As predicted, Oxford Nanopore Technology (ONT) made the biggest splash at the recently concluded Advances in Genome Biology & Technology meeting. There may have been apprehensions that Oxford wouldn’t live up to the pre-conference hype, but a full house of meeting attendees – plus loads of people following on Twitter – were not disappointed. Following the Steve Jobs school of product launch (website down on the morning of the talk and then a new one up just after the announcement, Clive Brown making a dramatic pause before announcing ‘one more thing……’) ONT presented a novel nanopore sequencing technology that caused audible gasps in the audience, hijacked all other discussions for the rest of the meeting, and was immediately being hailed as a ‘game-changer’.
Enthusiastic approvals also streamed through the Twitter feeds, and equally glowing reviews came in the form of blog-posts  almost immediately (by those who had been given early access to the presentation). The paradigm-shifting promise of an USB-powered, almost palm sized device that can sequence billion bases of DNA on your laptop has captured everyone’s imagination, with main-stream media like New York Times covering the story. The announcement rapidly affected the stock prices of other players in the sequencing field as well (suited financial types attending the meeting were observed to be furiously texting as the talk was progressing).
Many blogs and news-sites (check the list  below) have extensively covered the various features of the two sequencing platforms announced by ONT – the GRIDIon and the MINIons, as well as how it might affect the sequencing landscape (though unfortunately in some cases, the posts sound like rehashing of marketing materials). So I won’t go into those details here.
However, I would like to go a little beyond the hype, briefly recapping the talk by Clive Brown to analyze some of the scientific breakthroughs made by ONT that enabled the technology at the heart of these sequencing devices.
[As an aside, my pre-announcement predictions of the technology was off on many of the details, though the general ideas were as expected. Of course, the MINIon totally came from the left field.]
Basically the platform involves a chip containing densely arrayed nanopores, the electrical current through each of which can be read separately. DNA can be added to the chips with an enzyme and sequencing is performed as the enzyme bound DNA is pulled to the pore. The only requirement for sequencing to work is that the DNA should have a 5’- overhang. However, there are number of other DNA forms that will work as well. The two sequencing runs demonstrated during the talk were performed with DNA where a hairpin is added at one end (possibly so that the DNA is arrested at the top of the pore and can be run through the pore in the reverse direction once more).
Zooming into the sensor itself, the protein nanopore being utilized by ONT is still α-hemolysin (αHL), but it is an engineered protein with several amino acids within the pore mutated to other residues to improve base-discriminating abilities.
Rather than using a polymerase to control the speed of the DNA (as I was thinking), they have developed a novel motor-enzyme for the ratcheting motion. However, they would not say which particular protein acts as the rachet other than that it is certainly not a polymerase. Not having a polymerase is actually useful since additional nucleotides do not need to be added to the solution and the DNA is available in the original form for re-sequencing if needed.
For obtaining both the optimal pore and the motor-enzyme, the ONT scientists had to screen hundreds of mutations to hit upon the perfect ones.
ONT has also done away with the traditional lipid membrane that the αHL protein is usually embedded in, and replaced it with a robust synthetic polymer. The protein-polymer combination is preloaded on the chips and is extremely stable with 80% of the pores still functional after three days. It also has the ability to withstand dirty samples like blood and sewage wate. This I believe is actually a major material science-biochemistry interface innovation. While αHL is a relatively stable protein on its own, if the synthetic polymer can be adapted to other proteins, it could be useful to protein arrays in general. But it is mainly the stability of this polymer membranes (and to some extent the electronics) that enables the disposable USB drive-sized MINIon sequencer.
The synthetic polyemer-protein interface is combined with their own low-noise integrated circuit to produce the dense array of protein pores, each an individual sequencing machine, on their chips.
For actual base-reads, ONT still has not achieved single-base sensitivity (though Brown did mention they are working on it). Instead they are reading three bases at a time, leading to 64 different current levels. They then apply the Viterbi algorithm – a probabilistic tool that can determine hidden states – to these levels to make base calls at each position.
Using this technology, ONT was able to sequence two smaller sized genomes – a phiX viral DNA (5kbase) and the lamda DNA (48kbase). In both cases, DNA was sequenced as a single linearized fragment. Each fragment was read twice, once in each direction. The error was found to be ~4%, and mainly caused due to the nature of the predictive nature of base-calling, and fluctuations in currents due to DNA vibrating in the pore. The scientists at ONT are working on further pore mutations to remove this noise.
Considering that many different groups have struggled for over 25 years to produce sequencing information using nanopores, the presentation of this data is without doubt a significant scientific landmark in this field. The scientific team at ONT deserve rich kudos for making it happen. It was also heartening to see David Deamer and Dan Branton, two people from the group that were the first to envision nanopore sequencing, take in the talk from the front row. They must have been incredibly excited.
[One should mention here that Jens Gundlach of University of Washington, Seattle, also had a poster at the meeting that showcased nanopore sequencing data – albeit on a much shorter, 20-30 basepairs, scale – using the MspA protein and a phi29 polymerase enzyme].
In summary, Oxford Nanopore seems to have solved the three key technical challenges faced by protein nanopore sequencing technology: controlling speed of DNA, fragility of the biological membranes, and the lack of sharp sensing zones. They have demonstrated proof-of-principle of their pore by sequencing the phiX and lambda DNA at a relatively (compared to say, PacBio) high accuracy. They have unveiled a conceptual design of devices that will contain a dense array of protein pores that will work in parallel to sequence DNA, which they say will be available to customers at very competitive pricing (for GRIDIon) or unprecedented portability (MINIon). Combined the potential of unprecedented long reads, possibly upto 1000kB, no expensive or time-consuming library preparation, and potential direct RNA reads and epigenetic detections, the technology is indeed a massive game changer.
Until ONT demonstrates actual sequencing of a more complicated genome (a microbial one at minimum), there will be a healthy degree of skepticism. The reaction from Jonathan Rothberg, inventor of the 454 and Ion Torrent sequencing technologies, comparing the MINIon to ‘cold fusion’ might be a bit extreme. However, a majority of scientists I spoke to at AGBT agreed that while ONT’s technology is very promising, the proof will come from real world usage of the devices. The exceptional promises made by Pacific Biosciences two years ago, which they have only partially delivered on, is on everyone’s mind.
According to this post, ONT is providing about a dozen institutes with machines for beta testing, so hopefully we will have some real data very soon. Additionally, given the $1000 price range, I expect quite a few labs all over the world will buy a MINIon for simply testing the technology.
Still, the lack of data or some scientific details at the talk is a bit bothersome. I did not exactly see data that demonstrates that 64 levels of currents are being detected. Additionally, I was not quite sure if the phiX or the lambda DNA was sequenced using just one pore or the actual sensor array. There was mention of rabbit blood and waste-water being added to the chip for sequencing. This is very impressive in that the protein-polymer was not affected, but did they obtain actual sequencing data from these experiments? I do not recollect seeing that. Finally, while the sequencing of phiX or lambda DNA was quite exceptional as mentioned, one will have to wait for real sequencing data on complicated genomes to find out base length reads, error rates etc.
To be fair, it was a very short talk (20 minutes). But perhaps they could have released more data at the poster. Or allowed people to download some early data from their website.
In addition to these concerns, it seems to me there are couple of issues with the sequencing methods as it stands.
Firstly, since DNA is not amplified or modified, there will be modified nucleotides that will have a different current level. For future sequencing with real genomes, this will undoubtedly add to the complexity of base calling since the number of current levels being detected will be much higher than 64.
Secondly, if the DNA is added without any preparation, varying DNA lengths could cause some issues. Shorter DNAs are more likely to be pulled to the pore due to faster diffusion leading to a bias in the sequencing. I expect there will have to be some sort of fragment sizing, and there are quite a few commercial instruments out there (e.g the Pippin technology from Sage that was on demo at the conference) that can do this.
As such, the next few months will be extremely interesting as more details and data emerge confirming if ONT have indeed found one of the Holy Grails of sequencing (and yes, Brown showed an image of the rabbit from the Python movie, very appropriate).
[The ONT announcement completely overshadowed some other interesting technology news at the meeting, including Ion Torrent’s Ion Proton machine, some new data on very long reads from Illumina on their machines, and two other prospective next-generation sequencing technologies from GnuBio and LAserGen. More unfortunately, it overshadowed some really interesting basic scientific talks presented there. I will try to get a brief review of those very soon.]
 Further reading:
a. Kevin Davies at BioIT World possibly has the most comprehensive description of both the individual sequencers and the complete systems being offered by ONT. A must read, since he offers much more details than I have on how the GRIDIons can be stacked to produce sequencing information at a faster pace, as well as details on the costs etc.
Also read Ellen Clark’s summary.
(do let me know if there were any other blogs/news articles about the Oxford Nanopore technology, and I will add a link here).