

CD Genomics provides long-read sequencing for gene therapy and AAV characterization using complementary PacBio HiFi and Oxford Nanopore Technologies (ONT) workflows. We help research and biopharma teams characterize packaged recombinant AAV (rAAV) genomes, plasmid templates, complex vector forms, host-vector junctions, integration architecture, and selected epigenetic features with read-level structural evidence that can span entire vector molecules or long host-genome contexts.

Solution highlights

• Profile full-length, partial, truncated, rearranged, and concatemer-like AAV genome structures.
• Review ITR-to-ITR vector architecture and recurrent start/end or breakpoint patterns.
• Detect vector-associated plasmid backbone, helper/packaging, or host-cell DNA sequences when captured.
• Reconstruct host-vector junctions and complex integration structures in cell or tissue DNA.
• Compare production plasmids, vector preparations, cell banks, and experimental samples across development stages.
• Use PacBio HiFi, ONT, or a combined design according to the structural, accuracy, native-DNA, and throughput question.

Discuss Your AAV Characterization Study

Why AAV Characterization Is a Long-Range Sequence Problem

A recombinant AAV vector may be only a few kilobases long, yet the molecular population inside a vector preparation can be much more complex than a single expected reference sequence. Replication and packaging can generate incomplete genomes, asymmetric truncations, snapback or self-complementary structures, rearrangements, concatemers, reverse-packaged fragments, and DNA derived from production plasmids or host cells. A conventional coverage plot can show that all vector regions are represented while still obscuring how those regions are connected on individual molecules.

Long reads address that connectivity problem. When a read spans most or all of a packaged vector genome, the molecule can be interpreted as an intact structural unit rather than reconstructed from short fragments. This supports direct review of where molecules start and end, which vector elements remain linked, whether an unexpected fragment is fused to an ITR-bearing sequence, and whether distinct structural populations coexist in the same preparation. For upstream construct confirmation, our Full-Length Plasmid Sequencing service provides a complementary view of the DNA templates used during vector development.

The same long-range logic becomes even more important after transduction. Vector DNA in cells or tissues can exist as episomal forms, concatemers, truncated molecules, or host-integrated sequences. A host-vector junction may lie kilobases away from the transgene segment of interest, and an integration can contain multiple vector copies in different orientations. Targeted long-read or long-read whole-genome designs can retain these relationships on individual molecules when the relevant DNA is captured.

Long-read sequencing is therefore best viewed as a structural characterization layer. It complements, rather than replaces, assays for capsid concentration, vector genome titer, empty/full capsid ratio, potency, infectivity, residual DNA quantification, or validated release testing. The sequencing question is: what DNA structures are present, and how are their components connected?

Packaged AAV Genome Characterization: Read the Molecule, Not Only the Coverage

Vector identity and full-length architecture

For purified rAAV preparations, long-read sequencing can be used to compare captured vector-derived molecules with the expected ITR-to-ITR design. Depending on sample preparation and platform, reads may span the complete vector genome or large fractions of it, allowing direct confirmation of element order, orientation, payload continuity, and sequence identity. PacBio HiFi is particularly useful when high consensus accuracy is important for sequence-level review, while ONT provides a flexible route for long-molecule profiling and can be advantageous when native-DNA information is part of the design.

Short-read sequencing remains valuable for high-depth base-level coverage, but fragmentation breaks the physical linkage among distant vector elements. For AAV, that linkage is often the point of the experiment. If a 5′ ITR-associated fragment, promoter, transgene segment, and plasmid-backbone sequence occur on one long molecule, that read provides structural information that four independent short-read coverage peaks cannot.

Partial genomes, truncations, and rearrangements

Long-read data can be classified at the read level to distinguish expected full-vector structures from partial or structurally altered forms. Useful outputs include length distributions, alignment start/end maps, breakpoint positions, orientation patterns, and molecule-level schematics. Recurrent start or end positions can reveal preferred truncation regions, while complex mappings can identify rearranged or concatemer-like molecules.

Published AAV studies demonstrate why this matters. Direct ITR-to-ITR nanopore sequencing has identified truncation hotspots in single-stranded and self-complementary vectors, while PacBio-based AAV genome population sequencing has shown that preparations appearing homogeneous by lower-resolution methods can contain substantial structural heterogeneity. These studies also show that platform and library-preparation biases must be considered when interpreting molecule frequencies. We therefore report read-supported structure with the relevant assay limitations rather than convert every read count into an absolute product-quality claim.

ITRs require explicit interpretation

Inverted terminal repeats are central to AAV replication and packaging and are also challenging sequencing features because of their secondary structure. The goal is not to claim that every ITR base is recovered perfectly in every workflow. Instead, we evaluate ITR-associated read starts and ends, ITR-spanning or ITR-adjacent sequence evidence, and structural continuity across the vector while documenting platform- and library-specific limitations. When an ITR sequence discrepancy could drive a development decision, we recommend orthogonal confirmation using a method designed for that specific question.

Residual and chimeric DNA sequence evidence

A long-read reference search can include the intended vector, production plasmids, helper or packaging constructs, and relevant host genomes. Reads that map partly or fully to these references can reveal plasmid-backbone fragments, rep/cap/helper-derived sequences, host-cell DNA fragments, or chimeric molecules that contain both AAV and non-AAV sequence. The analysis is sequence-specific: it tells you what captured molecules contain. It does not replace a validated quantitative residual-DNA assay when a program needs a formal concentration limit.

For production-template and cell-line questions that extend beyond packaged AAV DNA, our Viral Whole-Genome Resequencing capability and related long-read genomic workflows can be integrated into a broader characterization plan.

PacBio HiFi, ONT, or a Combined AAV Strategy?

Our PacBio SMRT Sequencing Technology and Oxford Nanopore Sequencing Technology pages describe the broader platform principles. For this solution, platform choice is made from the AAV question, sample type, required sequence confidence, need for native-DNA information, and whether the study is targeted or genome-scale.

Bioinformatics Analysis and Evidence Reporting

Every final report distinguishes direct read evidence, algorithmic classification, inferred interpretation, and items recommended for orthogonal validation. This is especially important for rare structural forms, ITR-associated discrepancies, and low-frequency integration events where library preparation, enrichment, and mapping can influence observed frequencies.

Project Entry Modes and Sample Types

Exact input quantity and quality requirements depend on platform, assay scale, and whether native high-molecular-weight DNA must be preserved. We review sample requirements before project initiation rather than publish one universal threshold for all AAV workflows. General shipment and preparation guidance is available in our Sample Submission Guideline.

Independent Published Example: Long-Read Reconstruction of AAV Genomes and Integration Architecture

Greig JA, Martins KM, Breton C, et al. Integrated vector genomes may contribute to long-term expression in primate liver after AAV administration. Nature Biotechnology. 2024;42:1232–1242. doi:10.1038/s41587-023-01974-7.

Background

Greig and colleagues investigated the persistence and expression of liver-directed AAV8 and AAVrh10 vectors in nonhuman primates over extended follow-up. The study combined molecular, histologic, transcriptomic, integration-site, and long-read approaches to understand why vector DNA could persist even as transgene expression declined. This makes the work relevant to gene-therapy characterization because it separates the structure of administered vector preparations from the structure of vector DNA that remained in tissues.

Methods

For late tissue samples, the authors enriched vector-containing high-molecular-weight DNA with probes tiled across vector transcriptional units and performed PacBio Sequel II HiFi circular consensus sequencing. Reported CCS reads ranged from 51 to 50,419 bp, with average sequence sizes across samples ranging from 4,216 to 6,113 bp. Reads containing vector sequence were mapped against both vector and host genomes to characterize vector organization and host-genome junctions.

Results

The study first showed that complete ITR-to-ITR vector genomes represented the majority of DNA in the vector preparations used for dosing, while a minority population contained truncations that were not present in the input plasmid. In late liver samples, long-read analysis revealed extensive heterogeneity, including rearranged and truncated vector segments organized into complex concatemers. The authors observed head-to-tail, head-to-head, and tail-to-tail concatemeric configurations and captured host-vector junctions within long individual molecules.

Integration-site analysis showed that vector integration events persisted at broadly distributed loci. At later time points, reported integration levels stabilized in the range of 0.1–0.7 AAV integration events per 100 genomes in the evaluated samples. The authors also reported that the identified integration sites in these nonhuman primates were not located in the genic regions they evaluated as frequently mutated in human hepatocellular carcinoma or previously implicated in AAV-associated mouse liver tumors.

Figure 5 from Greig et al. (Nature Biotechnology, 2024, CC BY 4.0) shows representative structures of integrated vector DNA after AAV administration, including complex concatemeric configurations and host genomic junctions.

Conclusion

This independent published example illustrates two distinct uses of long-read sequencing in an AAV program: profiling the integrity of the administered vector genome and reconstructing the complex organization of vector DNA in host tissue. It also shows why the expected plasmid or packaged-vector sequence is not sufficient to predict every structure that may arise after delivery. The findings are presented here as literature evidence, not as a CD Genomics customer project or a performance guarantee.

Sample Deliverables

1. AAV genome integrity overview — read-level classification of expected full-vector structures, partial/truncated molecules, rearrangements, concatemer-like forms, and other supported categories.

2. Read start/end and breakpoint map — recurrent vector-genome start positions, end positions, structural breakpoints, and hotspots mapped against annotated vector elements.

3. Non-vector sequence report — reads mapping to production plasmids, helper/packaging sequences, host genomes, and vector-linked chimeric molecules.

4. Integration-site and architecture report — host-vector junction coordinates, surrounding host sequence, vector-copy orientation, concatemer structure, and representative long reads.

5. Comparative construct/lot summary — consistent structural metrics and plots for side-by-side review across vector designs, lots, or research time points.