By Kevin Davies
March 1, 2010 |
MARCO ISLAND, FL – In the end, Jonathan Rothberg stole the show.
With his customary quirkiness, passion and theatricality, and paying tribute to his heroes Jim Watson, Gordon Moore and Steve Jobs, the founder of Ion Torrent Systems closed last week’s superb Advances in Genome Biology and Technology conference by presenting the first technology that goes beyond next-generation sequencing, with its reliance on light and optics and lasers, into semiconductor sequencing.
In a ploy that was either cute or contrived, depending on one’s point of view, a pair of Ion Torrent staff carried in the prototype “Personal Genome Machine,” which had literally been sequencing DNA in a nearby hotel suite, early on during Rothberg’s presentation.
Whatever it’s called, Rothberg insisted to Bio-IT World it was not “next-generation sequencing.” Said Rothberg: “In ‘next-gen,’ the box was the machine. But now, the substrate is the machine.” Next-gen sequencing was empowering, but only for a few. Likening next-gen technology to the minicomputer, Rothberg said his ultimate goal was to develop the sequencer equivalent of the personal computer. He credited Broad Institute sequencing director Chad Nussbaum with coining the term: “post-light sequencing.”
Indeed, at the heart of the Ion Torrent instrument is a microfabricated chip, as sort made in semiconductor foundries around the world. The Ion Torrent sequencer initially uses chips containing 1.4 million sub-microscopic wells.
Details of the sample prep were not disclosed, although Rothberg said the system could use any DNA libraries prepared for any other commercial second-generation system. The instrument is small and light, weighing in at less than 50 pounds, involving no cameras or microscopes or lights. Because no imaging is required, the location of the wells is fixed and known, and the data are readily compressed, the demands on data storage and management are greatly reduced. Instead of a mini-supercomputer or blade center, the accompanying IT is a little more than a laptop.
As the sequencing method cycles in nucleotides, much like the 454 Life Sciences platform Rothberg conceived more than a decade earlier. But instead of detecting the pyrophosphate released upon incorporation, Ion Torrent directly measures the pH change caused by the release of a positively-charged hydrogen ion during each cycle. The pH change puts a charge on a sensing layer directly under the wells, which is turned into a voltage. Another name for the instrument, said Rothberg, is “the world’s smallest solid-state pH meter.”
If two identical bases are incorporated, then the chip detects a doubling of the pH change. That linear response is accurate for runs of at least six identical bases, and Rothberg says it is important to get to accurately call eight consecutive polymeric bases. Beyond that, the repeat is probably not biologically relevant, he asserted.
The machine reads each point and converts the pH into a voltage at 60x/second, so during a nucleotide incorporation, which Rothberg says takes 4 seconds, 240 data points are produced. “You never miss incorporation.” Another advantage is the Watson-Crick basepairing. Unlike other systems that measure DNA synthesis with modified bases and engineered or modified enzymes, Rothberg said he decided not to fight one billion years of evolution. Ion Torrent uses natural unmodified bases and polymerase.
It’s the simplest sequencing in the world,” says Rothberg. “One well, one sensor, one read.” (Rothberg has licensed other intellectual property in the semiconductor space, including technology emanating from Ron Davis’ lab at Stanford that focused on the negative charge added to the DNA rather than the hydrogen ion by-product.)
The result is a benchtop machine that, by comparison with the marvel of engineering that is the $695,000 PacBio RS machine that debuted 24 hours earlier, looks a bit like a child’s toy. But that’s sort of the point. Rothberg sees this machine, which will be priced at one tenth of any high-throughput next-gen sequencer – around $50,000 -- being used for multiple applications in research and clinical facilities around the world. “You can sequence on the back of a donkey if you want to,” said Rothberg.
As evidence of that international ambition, the tubes containing the four nucleotide solutions are labeled with rudimentary symbols – a cross, a plus, a square and a circle (closed symbols for purines are closed symbols, pyrimidines are open) – rather than the standard letters. “Our users are not going to know what As, Cs, Ts and Gs are,” said Rothberg somewhat debatably.
Currently, a run takes about an hour, during which time Rothberg claims it can read about 100-200 bases, with good quality sequence available from about half the wells on the chip. Regardless of the initial performance, there is a considerable upside. In principle, it takes no longer to produce 500 million reads than 5 million on a given chip, simply by engineering more wells on the chip.
The Personal Genome Machine will officially launch later this year, once launch specs are settled. The raw throughput will not come close to matching the existing next-gen instruments on the market, but Rothberg stressed, “We’re not competing with [the Illumina] HiSeq, etc. We’re when you want to get a couple of exons or a transcriptome.” Other potential applications include identifying structural variations, for example detecting rearrangements in a cancer sample; or rapid sequencing of a microbial genome, 100 genes of interest; or metagenomics.
“This chip is not for whole genome sequencing of humans,” stressed Rothberg. “This chip is for whole-genome sequencing of bacteria. This is for expression.” Rothberg presented preliminary data on a pair of microbial genomes, conducted with the Broad Institute and Stanford, and assembled with the aid of DNAStar.
Rothberg is so convinced that his personal genome machine will stimulate unanticipated applications that Ion Torrent will give away two machines for the most novel or exciting applications it receives in the next few weeks. “We’re going to have this at enough places that people will come up with amazing applications. You didn’t think about the apps on your iPhone [when it launched]. We want applications that change behavior.”
The winners will be announced in April.
This one goes to 1000
“Right now, we’re sequencing 100-200 bases,” says Rothberg, but “It’s just the beginning… Since we have highest signal-to-noise ratio and natural bases … there’s no reason it couldn’t go to 1000 bases. But that takes a lot of empirical work.”
Assuming Moore’s Law applies to the performance of semiconductor sequencing, then increasing the number of reads on the chip over time appears a formality. “If this chip has 1 million reads, it costs the same as if it has 7 million,” says Rothberg. Not only do the costs fall exponentially as more chips are made, but also Rothberg says, “As you get smaller, the sequence gets faster and more accurate.”
“Nothing is cheaper or more robust than the electronics industry,” Rothberg concluded. “We can deliver: if you give us an order for 1000 machines, we’ll give you 1000 machines.” Rothberg closed his well-received presentation by saying of post-light sequencing, “It’s going to be inevitable.”
In his 2007 keynote address at Bio-IT World Expo, Rothberg said: “Moore’s Law has been good to me.” It looks as if it might be again.