NEW YORK, Oct. 1 - For the last decade, genomics-technology cheerleaders have promised that dirt-cheap, idiot-proof DNA sequencing is right around the corner. Laborious and expensive parts of traditional capillary-based sequencing will soon be obsolete, they say, and a new era will bring better methods, faster machines, and reads so cheap that it'll be possible, just for kicks, to knock out a draft of Uncle Larry in one afternoon.
Needless to say, the thousand-dollar genome is still limited to all-star plenary gabfests and breathless speculation.
But a GenomeWeb survey of top technology developers and major genome-sequence centers shows that most believe steady, unspectacular improvements will keep pushing costs down over the next few years. And in the meantime, alternative sequencing methods now under development may finally deliver that order-of-magnitude kick in the pants that'll allow even the littlest genomicist to sequence and resequence at will.
DNA sequence types talk about their business much like computer people do, with cheerful faith that the process will inevitably keep getting cheaper and faster.
They should be so lucky. Though precise numbers are very hard to calculate, Mark Guyer, director of the division of extramural research at the National Human Genome Research Institute, estimates that the cost of gene sequencing has halved roughly every two years during the last decade. Consequently, the bottom price of a read is now about $1.50. Impressive, but hardly a Moore's law.
Most sequencers agree that incremental advances, like using less reagent, reducing sample volumes, and taking advantage of economies of scale, promise to keep shrinking costs at roughly the same rate over the next three to five years. "We don't think we've yet hit the bottom on the size vs. efficiency curve," says Guyer.
Faster progress could also come from technical changes like integrating sample prep and sequencing, or methodological boosts like finding a way to simplify--or even skip--the mapping step.
Some of those modest improvements may already be here: Applied Biosystems claims that despite having a higher pricetag, its new 3730 machine reduces costs by increasing throughput and read lengths. Richard Gibbs, director of the Human Genome Sequencing Center at Baylor College of Medicine, which tested the 3730 tool as an early-access partner, estimates that it will probably cost about 40 percent less to run than ABI's older machine. "I'm more optimistic than some of my buddies in the field," he admits.
Still, progress in capillary-based sequencing, the current industry standard, may soon hit the wall. "When you see the ultimate refinement of a technology, you know its death is near," warns David Schwartz, a genomicist working on optical mapping technologies at the University of Madison in Wisconsin. "Sanger sequencing is probably going to disappear in five years."
So what could possibly take its place?
Broadly, there are techniques based on incorporating fluorochromes into growing strands of DNA; on microfabricated platforms that sequence by pulling a long strand of DNA through a tiny pore; on atomic-force microscopy; and on microarrays and sequencing-by-hybridization. Approaches that seek to sequence a single molecule promise the biggest bang for the buck, but may be the most difficult to achieve.
In any case, the technology that ultimately wins out over the others must have a few things going for it, says Schwartz. It will have to be robust, cheap, and, well, stupid. "We've seen, certainly in terms of large-scale sequencing, the dumber [the technology] the better," he said. "Simple works. Shotgun sequencing, for example--all hail it--the damn thing works."
The commercial arena is crowded with competitors [see sidebar]. Stalwarts like Amersham and ABI are joined by US Genomics, Mobious, Solexa, 454, and Callida, among others. Each promises to deliver jaw-dropping throughput for pennies. (Craig Venter has already tapped US Genomics to join his new effort to reach the $2,000 genome.)
In fact, genomicists may soon see the promised land. Dan Ehrlich, director of the BioMEMS lab at Whitehead, points out that big money only began flowing into the Human Genome Project four or five years ago, and any significant new technology takes at least that long to develop. Now, with all that cash and time behind us, he says, "it really is not unrealistic to expect major advances over the next five years."
Still, most DNA sequencers aren't holding their breath waiting for The Next Big Sequencing Thing. "There's been a lot of hype, and not a lot of substance from a scientific standpoint" on some of these new techniques, says Elaine Mardis, director of technology development at the genome-sequencing center at Washington University. "To me, it means nothing until you publish a paper or stand up at a meeting. My gut feeling right now is that I don't know which new technology to back because we haven't been provided with a hell of a lot of detail."
While none of these methods is likely to soon replace the Sanger method for de novo sequencing, some of them may be well suited for resequencing, an area of growing interest now that the human genome has been nearly finished. "There are a lot of large-scale experiments that people are still not doing because of the cost and throughput requirements of resequencing--for example, looking at 1,000 genes associated with cancer across 1,000 people," points out Greg Yap, senior director of marketing for DNA analysis at Affymetrix.
Ultracheap sequencing could open up many other doors: clinic-based whole-genome diagnostics; large-scale comparative proteomic studies; and a huge range of agricultural projects. Evolutionary genomics, the study of gene regulation, and the investigation of duplications, deletions, and translocations would also benefit.
But even if a new technology manages to cut per-read costs down to pennies, obtaining a finished sequence will still be packed with a lot of other expenses; genome scientists may find themselves speeding through sequencing only to stall at the speed bumps of assembly, annotation, and data analysis.
Getting high-accuracy data may also be a prolonged struggle, says Baylor's Gibbs. "In three years, someone [may] say, 'We did a genome for a million dollars,'" he says. "That'll impress everybody, but getting those data in real finished form will still cost another $20 million to $30 million."
So while tedious technical tinkering looks like the best near-term bet for bringing down the price of the sequencing process, it'll probably take a shoot-the-moon technological breakthrough to reach the $1,000 genome.
So perhaps big plans and optimistic thinking aren't so bad after all. "I think this field is rife with dreamers, which is good," says Madison's Schwartz. "You latch onto a good idea, and convince yourself, 'Gosh, I'm one step away from sequencing a human genome in five minutes.' In fact, you're not. But we need these kind of people, and we need these kind of dreams."