Long Read Rna Sequencing

This guide explains long read RNA sequencing, a technology that captures full length RNA molecules to study transcript diversity, splicing, and repetitive regions. It is intended for researchers considering or starting a long read RNA sequencing project, especially those studying complex transcriptomes, isoform function, or repetitive genomic elements. For an authoritative background on sequencing technologies, see the NCBI Bookshelf NCBI Bookshelf.

Unlike short read sequencing, which fragments RNA into roughly 150 base pieces, long read platforms such as those from Pacific Biosciences (Iso Seq) and Oxford Nanopore Technologies (cDNA or direct RNA sequencing) can read transcripts from end to end. This enables direct detection of isoforms, polyA tails, and base modifications. Training resources from EMBL EBI describe the principles of long read sequencing EMBL-EBI Training.

At a Glance

Core Concepts and Decision Points

Long read RNA sequencing is not one monolithic method. You must decide between cDNA based and direct RNA sequencing, as well as platform (PacBio versus Oxford Nanopore). Direct RNA sequencing reads the native RNA molecule, preserving base modifications but requiring high quality RNA and yielding lower throughput. cDNA based approaches use reverse transcription and PCR, which can introduce artifacts but provide higher yield Galaxy Training Network.

Decision points to consider:

1. What is your primary research question? If you need full length isoform sequences, long reads are essential. If you require accurate quantification of known genes, short reads may be more cost effective.
2. Do you need to detect RNA base modifications? Only direct RNA sequencing (on Nanopore) preserves modifications like m6A.
3. What is your starting RNA quality? Long read protocols are sensitive to RNA degradation. Check RNA integrity numbers carefully.
4. Are you working with repetitive transcript regions? Long reads can span repeats, making them ideal for transposable element derived transcripts. A recent study used long read RNA sequencing to map transposable element variation in human brain tissue Long read sequencing maps transposable element variation and its regulatory and epigenetic effects in the human brain.
5. What is your budget? PacBio Iso Seq typically costs more per run than Nanopore but offers higher consensus accuracy. Nanopore allows real time data streaming and lower entry cost.

A systematic benchmark of computational methods for single cell and spatial long read RNA data provides practical guidance on tool selection for preprocessing, alignment, and isoform detection A systematic benchmark of bioinformatics methods for single cell and spatial RNA seq nanopore long reads data.

Practical Workflow or Implementation Steps

A typical long read RNA sequencing project follows these steps. Each step references publicly available training materials.

1. Library Preparation

For cDNA based approaches, choose between full length versus targeted enrichment. Use a reverse transcriptase capable of generating long cDNA, such as the template switching method. For direct RNA, ligate adapters to native RNA. Detailed protocols are available through the Galaxy Training Network Galaxy Training Network.

2. Sequencing

Run on PacBio Sequel IIe or Oxford Nanopore PromethION. For Nanopore, select the appropriate flow cell (R10.4 for basecalling accuracy). Basecall in real time using Guppy or Dorado. The data are deposited in the NCBI Sequence Read Archive for public access NCBI Sequence Read Archive.

3. Basecalling and Demultiplexing

Convert raw electrical signals to base sequences. For Nanopore, use super accuracy models. Demultiplex using Dorado or MinKNOW.

4. Adapter Trimming and Polishing

Trim adapter sequences and perform consensus polishing if using circular consensus (PacBio). For Nanopore, optional polishing with Medaka or Racon improves accuracy.

5. Read Alignment and Transcript Discovery

Align reads to the reference genome using long read aware aligners like Minimap2, deSALT, or GraphMap. Run transcript assembly tools such as StringTie2 (with long read mode) or FLAIR (for isoform detection). Open software from Bioconductor offers packages like BSgenome for reference management and rMATS for splicing analysis Bioconductor.

6. Quantification and Differential Expression

Quantify isoform expression using Salmon (long read mode) or Nanocount. Tools like DESeq2 can be adapted, but note that long read quantification is less mature. Consider using a dedicated long read differential expression method.

7. Downstream Analysis

Analyze isoform switching, alternative splicing, and novel transcripts. For de novo transcriptome assembly (e.g., for non model species), long reads alone can produce highly contiguous assemblies. A study on Heliconius butterflies demonstrated that long read sequencing alone enabled high quality de novo genome assembly de novo assembly of 5 highly contiguous Heliconius butterfly genomes with long read sequencing alone. Apply similar principles to transcriptome assembly.

Quality Checks and Common Mistakes

Quality Checks

• Check read length distribution. Short reads (<200 bases) may indicate RNA degradation or adapter dimers.
• Evaluate accuracy by aligning to a reference genome. Expect >85% for raw Nanopore reads, >99% for PacBio circular consensus.
• Confirm that polyA tails are detected in direct RNA data, indicating intact 3' ends.
• Use RNA integrity number (RIN) values from the input RNA, aim for RIN >8.

Common Mistakes

1. Assuming short read tools work out of the box. Many short read aligners and assemblers fail on long reads due to higher error rates. Always use long read specific software.
2. Ignoring the globin depletion paradox. In blood derived samples, globin mRNA represents up to 70% of transcripts. Short read protocols often include a depletion step, but certain depletion methods do not transfer well to long read RNA sequencing, leading to unintended loss of other transcripts. A recent study highlighted this issue The globin depletion paradox: when short read optimisation does not transfer to long read RNA sequencing. Evaluate depletion strategies empirically for long read experiments.
3. Overlooking base modification artifacts. In direct RNA sequencing, modified bases can cause basecalling errors. Use a modification aware basecaller or treat them as a source of signal rather than noise.
4. Neglecting to optimize reverse transcription. For cDNA methods, using a standard reverse transcriptase may produce truncated cDNA. Use a thermostable, processive enzyme and include a denaturation step for structured RNAs.
5. Failing to benchmark tools on your specific data. A tool that works well for bacterial transcriptomes may underperform on human repetitive regions. The systematic benchmark previously cited [6] provides a framework for testing.

Limits of Interpretation

Long read RNA sequencing has several important limitations that affect biological interpretation.

Accuracy. Raw Nanopore reads have error rates around 5,15%, which complicates single nucleotide variant calling. PacBio circular consensus improves accuracy to >99%, but requires high coverage. For applications requiring precise base identity (e.g., variant effect assays), short reads may still be superior. A scaled approach used for hypertrophic cardiomyopathy variants combined both short and long reads to achieve sequence function resolution Scaled Multidimensional Assays of Variant Effect Identify Sequence Function Relationships in Hypertrophic Cardiomyopathy.

Quantification bias. Long reads often underrepresent short transcripts and overrepresent long ones due to fragmentation during library preparation. Normalize by transcript length and use spike in controls. Absolute quantification is not reliable without external standards.

Isoform detection false positives. Due to sequencing errors, many apparent novel isoforms are artifacts. Apply strict filtering: require at least 2 reads supporting a unique splice junction, and compare against known annotations. Use tools like SQANTI3 to classify isoforms.

Reproducibility. Protocols are still evolving. A protocol that works today may need adjustment after a platform update. Always test a pilot experiment.

Interpretation of repetitive elements. While long reads excel at spanning repeats, the alignment of reads to highly repetitive regions (e.g., centromeres) remains ambiguous. The accompanying brain study on transposable elements [11] noted that many TE derived transcripts are multi mapped, requiring specialized quantification methods.

Frequently Asked Questions

1. Can I use long read RNA sequencing for gene expression quantification?
Yes, but with caution. Long read quantification is less accurate than short read due to higher error rates and systematic biases. Use dedicated tools like Salmon (long read mode) and include spike in controls. For absolute quantification, short read is still preferred.

2. Do I need a reference genome for long read RNA sequencing?
A reference genome is helpful for alignment and isoform classification, but it is not strictly required. For de novo transcriptome assembly, long reads can produce contigs without a reference. However, gene annotation quality is higher with a genome. The Durum wheat reference genome, for example, was improved using long reads Durum Wheat cv. Svevo Reference Genome Rel.2.0: A Comprehensive Tool for Wheat Genomics.

3. How much sequencing depth do I need?
For full length isoform detection in a mammalian transcriptome, aim for 5,10 million mapped reads for moderate complexity. For rare isoforms, higher depth (20,30 million) may be required. For direct RNA sequencing, throughput is lower, 1,2 million reads is typical.

4. Can I detect RNA modifications with long read sequencing?
Only direct RNA sequencing (Oxford Nanopore) can detect base modifications in native RNA. Computational tools like Nanopolish, Tombo, or m6Anet analyze electrical signals to identify modified positions. PCR based methods destroy modifications.

References and Further Reading

• NCBI Bookshelf on sequencing technologies NCBI Bookshelf
• EMBL EBI Training resources for long read RNA EMBL-EBI Training
• Galaxy Training Network workflow tutorials Galaxy Training Network
• Bioconductor packages for genomic data analysis Bioconductor
• NCBI Sequence Read Archive for public data NCBI Sequence Read Archive
• Benchmark of single cell and spatial long read methods A systematic benchmark of bioinformatics methods for single cell and spatial RNA seq nanopore long reads data
• Globin depletion paradox study The globin depletion paradox: when short read optimisation does not transfer to long read RNA sequencing
• Transposable element mapping in brain Long read sequencing maps transposable element variation and its regulatory and epigenetic effects in the human brain
• De novo assembly of Heliconius genomes de novo assembly of 5 highly contiguous Heliconius butterfly genomes with long read sequencing alone
• Variant effect assay using long reads Scaled Multidimensional Assays of Variant Effect Identify Sequence Function Relationships in Hypertrophic Cardiomyopathy

Related Articles

• Protein Synthesis
• Incomplete Dominance Gene
• Cell Membrane Function Biology
• Dna Structure
• Protein Structure