PacBio vs. Oxford Nanopore sequencing

Long-read sequencing developed by Pacific Biosciences and Oxford Nanopore overcome many of the limitations researchers face with short reads. Long reads improve de novo assembly, transcriptome analysis (gene isoform identification) and play an important role in the field of metagenomics. Longer reads are also useful when assembling genomes that include large stretches of repetitive regions.

Currently, there are two long read sequencing platforms. To help a researcher choose between which platform has greater utility for their application, we compare overall instrument specifications offered by PacBio and Oxford Nanopore, and published applications in the next-generation sequencing space.

Capturea Oxford Nanopore charges an access fee that gives users one MinION/PromethIon instrument, a starter pack of consumables, certain data services, and community-based support

* Insufficient data

Although both PacBio and Oxford Nanopore generate longer reads compared to short read Illumina or Ion sequencing, the higher error rate of both the PacBio and Oxford Nanopore sequencers remain an issue needs addressing. Whereas PacBio reads a molecule multiple times to generate high-quality consensus data, Oxford Nanopore can only sequence a molecule twice. As a result, PacBio generates data with lower error rates compared to Oxford Nanopore. PacBio has a slightly better overall performance for applications such as the discovery of transcriptome complexity and sensitive identification of isoforms. On the other hand, MinION provides higher throughput as nanopores can sequence multiple molecules simultaneously. Hence, it is best suited for applications that require a larger amount of data9

As long reads can provide large scaffolds, de novo assembly is one of the main applications of PacBio sequencing5. Though the error rate of PacBio data is higher than that of short read Illumina or Ion sequencing, increased coverage or hybrid sequencing can greatly improve the accuracy of genome assembly. PacBio sequencing has been successfully used to finish the 100-contig draft genome of Clostridium autoethanogenum DSM 10061, a Class III, the most complex genome classification in terms of repeat content and repeat type. It has a 31.1% GC content and contains repeats, prophage, and nine copies of rRNA gene operons. Using a single PacBio library and sequencing it with two SMRT cells, an entire genome can be assembled de novo with a single contig. When short read Illumina or Ion sequencing was used alone with the same genome, >22 contigs were needed, and each of the assemblies contained at least four collapsed repeat regions, PacBio assemblies had none10.

PacBio sequencing has also been used to assemble the chloroplast genome of Potentilla micrantha11, Saccharomyces cerevisiae, Aradopsis thaliana and Drosophila melanogaster using fewer contigs and CPU time for assembly compared to assemblies using Illumina sequencers12.

PacBio sequencing of PCR products can be used to improve the quality of current draft genomes by closing gaps and sequencing through hairpin structures and areas of high GC content more efficiently than Sanger sequencing13.

Pacific Biosciences has developed a protocol, Iso-Seq, for transcript sequencing. This includes library construction, size selection, sequencing data collection, and data processing. Iso-Seq allows direct sequencing of transcripts up to 10 kb without the use of a reference genome. Iso-Seq has been used to characterize alternative splicing events involved in the formation of blood cellular components14. This is essential for interpreting the effects of mutations leading to inherited disorders and blood cancers, and can be applied to design strategies to advance transplantation and regenerative medicine.

Another major application of PacBio sequencing is in epigenetics research. Recent studies demonstrate that investigation of intercellular heterogeneity in previously undetectable genome DNA modifications (such as m6A and m4C) is facilitated by the direct detection of modifications in single molecules by PacBio sequencing15.

Compared to PacBio, the Oxford Nanopore MinION is small (size of a USB thumb drive), affordable, utilizes a simple library prep and is field portable16. This is useful in situations such as a virus outbreak where a mobile diagnostic laboratory can be set up using MinIONS. In remote regions such as parts of Brazil and Africa where there are logistical issues associated with shipping samples for sequencing, MinION can provide immediate and real-time data to scientific investigators. The most notable clinical use of MinION has been the analysis of Ebola samples on-site during the viral outbreak in West Africa17,18.

The low cost of sequencing and portability of the MinION sequencer also make it a useful tool for teaching. It has been used to provide hands-on experience to students, most recently at Columbia University and the University of California Santa Cruz, where every student performed their own MinION sequencing19.

Perhaps the most ambitious MinION application is its potential to detect and identify bacteria and viruses on manned space flights. In a proof-of-concept experiment, Castro-Wallace et al. demonstrated successful sequencing and de novo assembly of a lambda phage genome, an E. coli genome, and a mouse mitochondrial genome. They observed that there was no significant difference in the quality of sequence data generated on the International Space Station and in control experiments that were performed in parallel on Earth22.

Recently, Oxford Nanopore developed a bench-top instrument, PromethION, that provides high-throughput sequencing and is modular in design. It contains 48 flow cells that can be run individually or in parallel. The PromethION flow cells contain 3000 channels each, and produce up to 40 Gb of data.

 

References:

  1. Pacific Biosciences – AllSeq. Available at: http://allseq.com/knowledge-bank/sequencing-platforms/pacific-biosciences/.
  2. Jain, M., Olsen, H. E., Paten, B. & Akeson, M. The Oxford Nanopore MinION: delivery of nanopore sequencing to the genomics community. Genome Biol. 17, 239 (2016).
  3. Lu, H., Giordano, F. & Ning, Z. Oxford Nanopore MinION Sequencing and Genome Assembly. Genomics. Proteomics Bioinformatics 14, 265–279 (2016).
  4. Jain, M. et al. Nanopore sequencing and assembly of a human genome with ultra-long reads. bioRxiv (2017).
  5. Jain, M. et al. MinION Analysis and Reference Consortium: Phase 2 data release and analysis of R9.0 chemistry [version 1; referees: awaiting peer review]. F1000Research 6, (2017).
  6. Rhoads, A. & Au, K. F. PacBio Sequencing and Its Applications. Genomics, Proteomics Bioinforma. 13, 278–289 (2015).
  7. MinION. Available at: https://nanoporetech.com/products/minion.
  8. PromethION Early Access Programme. Available at: https://nanoporetech.com/community/promethion-early-access-programme.
  9. Oxford Nanopore in 2016. Available at: http://blog.booleanbiotech.com/nanopore_2016.html.
  10. Weirather, J. L. et al. Comprehensive comparison of Pacific Biosciences and Oxford Nanopore Technologies and their applications to transcriptome analysis. F1000Research 6, 100 (2017).
  11. Brown, S. D. et al. Comparison of single-molecule sequencing and hybrid approaches for finishing the genome of Clostridium autoethanogenum and analysis of CRISPR systems in industrial relevant Clostridia. Biotechnol. Biofuels 7, 40 (2014).
  12. Ferrarini, M. et al. An evaluation of the PacBio RS platform for sequencing and de novo assembly of a chloroplast genome. BMC Genomics 14, 670 (2013).
  13. Berlin, K. et al. Assembling large genomes with single-molecule sequencing and locality-sensitive hashing. Nat Biotech 33, 623–630 (2015).
  14. Zhang, X. et al. Improving genome assemblies by sequencing PCR products with PacBio. Biotechniques 53, 61–62 (2012).
  15. Chen, L. et al. Transcriptional diversity during lineage commitment of human blood progenitors. Science (80-. ). 345, (2014).
  16. Feng, Z., Li, J., Zhang, J.-R. & Zhang, X. qDNAmod: a statistical model-based tool to reveal intercellular heterogeneity of DNA modification from SMRT sequencing data. Nucleic Acids Res. 42, 13488–13499 (2014).
  17. Jain, M., Olsen, H. E., Paten, B. & Akeson, M. Erratum to: The Oxford Nanopore MinION: delivery of nanopore sequencing to the genomics community. Genome Biol. 17, 256 (2016).
  18. Quick, J. et al. Real-time, portable genome sequencing for Ebola surveillance. Nature 530, 228–232 (2016).
  19. Hoenen, T. et al. Nanopore sequencing as a rapidly deployable Ebola outbreak tool. Emerg. Infect. Dis. 22, 331–334 (2016).
  20. Citizen Sequencers: Taking Oxford Nanopore’s MinION to the Classroom and Beyond – Bio-IT World. Available at: http://www.bio-itworld.com/2015/12/9/citizen-sequencers-taking-oxford-nanopores-minion-classroom-beyond.html.
  21. Castro-Wallace, S. L. et al. Nanopore DNA Sequencing and Genome Assembly on the International Space Station. bioRxiv (2016).

12 thoughts on “PacBio vs. Oxford Nanopore sequencing

  1. There is actually no good reason that quality and length would be different between PromethION and MinION, since the underlying technology is the same. As far as I know, there is also insufficient PromethION data available to make statements about the performance. Your quality score of MinION reads is also rather outdated, see for example https://f1000research.com/articles/6-760/v1 for a more recent report (median identity 89%), although that one too doesn’t discuss the latest chemistry.

    Like

    • Thanks for your feedback. Earlier I had added estimated specifications for PromethION as mentioned in other blog posts. I have since then updated my post so as not to make assumptions about the PromethION specs.
      The MinION specs have also been updated.

      Like

  2. Seems like folks were getting closer to 2-3 GB per minion flow cell over 48 hours, from recent publications (e.g. http://biorxiv.org/content/early/2017/04/20/128835). On a GB per hour bases that is not higher throughput than Sequel (nor even RSII). And the max consensus error rates at reasonable coverage would be much more interesting to compare than raw error rates. That paper showed ~Q23 with nanopore alone or ~Q33 with hybrid illumina polishing. PacBio regularly gets > Q50.

    Like

  3. Seems like folks were getting closer to 2-3 GB per minion flow cell over 48 hours, from recent publications (e.g. http://biorxiv.org/content/early/2017/04/20/128835). On a GB per hour bases that is not higher throughput than Sequel (nor even RSII). And the max consensus error rates at reasonable coverage would be much more interesting to compare than raw error rates. That paper showed ~Q23 with nanopore alone or ~Q33 with hybrid illumina polishing. PacBio regularly gets > Q50.

    Like

  4. The data in this article is way out of date already.
    Glaring errors (there are many more)

    1. Promethion is not really available as in an early access program and flowcells have been majorly delayed. First results have been mentioned recently on twitter by one group only to my knowledge. Specs are as such absolute speculation.

    2. The Minion has an error rate of around 10 % (5-15%) at the moment, definitely not 35%. That was in 2014. See the recent london calling conference talks for more.

    3. The Minion average read length depends on the fragment size, reads of ~900kbp have been generated in early 2017 (see Nick Lomans blog).

    4. Pacbio promised those specs for the Sequel but are certainly not delivering yet.

    Like

    • Hi Colin,

      Thanks for your feedback.
      I obtained the information about Promethion early access program through the Oxford Nanopore website – https://nanoporetech.com/community/promethion-early-access-programme.
      Regarding the other specs being outdated, you are right. I had earlier cited published research articles from 2012-2016. Thanks for bringing the new data to my attention.
      I have updated the blog post with these specs for MinION and cited the most recent articles.

      Like

  5. Yikes. All your numbers for ONT are off. Granted, that platform has such flexibility and variability that it is hard to nail down numbers, but these really aren’t right.
    Maximum reported ONT read length is just shy of 1Mb now.
    Error rate with 1D^2 chemistry is reported to be less than 5%
    Output on the current flowcells is up to 15Gb reported, though 5Gb is more typical. This is probably your best number, though some of the runs in that publication were stopped early and if you didn’t filter those out it will skew your numbers downwards.
    Millions of reads are pretty routine at this point – but it depends on library insert size distribution
    Run price of $300 is only with the cheapest GridION pricing; $500 is the lowest MinION flowcell pricing and that doesn’t include library preparation reagents

    Like

    • Hi Keith,
      Thanks for your feedback. As you rightly said, it’s hard to nail down the exact numbers with Oxford Nanopore due to its flexibility. I have updated the blog post with the latest numbers after looking up some sources based on your suggestions. Hope that the specs are accurate now.

      Like

  6. Raw accuracy doesn’t tell you much, really. It’s more misleading than useful. I would think **consensus accuracy** possible from a typical amount of coverage is a much more useful metric. And there are many papers that characterize that. PacBio consensus is routinely > Q55 without even resorting to CCS reads. ONT has been working hard to improve consensus accuracy for some time, and in one of your citations was up to about Q23 (without adding Illumina data for polishing.) Perhaps that has improved a bit further recently? Does anyone have publications showing a higher ONT consensus accuracy with a more recent chemistry?

    Like

Leave a comment