[Part 2 of my summary of the AGBT 2011 meeting. Mainly about the various sequencing technologies.
Also featuring: I get to anoint the most partying corporation.
Part 1 here]
In contrast to last year’s two new exciting sequencing technology announcements, PacBio’s SMRT single molecule sequencing and Ion Torrent’s Personal Genome Machine (PGM), nothing groundbreaking was presented.
The closest that came to a potentially novel and readily utilizable technology was the BioNanomatrix system. The platform consists of a dense array of nanochannels (~100nm thick) and very long stretches of genomic DNA can be entropically driven through these channels in a linear manner. The flowing DNA can be either imaged with a camera (if fluorescently labeled) or electrically detected. This movie demonstrates the concept for a 400kb DNA. I am not sure how the platform can be used for de novo sequencing, but it could certainly be useful for investigating DNA structural variations or particularly for epigenetic mapping of long DNA stretches. One thing we did wonder about was how much this technology differs from the US Genomics approach of DNA imaged in microfluidic channels.
There were also lots of posters, and quite a few talks on sample preparation for various NGS platforms, both from the point of view of speed, as well as quality. This is one field I am not very familiar with and hence, little to write about In fact, I had never realized how important the role sample preparation played till I heard talks about genome assembly talks!
I did finally have the opportunity to see in person both the Ion Torrent PGM and the recently announced Illumina MiSeq machines. As I am not an end-user, it was the technology in each that was the cool factor. Both machines are pretty much plug-and-play units: add your DNA sample at one spot and out comes the result few hours later (real-time on your iPhone app if you want). As an added bonus, you can even charge your iPhone on a dock on the Ion Torrent PGM, making it the most expensive (~$60K) charger on market.
In case of the PGM, the sample preparation is an 8-hours process involving emulsion-PCR. MiSeq, on the other hand, uses a 20 minute step before sample loading. MiSeq also has the advantage of being similar to Illumina’s HiSeq – and hence can take advantage of similar chemistry, and thus, reagents. Thus, while the MiSeq seems like a scaled down Solexa technology, Ion Torrent uses a novel approach where the pH change occurring due to the proton release from nucleotide addition is detected. Consequently, the PGM is camera-free and their chip, where the sequencing reaction occurs, takes advantage of the semiconductor manufacturing industry. Hence their throughput is roughly expected to double every six months.
Of course, there are important differences between the two in terms of read lengths and coverage. PGM can sequence ~50-100bp (>99.5% accuracy for 50bp reads) but 10Mb coverage vs 35bp reads and >120 Mb for the MiSeq (the Torrent data was information I got from a company representative, and the MiSeq from their website).
While the PGM is now available for sale, it seemed to me – and feel free to correct me here – that people are still not sure on where and how to utilize the platform. Big sequencing centers like The Broad Institute (Chad Nasbaum from the institute presented posters and talks on validating both the PGM and PacBio) seem to be planning on using the instrument for QC of library preparations and validation of certain sequencing results. In terms of actual sequencing, the most likely applications could be for microbial genomes or targeted genes. I am assuming that various academic PIs are writing RO1 grants that include $60,000 for instrument purchase as we speak. Ion Torrent is also crowd-sourcing for developing methods that will enable improved read lengths, lower read times and even faster sample preparation times.
The other major player in current third generation sequencing systems is Pacific Biosciences’ SMRT (single-molecule, real time) sequencer. Though their machines are not available commercially, quite a bit of data was presented by early users and from their own research. I was not able to attend their lunch workshop but did receive some feedback from other attendees. The main issue with the platform currently seems to be the low accuracy (86% as opposed to >99% on all other current instruments). Apparently their own people admitted to as much, indicating that the PacBio data needs to be combined with other 2nd generation sequencing systems (454, SOLiD etc) for higher accuracy. However, they do posses some advantages – they can cover long stretches of DNA (>1000bp) and single molecule sequencing is extremely rapid. The quick sequencing of the Haiti cholera genome was provided as an example , and I have already written about the disease weathermap project.
Additionally, Jonas Korlach from PacBio presented data on detection of methylated bases and other DNA damages. It seems that the polymerases used in the sequencing pauses when it reaches modified bases and this longer base addition interval can be detected. By using further base and polymerase modifications, they have been able to elongate this interval further, thereby allowing for a more accurate detection of these bases. This could prove very useful for epigenetic or DNA damage studies on unamplified DNA.
The most exciting new sequencing method on the horizon and one that has potential to be a disruptive technology is nanopore sequencing (disclosure: I work for a company that is performing research in this area). The basic idea of the technology is to force the DNA through a small pore (on few nanometers scale) using a voltage bias (much like electrophoresis) and then reading the differential current blocks produced by each base as it passes through the pore. The advantages of these system include possible long reads, no requirement for amplification (therefore useful for detecting epigenetic modifications), and simpler electronic detection that requires no cameras. The lack of labeling and the electronic detection makes the system potentially cheap as well.
Unfortunately, Hagan Bayley of Oxford University, one of the leading researchers on nanopore, and whose technology is being developed through Oxford Nanopore, was unable to attend the meeting. This resulted in some shuffling of talks, with his collaborator Mark Akeson – supposed to speak earlier during a technology session – taking Bayley’s place in the last session. Jens Gundlach of Washington University stepped in to provide an additional presentation. Both Gundlach and Bayeley (and Akeson’s) group use protein nanopores for sequencing, but they use different proteins, each with certain advantages. Since Gundlach did not approve of blogging/tweeting of his talk (“I will be the only one tweeting here” was his comment), I cannot write about it. However, his group did publish a nice paper last year, where he describes some success with his sequencing method, albeit using a preliminary amplification step .
All of Akeson’s work presented at the meeting has been published last year as well [2, 3]. He has been successful in tackling one of the major issues of nanopore sequencing, i.e. arresting the rapid rate at which DNA passes through the pore thereby allowing resolution of each base. His group achieved this by ratcheting the DNA through the pore using a special polymerase. Akeson did also mention – in response to a question from the session chair Eric Green – that short sequencing with this method is only ‘months’ away!
As opposed to the protein channels used by these groups, NABSys utilizes solid-state nanopores. The major disadvantage of solid-state nanopores versus protein is that their sizes, and therefore current block when the DNA is flowing through, is variable (additionally, protein nanopores can be mutated and/or chemically modified to provide greater versatility). On the other hand, since solid-state pores are not biological entities, they tend to be more stable and are easier to transport and include in a machine. Also, Jon Oliver, presenting for the company, stated that they were having some luck in producing more uniformly sized solid-state nanopores, or at any rate, differently sized pores that could be calibrated accordingly.
In addition to using solid-state pores, NABSys’ approach differs in the sense that instead of performing direct strand sequencing (as Oxford and collaborators are attempting), they plan to detect distances between hybridized regions of very long stretches of DNA. Data was sketchy, but the company is performing a lot of algorithm development that will allow contigs of extremely varied sizes to be read and assembled by this technique.
I was also hoping to see the new GRIDIon platform from Oxford Nanopore in action. This is a generalized nodal platform for DNA sequencing or nanopore detection, announced just before the conference. Videos of the system look pretty cool, but unfortunately, there weren’t any demonstrations at the AGBT (there some reviews out there by Luke Jostins, Dan MacArthur and Kevin Davies)
(Though not directly AGBT related, it is worth in this context to read NHGRI’s very good summary of various nanopore sequencing approaches including some I have described here.)
A different technology that seems much more ready for prime-time is Life Technology’s quantum-dot based platform, codenamed ‘Star-light’. Life Tech’s Joe Beechem only had a poster about the method but I have heard him talk on this topic before at a University of California, San Diego seminar. Personally, I find this the most exciting new sequencing technology out there (other than nanopores, of course!). This is probably because it intersects my previous interests in fluorescence detection at single molecule level with my current sequencing research. Like PacBio’s SMRT, this system too performs real-time single molecule sequencing during polymerase synthesis, but works on a different principle.
The heart of the method is a quantum-dot (QD) conjugated polymerase, what they call a ‘nanosequencing engine’. Quantum dots are highly stable and bright fluorescent semiconductor particles on the nanometer scale with all kinds of desirable optical properties (stability, broad excitation range etc). The DNA strand with the QD-polymerase is attached to a glass coverslip though a universal adapter and imaged in a solution with the four nucleotides, each with a unique fluorescent groups emitting at a unique wavelength. The sequencing is by synthesis: As a nucleotide is added, the QD – excited by a laser light – transfers its exited state energy to the fluorophore on the nucleotide by a process called Forster Resonance Energy Transfer (FRET). Consequently, the fluorescence from the QD goes down (said to be quenched) and at the same time the fluorescence of the acceptors jumps up from the zero baseline (these fluorescence emission are well separated on the spectrum to be detected individually).
Due the nature of the FRET process which happens only at very short (2-10nm) distances, this decrease in QD fluorescence and increase in nucleotide fluorescence occurs only when they are close together. Therefore, background interference from other fluorescent nucleotides is absent. Since each nucleotide carries a different colored fluorophore, the bases can be called in real-time based on two simultaneous observations: QD emission quenching and observation of an acceptor emission corresponding to the particular base. After incorporation, the fluorescent moiety is cleaved off by the polymerase and the QD is un-quenched, ready for the next nucleotide. A major advantage of this system over the PacBio technology is that if the QD-polymerase conjugates fail after a while, it can be washed away and replaced with fresh conjugates. Additionally, the growing DNA strand can be denatured for re-sequencing and thereby reduce errors.
The neatest application for this technology is where a very long DNA (even in the range of 40-50kb) can be placed horizontally on the cover-slide, with several of the QD-nanosequencing engines performing parallel sequencing reaction on the strand!
There was no word on when Life Technology plans to launch an actual machine based on this concept, or what the price point, accuracy etc will be. As I have written before, Life Technology’s acquisition of Ion Torrent and its current ownership of the SOLiD technology makes it an interesting question on where they can fit in this quantum-dot based technology.
In summary, while no new sequencing technologies were released this year, there are a few exciting ones in the pipeline that could see light of day in the next few years. I should also point out that have not covered sequencing solution providers like Complete Genomics, and recently, Perkin-Elmer. Complete did have the one major announcement at this conference: the release of 60 fully sequenced human genomes for open-source use by researchers.
Other General Meeting Notes:
AGBT does not have the traditional vendor show. Instead, sponsors are allotted suites or small conference rooms in the hotel where they can display their products and services. It also provides them an opportunity to give away free stuff and host parties. I was fairly restrained (or so I thought), but it almost required an extra baggage on the trip home.
Speaking of parties, Thursday evening on the printed meeting agenda was designated as ‘Dinner on your Own’. This seems to be a euphemism for ‘finding the right party’. I found myself having dinner at the PacBio ‘Dinner and Movie’ event (will write later about the documentary they showcased at the dinner) and then at various parties hosted by Life Technology, Agilent and Caliper. Life Technology possibly had the best location, the hotel rooftop with views out to the ocean; and they served shrimp cocktails as h’dourves to boot. But their insistence on serving something called the “Torrentini – Passion for Life”(vodka+passion fruit pucker, get it?) was slightly off-putting to a martini purist like myself.
I am not sure if Caliper’s ‘All Night Long’ party lived up to its name, but the aged Glenmoraigne and the better than average beer selection was certainly welcome. However, when it comes to hospitality, one has to hand it to Agilent. They were possibly the most active in giving away swags and pretty much insisting that we don’t pass by without a drink at their suite every evening at the conference.
On this note, thanks to all the sponsors for the meals and drinks, and most importantly, also for enabling all the great science to be presented. And huge appreciation for the organizing committee too for pulling off the logistics of hosting 800 odd people at a single site; it was quite neat how they used big screens to project talks all through the humongous conference room such that everyone could attend the day sessions simultaneously.
 I. M. Derrington, T. Z. Butler, M. D. Collins, E. Manrao, M. Pavlenok, M. Niederweis, and J. H. Gundlach, “Nanopore DNA sequencing with MspA,” Proc Natl Acad Sci U S A, vol. 107, pp. 16060-5, Sep 14.
 K. R. Lieberman, G. M. Cherf, M. J. Doody, F. Olasagasti, Y. Kolodji, and M. Akeson, “Processive replication of single DNA molecules in a nanopore catalyzed by phi29 DNA polymerase,” J Am Chem Soc, vol. 132, pp. 17961-72, Dec 22.
 F. Olasagasti, K. R. Lieberman, S. Benner, G. M. Cherf, J. M. Dahl, D. W. Deamer, and M. Akeson, “Replication of individual DNA molecules under electronic control using a protein nanopore,” Nat Nanotechnol, vol. 5, pp. 798-806, Nov.
 C. S. Chin, J. Sorenson, J. B. Harris, W. P. Robins, R. C. Charles, R. R. Jean-Charles, J. Bullard, D. R. Webster, A. Kasarskis, P. Peluso, E. E. Paxinos, Y. Yamaichi, S. B. Calderwood, J. J. Mekalanos, E. E. Schadt, and M. K. Waldor, “The origin of the Haitian cholera outbreak strain,” N Engl J Med, vol. 364, pp. 33-42, Jan 6.