Pacbio Long Read Sequencing
PacBio long read sequencing (single molecule real time sequencing) produces reads that average 10 to 25 kilobases, with some exceeding 100 kilobases, and offers a direct view of structural variation, repetitive regions, and base modifications. This guide is for researchers and bioinformaticians who need a practical, source bounded framework to decide when and how to use this technology, from project design through data interpretation.
Long read sequencing has transformed genomics by resolving regions that short reads cannot span consistently. If you work on de novo genome assembly, structural variant detection, or transcript isoform analysis, you need to understand how PacBio HiFi (high fidelity) and continuous long read (CLR) data differ, where they excel, and where they fall short. Trustworthy technical references are available on the NCBI Bookshelf and through EMBL EBI Training resources.
At a Glance
| Aspect | Clarification |
|---|---|
| What it does | Directly sequences single DNA molecules in real time, producing long reads (typically 10 25 kb for HiFi, up to 100+ kb for CLR). |
| Key advantage | Resolves repetitive regions, detects structural variants, and enables contiguous de novo assemblies without a reference. |
| Read accuracy | HiFi reads have 99.9% per base accuracy (Q30+). Traditional CLR reads have lower single pass accuracy (around 85 90%) but can be polished. |
| Best suited for | De novo assembly of microbial and eukaryotic genomes, haplotype phasing, structural variant discovery, full length transcript sequencing, and base modification detection. |
| Primary limitation | Higher per base cost compared to short reads, higher DNA input and quality requirements, computationally intensive data processing. |
| Typical output file type | BAM (aligned) or FASTA/FASTQ (raw reads). |
Core Concepts and Decision Points
PacBio sequencing works by immobilizing a single DNA polymerase molecule at the bottom of a zero mode waveguide (ZMW). As nucleotides labeled with different fluorophores are incorporated, light pulses are recorded in real time. The pattern of pulses is translated into a sequence. This process was described in detail in authoritative sources such as the NCBI Bookshelf.
Today the decision point is usually between HiFi reads and CLR reads. HiFi reads are generated by circular consensus sequencing (CCS). The polymerase reads the same template multiple times, and the consensus is highly accurate. For most applications including genome assembly and variant detection, HiFi is the recommended choice. CLR reads are longer but less accurate. They remain useful when the goal is maximal contiguity at the expense of per base accuracy, for example in scaffolding a reference genome.
A second decision point is sequencing depth. For a human genome, 15 to 30 fold coverage of HiFi reads yields a highly contiguous assembly. For bacterial genomes, 50 to 100 fold coverage is typical. You can explore public PacBio datasets on the NCBI Sequence Read Archive to see what depths were used in published studies.
A third decision point is whether to combine PacBio with other technologies. For example, the genome of the rat tapeworm Hymenolepis diminuta was assembled entirely from PacBio long reads, demonstrating that a single technology can suffice for a high quality reference genome source: Assembly of a high quality reference genome for the rat tapeworm Hymenolepis diminuta. But for some projects, optical maps or Hi C data may help resolve complex repeats.
Practical Workflow
1. Library Preparation
Extract high molecular weight (HMW) DNA. Shearing methods vary by target read length. For HiFi, shear to 10 20 kb. For CLR, shear to 50 kb or more. Size selection is critical. Short fragments degrade assembly quality. Quality control with pulsed field gel electrophoresis or a TapeStation ensures that the majority of fragments are above the target size. The Galaxy Training Network offers step by step tutorials for QC of long read libraries.
2. Sequencing
Load the library onto a PacBio Sequel II, IIe, or Revio system. HiFi sequencing typically runs for 15 to 30 hours depending on the polymerase kinetics. CLR runs can be shorter but produce lower quality bases. Monitor the run metrics such as polymerase read length and number of productive ZMWs.
3. Basecalling and Demultiplexing
The instrument output is raw signal data (h5 or pod5 files). Basecalling is performed by PacBio software (e.g., SMRT Link) or by open source tools like ccs for HiFi or lima for barcode demultiplexing. Always verify the yield and read length distribution.
4. Quality Control
Filter reads by predicted accuracy. For HiFi, keep only reads with QV >= 20 (99% accuracy). Remove short reads (< 1000 bp) as they tend to be low quality. Tools like FastQC and Nanoplot (originally for Oxford Nanopore) can be adapted for PacBio, but the Bioconductor package qckitfastq provides metrics specific to long reads.
5. Assembly or Analysis
Use an assembler designed for long reads. Popular options include hifiasm (for HiFi), Flye, and Canu. For structural variant detection, mapping with minimap2 followed by Sniffles or pbsv works well. For transcriptome analysis, IsoSeq with cupcake or SQANTI3 is standard.
A noteworthy application is the detection of rare structural variants using reduced representation approaches. The HiFiRE3 method uses restriction enzyme ends to reduce genome complexity while preserving structural variant detection at high fidelity source: High fidelity rare structural variant detection with HiFiRE3 reduced representation via restriction enzyme ends.
6. Polishing and Validation
If you used CLR reads, polish with a second round of short reads or with HiFi reads. Even HiFi assemblies may benefit from a final polishing pass with polish or DeepConsensus. Validate using paired end short reads aligned to the assembly to check for misassemblies. Tools like Stash can detect misassemblies using a data structure based on stochastic tile hashing source: Genome misassembly detection using Stash.
Quality Checks
A successful PacBio run should produce more than 10 Gb of Q20 bases per SMRT Cell 8M. The N50 read length for HiFi should be above 10 kb. Run ccs statistics to see the number of passes and the resulting QV. For assembly, check the contig N50 and the percentage of complete BUSCO genes. The Galaxy Training Network has workflows that include these metrics.
If you are working on a specific variant, such as STRC in hearing impairment, long read sequencing can resolve the repeat region, but you must validate with orthogonal methods. One study used maker mediated refinement of long read sequencing with MLPA validation source: Integrated analysis of STRC variants in hereditary hearing impairment using maker mediated refinement of long read sequencing with MLPA validation.
Common Mistakes
1. Insufficient DNA quality. Degraded DNA leads to short reads. Always check for shearing and use HMW extraction protocols. A single DNase contamination can ruin the library.
2. Overlooking depth requirements. For a bacterial genome, 50x HiFi may be overkill, but for a eukaryotic genome, 15x is often too low for complete contiguity. Check similar genomes in the NCBI SRA.
3. Using default parameters for every assembler. Each assembler has flags for genome size, heterozygosity, and expected coverage. Tune these parameters.
4. Ignoring the effects of repeats. Even with long reads, highly repetitive regions like telomeres and centromeres may not assemble fully. Sacrificing read length by shearing too short can make this worse.
5. Not performing polishing after CLR assembly. CLR assemblies have many systematic errors, polishing with HiFi or short reads is mandatory.
6. Forgetting to check for contamination. Microbial contamination in a sample can produce high quality but spurious assemblies. Always taxonomically assign contigs.
Limits of Interpretation
PacBio long reads cannot guarantee completeness in all repeats. Highly homopolymorphic stretches or regions with extreme GC bias (above 70% or below 30%) may still have errors. Base modification detection (e.g., methylation) requires specialized kinetic analysis and should not be treated as definitive without validation.
Structural variant detection recalls large insertions and deletions well, but identifying exact breakpoints in nested repeats remains uncertain. The study of Morganella morganii NM 11 from snake origin used PacBio to gain a near complete genome but still noted limitations in resolving plasmid fusions source: Morganella morganii NM 11 of snake origin.
Genetic interpretation of sequence variants, especially in non coding regions, requires additional functional evidence. Long read data alone does not replace clinical validation.
Frequently Asked Questions
1. Can I use PacBio for human whole genome sequencing? Yes. HiFi reads at 15 30x coverage produce a high quality assembly suitable for detecting structural variants and phasing haplotypes. The cost is higher than short read sequencing, but the added information on structural variation and repetitive regions can be essential for certain research questions.
2. How do PacBio and Oxford Nanopore compare? Both are long read platforms. PacBio HiFi offers higher base accuracy (99.9%) while Oxford Nanopore can produce reads longer than 2 Mb. PacBio is generally preferred for de novo assembly of eukaryotic genomes, while Nanopore is often used for extreme structural variant detection or real time sequencing.
3. Do I need a bioinformatics pipeline to process PacBio data? Yes. At minimum you need tools for basecalling, read QC, and assembly or alignment. The Bioconductor project provides packages for downstream analysis, and the Galaxy Training Network has ready to use workflows.
4. Can PacBio data detect copy number variations (CNVs)? It can, but with caution. Long reads make breakpoint detection more reliable, but CNV calling from depth of coverage is still influenced by mappability of repeats. Use specialized tools like pbsv or Sniffles that incorporate split read and discordant read pair information.
References and Further Reading
- NCBI Bookshelf: Bioinformatics and Computational Biology - authoritative textbook chapters on sequencing technologies.
- EMBL EBI Training: Long Read Sequencing - online courses and tutorials.
- Galaxy Training Network: PacBio data analysis - hands on workflows for QC, assembly, and variant calling.
- Bioconductor: Genomic Data Analysis - R packages for long read analysis.
- NCBI Sequence Read Archive - repository for PacBio datasets used in peer reviewed studies.
- Genome misassembly detection using Stash: A data structure based on stochastic tile hashing - method for validating genome assemblies.
- Integrated analysis of STRC variants in hereditary hearing impairment using maker mediated refinement of long read sequencing with MLPA validation - practical example of variant detection with validation.
- de novo assembly of 5 highly contiguous Heliconius butterfly genomes with long read sequencing alone - demonstrates what is achievable with only long reads.
- Assembly of a high quality reference genome for the rat tapeworm Hymenolepis diminuta - example of a single technology genome.
- High fidelity rare structural variant detection with HiFiRE3 reduced representation via restriction enzyme ends - technique for targeted long read sequencing.