- Long-read and ultra-long-read sequencing are advanced DNA sequencing approaches that analyse long, continuous nucleic acid molecules rather than the short DNA fragments typically generated by conventional next-generation sequencing technologies. These approaches are important in modern bioinformatics because they provide extensive sequence context, making it possible to investigate complex genomic regions, repetitive sequences, structural variations, chromosome organization, and complete transcript structures. Long-read sequencing generally produces reads ranging from several thousand bases to tens of thousands of bases, while ultra-long-read sequencing focuses on exceptionally long DNA molecules that may extend beyond 100 kilobases and, in suitable experiments, reach several hundred kilobases or more. These categories are descriptive rather than universally standardized, and their boundaries vary among research fields and sequencing platforms.
- Conventional short-read sequencing technologies, such as Illumina sequencing, commonly produce reads of approximately 50–300 base pairs. Although short reads can provide high per-base accuracy and high sequencing throughput, they may be insufficient to resolve repetitive genomic regions, duplicated sequences, large structural variants, and complex rearrangements. When a sequence contains repeated regions longer than the reads themselves, it can be difficult to determine where individual fragments belong or how the repeated regions are connected. Long-read sequencing addresses this limitation by analysing much larger sections of the original DNA molecule, preserving information about the relationships among distant genomic features.
- The fundamental principle of long-read sequencing is the analysis of individual DNA molecules that are substantially longer than those typically used in short-read sequencing. The sequencing instrument generates a continuous sequence from each molecule, and computational methods use these reads to reconstruct genomic regions, identify variants, and investigate molecular structures. Two major technologies dominate modern long-read sequencing: Pacific Biosciences (PacBio) Single-Molecule Real-Time sequencing and Oxford Nanopore Technologies (ONT) sequencing. Both can generate long reads, but their sequencing principles and performance characteristics differ.
- PacBio sequencing uses Single-Molecule Real-Time technology, in which DNA polymerase activity is monitored through fluorescent signals as nucleotides are incorporated into a growing DNA strand. The DNA fragment is prepared as a circular SMRTbell template, allowing the polymerase to pass repeatedly around the same insert. Multiple observations of the original DNA sequence can then be combined through circular consensus sequencing (CCS) to generate highly accurate HiFi reads. PacBio HiFi reads commonly span approximately 10–25 kilobases, although the actual length depends on library preparation, insert size, sequencing chemistry, and instrument configuration. Their combination of long read length and high consensus accuracy makes them useful for genome assembly, small-variant detection, structural variant analysis, and haplotype phasing.
- PacBio has also produced Continuous Long Reads (CLR), which are based on polymerase observations of DNA templates without the same repeated-pass consensus strategy used for HiFi reads. CLR reads can provide long sequence information, but their individual-read accuracy is generally lower than that of modern HiFi reads. HiFi sequencing is now an important PacBio approach when both long read length and high base accuracy are required. Researchers should distinguish the different PacBio read types when selecting assembly and variant-analysis tools because their error profiles and analytical requirements differ.
- Oxford Nanopore sequencing uses a different principle. DNA or RNA molecules pass through nanopores embedded in a membrane, and changes in ionic current are measured as the nucleic acid moves through each pore. A motor protein regulates the movement of the molecule, allowing the electrical signal to be recorded and interpreted by basecalling software. Because the molecule passes directly through the pore, read length is largely determined by the length and integrity of the input nucleic acid rather than by a fixed sequencing-cycle limit. Oxford Nanopore sequencing can therefore generate ordinary long reads as well as ultra-long reads, including molecules extending beyond 100 kilobases.
- Ultra-long-read sequencing is particularly associated with Oxford Nanopore technology because exceptionally long DNA molecules can be sequenced directly when they remain intact during extraction and library preparation. Reads exceeding 100 kilobases can provide valuable information about long repetitive regions, segmental duplications, centromeres, telomeres, and complex structural rearrangements. In specialized experiments, reads may reach hundreds of kilobases or even longer. However, obtaining such reads requires careful preservation of high-molecular-weight DNA. Mechanical shearing, vigorous pipetting, repeated mixing, harsh extraction procedures, and other forms of physical damage can break long DNA molecules and reduce the proportion of ultra-long reads.
- The preparation of long-read sequencing libraries begins with the extraction of high-quality DNA or RNA. For genomic applications, high-molecular-weight DNA is especially important because the maximum possible read length depends strongly on the length of the original DNA fragments. Researchers commonly evaluate DNA concentration, purity, integrity, and fragment-size distribution before library preparation. Fluorometric methods can be used to quantify DNA, while electrophoresis-based methods can help assess fragment size and integrity. The extraction procedure must be selected according to the biological sample, because tissues, blood, microbial cultures, plants, and environmental samples may require different approaches.
- For ultra-long-read experiments, DNA extraction procedures are designed to minimize fragmentation. Gentle cell lysis, careful handling, suitable purification methods, and reduced mechanical stress help preserve long DNA molecules. The use of specialized high-molecular-weight DNA extraction protocols may be necessary for samples with large and complex genomes. Researchers should also avoid unnecessary freeze–thaw cycles and excessive sample manipulation. The quality of the extracted DNA can have a greater influence on the production of ultra-long reads than the nominal maximum read length supported by the sequencing instrument.
- Library preparation differs between PacBio and Oxford Nanopore sequencing. PacBio workflows convert DNA fragments into circular SMRTbell templates by attaching specialized adapters to both ends of the DNA. These templates are then prepared for polymerase-based sequencing. For HiFi sequencing, the insert length and number of polymerase passes influence the final consensus-read length and accuracy. Oxford Nanopore workflows attach sequencing adapters and motor proteins to DNA molecules, allowing them to pass through nanopores. Some library preparation methods are designed to preserve very long DNA fragments, while others prioritize throughput, rapid preparation, or particular sample types.
- Sequencing is performed using specialized instruments and consumables. PacBio systems, including platforms such as Revio and Vega, use SMRT Cells containing arrays of zero-mode waveguides. Oxford Nanopore provides platforms such as MinION, GridION, and PromethION, which use flow cells containing arrays of nanopores. The appropriate platform depends on the required throughput, sample number, genome size, desired read length, and available resources. Platform specifications and supported chemistries change over time, so experimental planning should use current manufacturer documentation.
- After sequencing, the resulting reads undergo computational processing. PacBio HiFi data are generated through consensus processing, while Oxford Nanopore data are produced by basecalling the electrical signals recorded during sequencing. Oxford Nanopore raw signal files can be stored in POD5 format, and basecalled sequences are commonly stored in FASTQ format. PacBio data workflows may include BAM files containing sequencing reads and associated information. Downstream analysis can use standard alignment formats such as SAM, BAM, and CRAM, together with reference-based or de novo analysis tools.
- Quality control is essential for both long-read and ultra-long-read sequencing. Important measurements include total sequencing yield, number of reads, read-length distribution, mean and median read length, maximum read length, read quality, and coverage. For ultra-long-read experiments, researchers should pay particular attention to the proportion of reads exceeding selected length thresholds, such as 50 kilobases or 100 kilobases. A dataset may contain a small number of extremely long reads but still have insufficient total coverage for reliable genome assembly. Therefore, read length should always be evaluated together with sequencing depth and overall data quality.
- Several bioinformatics tools support long-read quality assessment. NanoPlot and NanoStat can summarize Oxford Nanopore read lengths and quality distributions. pycoQC can provide sequencing quality reports for compatible nanopore data. MultiQC can combine reports from multiple analysis tools into a consolidated summary. For PacBio data, SMRT Link and associated PacBio tools provide information about sequencing yield, read quality, polymerase performance, and HiFi read generation. These reports help identify problems associated with DNA fragmentation, low sequencing yield, poor-quality reads, and library preparation.
- A major application of long-read sequencing is de novo genome assembly. De novo assembly reconstructs a genome from sequencing reads without requiring an existing reference genome. Long reads provide extended overlaps and connections among genomic regions, helping assembly algorithms resolve repetitive sequences and construct longer contigs. This is particularly valuable for organisms with large genomes, extensive repetitive DNA, duplicated genes, or complex chromosome structures. Long-read sequencing has improved the assembly of human, animal, plant, fungal, bacterial, and archaeal genomes.
- Ultra-long reads can provide additional advantages during genome assembly because they may span entire repetitive regions and connect unique sequences located on either side of those regions. For example, a repeat that is longer than ordinary long reads may remain unresolved, whereas an ultra-long read may extend across the entire repeat and provide evidence about its correct genomic position. This can improve assembly continuity and reduce ambiguity in complex genomic regions. However, ultra-long reads do not automatically produce complete or error-free assemblies. Adequate coverage, reliable basecalling, appropriate assembly algorithms, and additional evidence may still be required.
- Several genome assemblers are available for long-read data. Flye supports long-read genome assembly and is used with suitable PacBio and Oxford Nanopore datasets. Canu is designed for long-read assembly and includes correction and assembly procedures appropriate to supported data types. hifiasm is widely used for PacBio HiFi genome assembly and can produce phased assemblies in suitable diploid datasets. Shasta is another assembler developed for long-read sequencing data. The choice of assembler depends on read type, accuracy, genome size, ploidy, coverage, and the biological objective. Researchers should follow the recommended input settings for each program rather than assuming that all long-read datasets can be processed identically.
- A simplified Oxford Nanopore assembly workflow may use the following commands:
# Assemble long reads using Flye
flye –nano-hq reads.fastq \
–out-dir flye_assembly \
–threads 16 \
–genome-size 3g
- The –nano-hq option is intended for suitable higher-quality nanopore reads. Other Flye input options may be more appropriate for different read types and quality levels. The genome-size parameter should be adjusted to the expected genome size of the organism. Assembly parameters should be selected according to the software version and sequencing data.
- PacBio HiFi reads can be assembled using hifiasm:
# Assemble PacBio HiFi reads
hifiasm -o genome.asm \
-t 16 \
reads.hifi.fastq
- The resulting assembly graphs and sequence files can be processed and evaluated further. For diploid organisms, suitable hifiasm options and downstream procedures can be used to investigate haplotype-resolved assemblies. The exact workflow depends on the organism, ploidy, sequencing coverage, and whether phased or primary assemblies are required.
- After assembly, quality assessment is performed to evaluate completeness, continuity, accuracy, and possible contamination. QUAST can calculate assembly statistics such as contig number, total assembly length, N50, and other measures. BUSCO evaluates the presence of conserved single-copy orthologs expected for a particular taxonomic group. Read mapping can help identify unsupported assembly regions, coverage inconsistencies, and possible misassemblies. For microbial genomes, tools such as CheckM2 can help assess genome completeness and contamination. No single metric is sufficient to establish assembly quality, so several independent measures should be considered.
- Long-read sequencing is particularly useful for detecting structural variants. Structural variants include insertions, deletions, inversions, duplications, translocations, and complex genomic rearrangements. Long reads can span variant breakpoints and provide information about the surrounding sequence, making it easier to identify and characterize events that may be difficult to detect with short reads. Tools such as Sniffles2, cuteSV, and SVIM can identify structural variants from suitable long-read alignments. PacBio and Oxford Nanopore data may require different calling parameters and quality filters, so the selected workflow should be appropriate for the sequencing technology.
- Long reads also support the detection of small genetic variants, including single-nucleotide variants and short insertions or deletions. PacBio HiFi reads are particularly useful for accurate small-variant analysis because their consensus accuracy is high. Oxford Nanopore small-variant calling has also improved through advances in sequencing chemistry and basecalling models. Tools such as DeepVariant and Clair3 are used in suitable long-read variant-calling workflows, although their supported data types and recommended configurations should be checked. Variant interpretation should consider sequencing depth, mapping quality, sequence context, allele balance, and potential technical errors.
- Haplotype phasing is another important application. In diploid organisms, two copies of each chromosome are inherited, one from each parent. Haplotype phasing determines which variants belong to each chromosome copy. Because long reads can span multiple heterozygous positions, they provide direct evidence about the physical relationships among variants. Tools such as WhatsHap can use long-read alignments to phase variants and reconstruct haplotypes. Haplotype-resolved assemblies are particularly useful in human genetics, plant and animal breeding, population genomics, and studies of allele-specific gene expression.
- Long-read sequencing has also transformed transcriptome analysis. Conventional short-read RNA sequencing often fragments transcripts before sequencing, requiring computational reconstruction of transcript structures. Long-read transcript sequencing can cover complete or near-complete RNA molecules, helping identify alternative splicing, transcript isoforms, alternative transcription start sites, and alternative polyadenylation. PacBio Iso-Seq uses circular consensus sequencing to produce accurate full-length transcript reads, while Oxford Nanopore supports direct RNA sequencing and complementary-DNA-based long-read workflows. Tools such as FLAIR, TALON, StringTie2, and Bambu can support long-read transcriptome analysis, depending on the data and research objective.
- Ultra-long reads are especially useful for resolving centromeres, telomeres, large tandem repeats, segmental duplications, and other difficult genomic regions. Centromeres contain repetitive DNA sequences that can be challenging to reconstruct using short reads. Telomeres consist of repetitive sequences at chromosome ends and are important for chromosome stability. Ultra-long reads can span larger portions of these regions and help establish their sequence organization. They have also contributed to the reconstruction of previously incomplete regions of complex reference genomes.
- In cancer genomics, long-read sequencing can identify complex rearrangements, gene fusions, copy-number changes, and structural alterations associated with tumour development. Some cancer genomes contain multiple rearrangements that are difficult to reconstruct from short-read data because the relevant breakpoints are separated by large distances or occur within repetitive regions. Long reads can span these breakpoints and provide more complete structural information. They can also support haplotype analysis and investigation of allele-specific changes. However, tumour purity, sample heterogeneity, sequencing coverage, and the presence of subclonal variants must be considered during interpretation.
- In microbial genomics and metagenomics, long-read sequencing can improve the reconstruction of complete microbial genomes and plasmids. It can help connect antimicrobial resistance genes, virulence factors, and mobile genetic elements to the organisms carrying them. Ultra-long reads can span complex plasmids and repeated genomic regions that may be difficult to assemble using short reads. Tools such as Kraken2 and other taxonomic classifiers can support the identification of microbial sequences, while long-read assemblers and metagenomic workflows can reconstruct microbial genomes from complex samples. Reference database quality and contamination control remain important for reliable classification.
- Long-read sequencing is also valuable in plant and animal genomics. Many plant genomes contain large amounts of repetitive DNA, polyploid chromosome structures, and duplicated genomic regions. Animal genomes may contain structural variation, complex gene families, and haplotypes that are difficult to resolve using short reads. Long-read data can improve reference genome construction, gene annotation, comparative genomics, and the identification of genetic variation. In livestock and crop improvement, these approaches can support studies of production traits, disease resistance, fertility, adaptation, and genetic diversity.
- A further application is the analysis of DNA methylation and other epigenetic modifications. PacBio SMRT sequencing can detect certain DNA modifications through changes in polymerase kinetics, while Oxford Nanopore sequencing can identify modified bases through their effects on electrical signals. These methods can provide information about DNA modifications without conventional chemical conversion procedures in suitable workflows. Modified-base detection depends on the sequencing platform, chemistry, computational model, sequence context, and validation strategy. Researchers should distinguish direct signal-based evidence from computational predictions and use appropriate controls.
- Long-read sequencing can be combined with short-read sequencing to improve the quality and reliability of genomic analysis. In hybrid sequencing, long reads provide continuity and structural information, while short reads can provide additional high-accuracy evidence for polishing or validation. PacBio HiFi data may already provide sufficient per-read accuracy for many applications, reducing the need for extensive short-read correction. Oxford Nanopore assemblies may benefit from suitable polishing workflows depending on the chemistry and data quality. Other technologies, such as Hi-C sequencing and optical mapping, can provide long-range information for chromosome-scale scaffolding and structural validation.
- The choice between long-read and ultra-long-read sequencing depends on the biological question. PacBio HiFi sequencing is particularly useful when high per-base accuracy and reliable variant detection are important. Oxford Nanopore long-read sequencing is useful when flexible read lengths, real-time analysis, direct RNA sequencing, or modified-base detection are required. Ultra-long Oxford Nanopore reads are especially valuable when the objective is to span very large repetitive regions or complex structural features. In some genome projects, combining PacBio HiFi reads with Oxford Nanopore ultra-long reads provides complementary information: HiFi reads contribute high sequence accuracy, while ultra-long reads help resolve long-range genomic structures.
- Long-read sequencing also has limitations. The generation of very long reads requires high-quality, minimally fragmented nucleic acids, which may be difficult to obtain from degraded samples, formalin-fixed tissues, or some environmental materials. Ultra-long-read experiments can produce a smaller number of exceptionally long reads than ordinary long-read experiments, and their overall coverage may be insufficient if sequencing yield is not carefully planned. Some long-read technologies may have higher per-base error rates than short-read sequencing or PacBio HiFi sequencing, although performance varies with chemistry and computational processing. Data storage, basecalling, alignment, assembly, and variant analysis can also require substantial computational resources.
- For reproducible long-read analysis, researchers should document the sequencing platform, instrument, chemistry, library preparation protocol, DNA extraction method, sample quality, and sequencing coverage. Read-length distributions and quality-control reports should be preserved. The software versions and parameters used for basecalling, read filtering, alignment, assembly, polishing, variant calling, and phasing should be recorded. Reference genome versions, annotation releases, and quality thresholds should also be documented. Original raw data and intermediate files should be retained whenever possible so that the analysis can be reprocessed as computational methods improve.
- Long-read and ultra-long-read sequencing have become essential approaches in modern bioinformatics because they provide genomic information that is difficult to obtain using conventional short-read methods. Long reads improve genome assembly, variant detection, haplotype phasing, transcript characterization, and the investigation of complex genomic regions. Ultra-long reads extend these capabilities by spanning exceptionally large repetitive sequences and structural features, supporting more complete genome reconstruction. Understanding the differences between PacBio HiFi, PacBio CLR, Oxford Nanopore long reads, and Oxford Nanopore ultra-long reads is important for selecting the appropriate sequencing strategy and interpreting genomic data accurately.