MAS-Seq and Kinnex Explained: Scaling Full-Length Isoform Sequencing in Single Cells

Inquiry

Single-cell RNA sequencing has reshaped biomedical research, but standard 3'/5' tag-based short-read workflows provide limited information about complete transcript structure. They are highly effective for gene-level abundance and cell-state profiling, yet individual short reads generally do not preserve full exon connectivity, alternative transcript start or end sites, or complete splice isoforms. Long-read single-cell sequencing addresses this structural-information gap, although historically lower long-read throughput made large cell-resolved isoform studies more difficult to scale.

Multiplexed Arrays Sequencing (MAS-Seq), including the MAS-ISO-seq implementation, was developed to improve the efficiency of long-read transcript sequencing by concatenating barcoded cDNA molecules into ordered arrays before sequencing. The original MAS-ISO-seq study reported a greater than 15-fold increase in cDNA-read throughput under the evaluated Sequel IIe workflow, while the commercial Kinnex single-cell RNA workflow uses approximately 16-segment arrays to increase effective long-read throughput. This article explains the underlying concept, clarifies the relationship between MAS-Seq and Kinnex, and provides a practical framework for deciding when a single-cell project benefits from full-length isoform resolution.

Quick Method Picker: What Problem Does MAS-Seq / Kinnex Solve?

• Need to identify cell-type-specific alternative splicing and isoform switching → MAS-Seq / Kinnex Full-Length scRNA-Seq
• Need standard cell type clustering and broad differential gene expression → Standard Short-Read 3'/5' scRNA-Seq
• Need to detect candidate fusion transcripts or transcript-structure variation in specific cell populations → Full-Length Long-Read scRNA-Seq
• Need bulk tissue transcriptome re-annotation or baseline isoform discovery → Bulk Long-Read Iso-Seq
  Key technical check: confirm that the biological question depends on transcript structure and that the starting full-length cDNA is suitable for long-read library construction. This discussion focuses on research-use discovery workflows and computational study design, not clinical diagnostic testing.

For an overview of broader single-cell and spatial integration principles, see Bulk RNA-seq vs Single-Cell RNA-seq vs Spatial Transcriptomics: How to Choose for Tissue Studies.

At-a-Glance Comparison of Single-Cell Transcriptomic Readouts

Evaluating single-cell sequencing strategies requires understanding what each molecular readout measures, what it infers indirectly, and whether the added structural information is necessary for the study hypothesis.

Defining MAS-Seq vs Kinnex: Conceptual and Commercial Architecture

• MAS-Seq / MAS-ISO-seq: The foundational approach described by Al'Khafaji et al. uses programmed, directional cDNA concatenation to assemble multiple transcript molecules into ordered arrays that better use the sequencing capacity of long-read platforms.
• Kinnex Single-Cell RNA: The standardized commercial workflow from Pacific Biosciences builds on MAS-Seq concatenation chemistry for scalable single-cell isoform profiling. Current PacBio documentation supports Kinnex single-cell RNA sequencing on compatible HiFi systems, including Sequel II/IIe, Revio, and Vega configurations.
• cDNA Concatenation Array: A long library molecule containing multiple individual cDNA segments. Current Kinnex guidance describes 16-segment array construction, with expected library sizes commonly around 11–14 kb under the specified workflow.
• De-concatenation (Segmentation): The computational process, implemented in PacBio workflows with tools such as skera, that identifies array adapters and separates a HiFi read back into the original cDNA segments while preserving cell-barcode and molecule-level tagging information for downstream analysis.
• Differential Isoform Usage (DIU): A change in the relative abundance of transcript isoforms from the same gene across cell types, states, or experimental conditions, which can occur even when total gene-level expression changes little.

The Mechanism of Programmed cDNA Concatenation

The key engineering problem is a mismatch between the relatively short length of many single-cell cDNA molecules and the much longer templates that long-read sequencing can process efficiently. In the original MAS-ISO-seq study and current Kinnex workflows, multiple cDNA molecules are assembled into longer arrays so that one long-read template can yield many segmented transcript reads after computational processing. Current Kinnex application guidance describes expected single-cell library sizes around 11–14 kb, while the original MAS-ISO-seq design demonstrated the same general strategy using ordered multi-cDNA arrays.

Figure 2. Conceptual cDNA concatenation workflow: single-cell cDNA generation, multi-segment array formation, HiFi sequencing, computational segmentation, and downstream isoform analysis.

MAS-Seq and Kinnex address this mismatch through four linked stages:

1. Single-Cell Full-Length cDNA Generation

Single-cell cDNA can be generated using compatible upstream platforms such as 10x Genomics Chromium Single-Cell RNA-Seq. Each transcript is associated with a cell barcode and the molecule-level tagging information used by the upstream workflow. Full-length cDNA integrity is important because truncation introduced before long-read library preparation cannot be recovered later by concatenation.

2. Directional Array Assembly

Rather than sequencing each cDNA molecule as an independent long-read template, MAS-Seq-style workflows use directional adapter chemistry to organize multiple cDNA segments into ordered arrays. The original method and current Kinnex implementation use approximately 16 segments per full array. Exact reaction configuration, input requirements, and acceptance criteria should follow the applicable validated workflow or current kit documentation rather than being generalized from a single study.

3. High-Accuracy Long-Read Sequencing

The concatenated library is sequenced on a compatible PacBio HiFi system. Because the library molecule contains multiple transcript segments, a single HiFi read can contribute multiple cDNA sequences after segmentation, improving effective transcript throughput compared with sequencing short cDNA templates individually.

4. Bioinformatic Segmentation and Molecule Recovery

During data processing, known array-adapter patterns are identified and the long HiFi read is segmented into its constituent cDNA sequences. Cell barcodes and molecule-level tags are then recovered for mapping, transcript classification, isoform quantification, and downstream single-cell analysis.

For downstream analytical methods, see Single-Cell RNA-Seq Data Analysis Service.

When Is Isoform-Level Resolution Worth the Investment?

Moving from standard short-read gene counting to full-length isoform sequencing is most useful when the biological hypothesis depends on transcript structure rather than gene abundance alone. The decision should be driven by the expected biological signal, sample quality, cell numbers, and whether orthogonal validation is feasible.

Figure 3. Decision framework for choosing between short-read gene-level profiling and full-length isoform-resolved single-cell sequencing.

Scenario A: Cell-Type-Specific Splicing and Isoform Switching

• Biological Question: Does a specific cell population express a different transcript isoform even when total gene-level expression is similar?
• Molecular Readout: Alternative exon inclusion or exclusion, mutually exclusive exons, alternative transcript start or end sites, and differential isoform usage.
• Project Implication: Gene-level counts can remain similar while isoform usage changes. Full-length reads provide direct transcript-structure evidence that 3'/5' tag-based counting may not resolve. PTPRC isoform transitions in T-cell biology are a useful published example of why transcript structure can matter, but the relevance of that example should not be generalized to unrelated tissues, species, or mechanisms without supporting evidence.

Scenario B: Fusion Transcript Characterization and Cell-Population Assignment

• Biological Question: Which cell populations express candidate chimeric transcripts or fusion isoforms?
• Molecular Readout: Full-length reads spanning transcript-level fusion junctions together with cell-barcode information.
• Project Implication: Long reads can associate a candidate fusion transcript with particular cell populations and co-expression states more directly than short junction-spanning reads. However, an RNA fusion transcript does not by itself establish the corresponding genomic rearrangement; DNA-level breakpoint validation is required when a genomic structural event is being claimed.

Scenario C: Unannotated Transcripts and Novel Isoform Discovery

• Biological Question: Does a specialized tissue, developmental state, or non-model organism express transcript structures that are missing from the current annotation?
• Molecular Readout: Molecule-spanning exon chains, novel splice junction combinations, and alternative transcript boundaries.
• Project Implication: Short-read reconstruction becomes ambiguous when multiple isoforms share exons and splice junctions. Long reads provide molecule-spanning evidence for transcript structure, but candidate novel isoforms still require stringent computational filtering and, when biologically important, orthogonal experimental validation.

Scenario D: When Full-Length Isoform Sequencing May Not Add Enough Value

• Biological Question: Is the primary goal to identify major cell populations, broad expression programs, or strongly differentially expressed genes?
• Project Implication: If transcript structure is not central to the hypothesis, standard short-read single-cell RNA sequencing may provide sufficient resolution with simpler analysis and broader cell coverage. Long-read sequencing should be added when it is expected to change biological interpretation, not simply because more molecular detail is technically available.

To evaluate wider single-cell study designs, explore our comprehensive Single-Cell Sequencing Service and our guide on Short-Read vs Full-Length Transcriptomics in Single-Cell Research: What Will Matter Most in the Next Decade?.

Key Quality Control Checkpoints and Technical Considerations

High-quality single-cell full-length transcriptome data depend on protocol-aware quality control rather than one universal threshold. The numerical values below should be interpreted as published or manufacturer-recommended starting points for specific workflows, not as pass/fail rules for every sample or project.

1. Initial Cell and cDNA Quality
   For many droplet-based single-cell workflows, viability above approximately 85% is a useful practical starting point when fresh cells are available, but acceptable input depends on tissue type, dissociation strategy, and upstream platform. The amplified cDNA distribution should also be inspected for truncation and excessive short products. Many single-cell cDNA preparations are concentrated in the ~1–2.5 kb range, but the complete size distribution and current platform-specific specifications are more informative than a single peak value.
2. Array Assembly Efficiency
   Current PacBio application guidance for Kinnex single-cell RNA libraries lists an expected library size of approximately 11–14 kb and an expected full-array fraction around 85–92% under the specified workflow. These values are useful run-level reference points, but should not be treated as universal acceptance thresholds for every chemistry version, input type, or instrument configuration.
3. Read Segmentation Performance
   Track the number and fraction of HiFi reads that can be segmented into valid cDNA reads, the distribution of array sizes, and the abundance of incomplete or malformed arrays. Current PacBio technical guidance commonly shows full arrays containing about 16 segments, with mean array sizes near 15 or more segments in well-performing example runs. Interpretation should remain tied to the current kit version and run controls.
4. Cell-Barcode Recovery and Representation
   Evaluate valid barcode recovery, reads assigned to cells, cell representation across clusters, and molecule-level deduplication. When matched short-read data are available, compare cell-barcode composition and major cell populations across platforms to detect disproportionate dropout or library-specific bias.

For further validation workflows, see How To Validate Single-Cell RNA-Seq Data?.

Common Pitfalls and Practical Tips

• Do not assume high gene expression guarantees adequate minor-isoform coverage.
  Highly expressed genes can contain several competing isoforms, and the dominant transcript may consume much of the available evidence. Evaluate isoform-level saturation or down-sampling rather than relying only on gene-level counts.
• Avoid relying solely on a reference annotation.
  Reference databases such as Ensembl and RefSeq are valuable starting points but may not represent rare, cell-state-specific, developmental, or species-specific isoforms. Use long-read-aware tools such as IsoQuant, Bambu, or SQANTI3 to classify candidate transcript structures and flag lower-confidence models.
• Protect full-length cDNA before long-read library construction.
  Degraded, truncated, or heavily contaminated cDNA reduces the chance of recovering complete transcript structures. Cleanup and integrity assessment should follow the validated upstream and Kinnex workflow rather than being inferred from array-sequencing performance alone.
• Distinguish biological transcript diversity from technical truncation.
  Apparent alternative transcript starts or incomplete splice structures can arise from reverse-transcription or library-preparation artifacts. Evaluate canonical splice support, transcript completeness, recurrence across cells or replicates, and independent evidence before interpreting a novel isoform as a biological state marker.
• Do not infer causality from differential isoform usage alone.
  An isoform associated with a cell state, treatment, or phenotype may be a marker rather than a driver. Functional perturbation, targeted transcript validation, or protein-level follow-up may be necessary when a causal mechanism is proposed.

What MAS-Seq / Kinnex Cannot Establish on Its Own

• Transcript structure does not establish protein function. A full-length RNA isoform can be detected confidently without demonstrating that the encoded protein is translated, stable, or functionally distinct. Protein-level or functional validation may be required.
• A fusion transcript does not prove a genomic breakpoint. When the biological claim concerns a DNA rearrangement, orthogonal DNA-level evidence should be obtained.
• Differential isoform usage is an association unless experimentally tested. Observed relationships between isoforms and cell states should not be described as causal without perturbation or other functional evidence.
• Long-read sequencing does not replace biological replication. Appropriate biological replicates, batch-aware design, controls, and independent validation remain important for reproducible conclusions.

Data and Code Traceability

Reproducible analysis should record software versions, reference genome and annotation releases, barcode-processing settings, transcript-classification rules, and filtering parameters. PacBio workflows use read segmentation, including skera, before single-cell Iso-Seq processing. Spliced alignment can be performed with tools such as pbmm2 or minimap2, while downstream transcript discovery and classification can use packages such as IsoQuant, Bambu, or SQANTI3. Tool choice should be documented because transcript models can vary with annotation, filtering strategy, and software version.

FAQs

Ready to Advance Your Single-Cell Research

Full-length single-cell transcript sequencing can add a structural layer that gene-level counting alone does not provide, revealing cell-associated splicing patterns, transcript boundaries, and candidate isoform changes. MAS-Seq established a scalable concatenation strategy, while Kinnex provides a standardized commercial implementation for current PacBio long-read workflows.
For research teams evaluating whether isoform-level resolution will change project interpretation, CD Genomics supports research-use-only single-cell full-length transcriptome projects through project consultation, sample quality assessment, long-read sequencing, and bioinformatics analysis. Explore our dedicated Single-Cell Full-Length RNA Sequencing Service for Isoform and Splicing Analysis.

References

1. Al'Khafaji AM, Smith JT, Garimella KV, et al. High-throughput RNA isoform sequencing using programmed cDNA concatenation. Nature Biotechnology. 2024;42(4):582–586.
2. Zajac N, Zhang Q, Bratus-Neuenschwander A, et al. Comparison of single-cell long-read and short-read transcriptome sequencing via cDNA molecule matching: quality evaluation of the MAS-ISO-seq approach. NAR Genomics and Bioinformatics. 2025;7(3):lqaf089.
3. Kumari P, Kaur M, Dindhoria K, et al. Advances in long-read single-cell transcriptomics. Human Genetics. 2024;143(9-10):1005–1020.
4. Prjibelski AD, Mikheenko A, Joglekar A, et al. Accurate isoform discovery with IsoQuant using long reads. Nature Biotechnology. 2023;41(7):915–918.
5. Pacific Biosciences. Application note: Kinnex single-cell RNA kit for single-cell isoform sequencing. Updated 2025.