Pacific Bio lifts the veil on its high-speed genome-sequencing effort

Competition to analyze human genomes faster and cheaper — a subject I’ve discussed at length here and here — keeps heating up. The latest shot came yesterday, when Menlo Park, Calif.-based Pacific Biosciences granted the NYT an exclusive look at technology it says should eventually make it possible to sequence a genome in just a few minutes for under $1,000.

Although the “$1,000 genome” is a purely arbitrary goal, it’s become a Holy Grail of sorts for the genomics field. (The startup Knome, which we covered here, currently offers full-genome sequencing for $350,000.) Cheap, fast sequencing of all six billion DNA “letters,” or bases, in humans could make it possible, for instance, for doctors to better tailor treatments to a patient’s own genetic quirks or to identify the specific weaknesses of tumor cells. More broadly, it would also vastly increase our understanding of the genome, which has turned out to be a much more mysterious realm than just about anyone expected only a decade ago, and illuminate the ways DNA varies between individuals, groups and even among cells and tissues within a single individual.

PacBio has long been secretive about its work, although it’s been clear that the company has some heavy hitters backing it, including Kleiner Perkins Caufield & Byers, Alloy Ventures and Mohr-Davidow Ventures. (Our earlier coverage of the company’s $50 million financing last year is here.) A research team including scientists from PacBio — then known as NanoFluidics — and Cornell did publish a 2003 paper in Science describing some of the company’s work, but it’s a fairly abstruse piece even by the standards of the scientific literature, unless ploughing through mathematical descriptions of zero-mode waveguides is your idea of fun.

Before getting into the PacBio technology, however, a word of caution. The high-speed sequencing race is pretty crowded, with more than a half-dozen companies pushing forward with new technological approachs and making grand claims for their speed, accuracy and ultimate utility. None of these claims are really possible to evaluate at the moment, and of course the companies involved have every incentive to play up their strengths. PacBio is no different in this respect, and while the NYT story focuses on this particular startup, reporter Andrew Pollack is careful to quote several skeptics who raise perfectly reasonable concerns about the company’s expansive claims.

According to a sidebar to the main NYT story, PacBio’s basic approach involves trapping individual DNA molecules in thousands of tiny holes, or “wells,” fabricated in a thin sheet of metal. These holes are only 70 nanometers wide, or about one-thousandth the width of a human hair — small enough to hold a single stretch of DNA, but large enough for various biochemical reactions to take place inside.

Once inside the wells, the double-helix structure of DNA “unzips” into two single strands, each of which then rebuilds a complementary strand from chemical bases floating in the wells with the help of an enzyme called DNA polymerase. Those bases are tagged with fluorescent molecules that make it possible to identify them under a microscope. PacBio’s system uses digital-camera technology to observe the process in real time, allowing it to analyze and “read” the order in which the bases are strung together.

PacBio claims several advantages over other high-speed sequencing techniques, all of which involve chopping up DNA into short pieces that can be read more easily and then reassembled into a full genome. PacBio says it can read 1,000 bases of DNA in one go — several times greater than other high-speed techniques, which are capable of analyzing stretches of only 30 to 450 bases at once.

By observing the same reaction across thousands of wells, the PacBio sequencing system should read genomes with great speed. The company is currently able to read about 10 bases a second in any given well, so if 1,000 wells all worked at once — itself something of a tall order — the system should run about twice as fast as existing sequencers.

The company certainly doesn’t lack self-confidence. “If we ever make this work, there would be no other technology applicable in the sequencing field,” Hugh Martin, the company’s CEO, told the NYT. Martin went on to discuss the Archon X Prize for Genomics, which offers $10 million to anyone who can sequence 100 genomes in ten days. “When we’re ready,” Martin said, “we’re just going to win the X Prize.” (PacBio hasn’t even officially entered the contest yet.)

PacBio, however, doesn’t plan to release a commercial version of its system until early 2010, and won’t have a model capable of sequencing a full genome until 2013. Some of its competitors could leapfrog it in the meantime — Intelligent BioSystems, for instance, claims it will have a machine that can sequence a full genome in 24 hours for $5,000 by the end of this year. (The NYT says it will cost $10,000, presumably reflecting the difference between the full six billion bases in our 23 pairs of chromosomes and the three billion bases that make up just one set of chromosomes.) Complete Genomics and BioNanomatrix, who haven’t disclosed details of their technological approach except to say they plan to read a stunning 100,000 bases at a time, assert they’ll be able to decode a full genome in eight hours for $100, although it’s not clear by when.

As I noted earlier, however, it’s best to take all such projections with a grain of salt. Hitting goals like these are best-case scenarios that could be tripped up by any number of unexpected wrinkles in the engineering and technology-development process. Still, it’s a heck of a lot of fun to watch.

PacBio has raised a total of $78 million to date, and will probably need another $80 million to bring its systems to market, the company’s CEO told the NYT.