It’s true that announcements about new genomics technologies are as abundant as fliers for Cambridge Healthtech Institute conferences. And it’s not always clear which are worth heeding.
To help you out, this month we surveyed the field for a few of the more remarkable instruments and ideas on the horizon. Here, in the technological categories of bioinformatics, wireless computational biology, sequencing, genotyping, mass spectrometry, and bioarrays, we offer a sneak peek at some of the cooler new tools you can expect to see in the state-of-the-art genomics lab in 2002 and beyond.
SNPs for Pennies, Machine for Millions
While every airport in the nation was on alert for suspicious cargo late last year, the world’s most expensive genotyping system was being escorted part by part through US Customs by a team of European engineers. After 18 months in development at Grohmann Engineering in the out-of-the-way German village of Pr m, the platform passed a factory-acceptance test in November 2001 and arrived in giant crates on its customer’s doorstep in La Jolla, Calif.
Sequenom retained Grohmann, an industrial automation firm that counts Intel, Ford, and Motorola among its customers, to build a system of instruments and automated workstations that would enable genotyping at a rate of at least 100,000 samples per day, scalable to 1 million per day.
The platform would integrate every step in the genotyping process, from automated PCR reaction to the transfer of samples to Sequenom’s SpectroChip product, with a LIMS, sample tracking, and other efficiencies built in to bring the cost of running samples down to less than a nickel a genotype — a rate as yet unheard of even in state-of-the-art genotyping facilities.
Says Paul Heaney, Sequenom’s vice president for advanced systems, “We talked to all the bioautomation technology companies. But most of them really hadn’t a view as to how to get to [1 million samples per day]. We had to look out of the box.”
In fact, the endeavor wound up sending project manager Betsy Nanthakumar out of her office: she spent 250 days in Germany last year overseeing design and construction, and then testing each module in a makeshift laboratory she set up at the Grohmann site.
Because the final product can consistently achieve cycle times to run 200,000 samples in a 20-hour period, the platform has been dubbed the MassArray200k. Ramping up to 1 million samples will require multiplexing or scaling the thermal cycling stations up to meet the 26-second cycle time of other modules. At 31.5 square meters, the platform takes up just half the space required by standard technology to conduct the same processes. Watching it work, says Nanthakumar, is a little like watching a toy train set. Plates move down a track and are directed into stations for thermal cycling, enzyme dispensing, and six subsequent processes.
How did Nanthakumar get across Sequenom’s needs to a team of German engineers who had never heard the term single-nucleotide polymorphism? She laughs, “We learned a lot about engineering and they learned a lot about biotechnology.” Adds Heaney, who made eight trips overseas during the project, “There were definite periods when people would look at each other and say, ‘Are you serious? You want to do what?’”
Heaney says Sequenom never intended to commercialize the platform: “We just saw it as an internal need and something that would give us an edge over the competition.”
Sequenom CSO Charles Cantor adds, “We didn’t think anyone in the world would be interested in this.”
Nevertheless, the thing is already being drooled over by big pharma and high-throughput labs. Though it would take Grohmann at least six months to build another one, “at this point it’s three or four different groups that are after us and really interested,” Cantor says. One of them, he adds, is considering ordering more than one.
The price tag? $6 million.
WIRELESS COMPUTATIONAL BIOLOGY
Breakfast (and Blast) in Bed
Maybe you don’t feel the need to do whole-genome Blast searches on your Palm Pilot during your morning commute, but if Mauno Vihinen and his colleagues are right, the days of wireless bioinformatics aren’t that far off.
“People want to do things not only on their desk or their lap, but wherever,” says Vihinen, of the Finland Institute of Medical Technology at the University of Tampere, who developed BioWAP — an Internet service based on the Wireless Application Protocol that provides mobile access to bioinformatics databases and software. Any mobile phone or PDA equipped with WAP browser software can be used to access a BioWAP terminal.
Conceding that current demand for such a tool “is not that great,” Vihinen says the growth of mobile services and devices points to the inevitable need for mobile bioinformatics capability.
So far, the display, compute power, and memory capacity of mobile devices are not up to the demands of full-scale bioinformatics, but these features are improving rapidly, Vihinen says. “It’s the same situation as a decade ago when the capabilities of computers were the limiting step in bioinformatics.”
Although Vihinen expects BioWAP use to be limited for the next year or so, he is confident that advances in mobile technology will soon make it a viable tool for bioinformatics research. Some research labs already use mobile devices in local area networks, he notes, the first sign that more widespread adoption can’t be that far off.
Obviously, detailed graphics and long alignment strings are out of the question for the tiny screen on your wireless, but Vihinen says that “many basic parameters or references” work just fine with a cell phone or PDA. “For example, looking for what sort of motifs or patterns from Prosite are present in a certain sequence or just to check for the molecular weight,” Vihinen says.
EMBL, Swiss-Prot, Trembl, the Protein Data Bank, and 14 immunodeficiency mutation databases are also accessible with BioWAP. The categories of analysis and search services include: amino acids and codons, DNA/RNA sequences, protein sequences, PDB structures, Enzyme Commission nomenclature, mutation databases, and an SRS search.
BioWAP was developed without any industrial support, and Vihinen foresees keeping it available as a free service for the time being. However, he hasn’t ruled out licensing it to an industrial partner if — when? — the technology does catch on. “This is the first and only mobile bioinformatics service, so people would have to build what we already have done,” he says. “And I don’t know how much industry is interested in reinventing the wheel.”
— Bernadette Toner
Making Motifs Marketable
More than a few bioinformaticians who’ve been around the block were dreading yet another lecture on cluster analysis at TIGR’s Computational Genomics meeting in Baltimore last fall when, to their delight, Denise Wolf took the stage. She gave what one attendee calls “the most thought-provoking talk I’ve seen since David Searls’ Chris Overton memorial lecture at ISMB 2000.”
Wolf, a 37-year-old electrical engineer and self-professed “interdisciplinary gal” who works as a senior computer scientist engineer in Adam Arkin’s group at Lawrence Berkeley National Laboratory, presented “Motifs, Modules, and Mechanisms” — a sort of motivational speech on regulatory networks analysis for the 21st century biologist.
“We want it all,” says Wolf. “We want our multi-disciplinary teams of biologists, chemists, mathematicians, and engineers to generate, in a high-throughput fashion, genomic data, macromolecular structure data, mRNA expression data, molecular interaction data, spatio-temporal image data, kinetic mechanistic data, mutation data, and gross phenotype data, and to derive from it a deep, nuanced understanding of the organism ranging from genomic organization to biochemical pathways to cytomechanical and spatial processes.”
Her proposal: Employ electrical engineering methods and mathematical analysis to understand human systems. Instead of working at the gene or protein level, Wolf says she works at the “level of regulatory modules, analyzing networks of interacting proteins, DNA, and so forth that perform functions in the cell [such as] networks that control apoptosis or the development of organisms.”
She explains: “If you’re given a diagram of a computer, one type of diagram might be at the transistor level. You see that and you don’t know what the hell is going on. But if someone draws boxes around it and says, ‘This is a multiplexer and this is an adder,’ and points out the functional units within the network, it reduces the level of complexity and brings it up a level of abstraction. We’re trying to do the same thing with networks.”
Starting with genetic networks that are already fairly well understood, she develops mathematical models and attempts to figure out their basic operating principles using nonlinear analytical tools such as bifurcation theory.
For instance, in a paper that she expects to see published this month in a new journal called Omics, Wolf looks at the tiny genetic network that controls the expression of pillai that form on a bacterium such as E. coli and make it pathogenic by enabling it to adhere to epithileal tissues in mammals.
One by one, she and her team are analyzing such networks in order to build what she describes as “a parts store” of regulatory motifs for the community by identifying, characterizing, and collecting them.
She also urges the computational genomics crowd to develop algorithms and software for systems identification. “How do you take a network diagram and break it up into modules of motifs? And if you have motifs, how do you use them to perform higher-level network analysis?” she asks.
On the experimental side, Wolf, who is presently seeking an academic professorship that would enable her to continue her research while mentoring students, says the world needs synergistic protocols to figure out what data are required to identify a motif. “You don’t want to just keep gathering data in the same way if you’re thinking about the problem in a different way,” she says.
Says Ted Slater, director of bioinformatics at Paradigm Genetics who heard Wolf speak in Baltimore, “Denise is helping us start on the path of taking those kinds of switches and those kinds of modular components and someday we’ll be able to take all those things and put them together and model the way a cell works. Her contribution is giving those of us in the field the tools to get a handle on the complexity that we’re facing.”
To be sure, Wolf’s work is still early stage — at this point, not much more than a thought experiment. But she believes it holds great promise. “The way people react to this stuff, I think it’s going to go places,” she says. “At conferences a year from now you’ll have 30 papers doing motifs. It’s going to be the next wave.”
— Adrienne Burke
Glorious Dubious Mobious
When word of Mobious Genomics’ hush-hush new sequencing technology, boasting speeds of 1,000 bases per second, got out a couple years ago, qualifiers such as “emerging, amazing, beyond-belief” were understatements. A year later, Mobious’ reputation has gone from undercover and promising to the industry’s urban myth — the technology is still under wraps, and most people have mentally added Mobious’ molecular resonance sequencing to the long line of no-hit genomics wonders.
The rampant skepticism leaves Mobious founder Daniel Densham unfazed. Asked why people should believe his company to be anything other than a scam, he laughs and responds, “I would ask the same question.” In truth, he says, the Years of Silence were a result of time needed to build custom machinery and highly regulated facilities including clean rooms, establish collaborations, file for follow-on patents, and secure funding. Densham, 27, expects to be able to publicize Mobious sometime in the first half of this year. “If it wasn’t so valuable, we could kind of announce it half-finished,” he says, adding that the company will sequence some genomes for evidence of its process. “I want to be able to prove it without any doubt whatsoever.”
Established in late 2000, Mobious was created to commercialize the sequencing patents held by Medical Biosystems, a holding company Densham formed three years earlier. The first inklings of what would become molecular resonance sequencing occurred to him when he was about 18, and he began taking it seriously a couple of years later.
A completely different technology from Sanger’s 1970s-era sequencing method, Densham says, “We’ve gone back to the drawing board, and we’re looking at replication as it occurs.” With cell division as a template, Mobious uses chips to sequence each base real-time as it’s added to a strand in replication. Densham says the company has hit the 1,000-bases-per-second mark, but data acquisition tools are too slow to keep up.
New custom robots and detection equipment that came in at the end of December should enable sequencing at 300 to 500 base pairs per second. Typical reads are 150 kilobases — long enough for most complete BACs. Also, because sequencing occurs in real-time without shearing, repeats are read as easily as any other section. “It’s between 100 and 1,000 times cheaper, at least, than existing technology,” Densham says.
In the past year, Mobious has set up research collaborations with universities and drug companies, and should be signing an international pharma as a full-fledged customer this quarter.
Densham was a medical student with a specialization in genetics at Guy’s, King’s and St. Thomas’ School of Medicine when he began to really obsess about this idea for a revolutionary sequencing method that he’d thought about a few years before. “I couldn’t carry on with my studies without knowing if it worked,” he remembers, so he dropped out.
After consulting a biotech patent attorney who encouraged Densham to prove the concept, he managed to convince his reluctant family that it was a worthwhile endeavor. The plea probably didn’t surprise them — as a child, he was always taking things apart and studying them. “They built this sort of shed down by the bottom of the garden … so I didn’t blow up the house,” Densham recalls. At 11, he started a pseudo-company through which he worked with and repaired electronics, and by 13 he was trying to build an artificial eye. Faced with this latest project, his family put up the money to hire a lab for the summer. After weeks of honing the sequencing process, Densham got it to work with just two days of funding left.
He eventually bagged venture funding and launched Mobious, but “I wasn’t willing to sign up a large percentage of the company to a VC, [so] I had to grow the company at a slower rate,” Densham says. Based out of facilities at the University of Exeter, Mobious (named for artificially created loops of DNA) now has 16 employees. Densham plans to start out with contract sequencing. Another major potential source of business will be to study patient samples from failed phase-3 clinical trials and discover the SNPs that fouled up the tests — giving pharma clients a way to recoup the hundreds of millions of dollars spent on drugs that got shelved at the last minute.
Even with the wheels in motion, he’s still nervous about letting people in on Mobious’ technology, though he knows it’ll happen sooner or later. “There’s going to be a point soon where I can’t keep the thing secret because the patents have to be published,” Densham says. And he refuses to worry about the public’s perception of whether Mobious is viable. “Our patents speak for themselves. You can tell if you know the field whether we’re onto something or not.”
— Meredith Salisbury
The Next Dimension in Protein Analysis
David Clemmer fancies himself a modest guy from a small town who put himself through school playing folk music. Little do his old band buddies from Adams State College in Alamosa, Colo., know that he’s also responsible for building a spectacular new mass spectrometry instrument.
The 36-year-old, newly appointed chair of the Indiana University chemistry department has for a few years now been building and perfecting an ion mobility time-of-flight mass spectrometer that has grand possibilities for proteomics. Two years ago his IMS-TOF created a buzz when Technology Review named it a top technology for the 21st century. Today, it is under license to the Cambridge, Mass., startup Beyond Genomics, of which Clemmer is a co-founder.
Clemmer’s IMS-TOF uses gas as a high-pressure tool to separate ions before introducing them to the mass spectrometer. Crediting precedents set by people such as Herbert Hill at Washington State University, Mike Bowers of University of California, Santa Barbara, and his own postdoc advisor Martin Jarrold of Northwestern University, Clemmer says his contribution to mass spec has been to realize that the separation of gas occurs on millisecond timescales while time-of-flight is on microsecond timescales.
“Because you do things so quickly on millisecond timescales, you can essentially connect any other separation technique to this and then do a multidimensional experiment,” he explains.
That’s especially helpful for analysis of things as complex as proteomes. “If you have a very complicated sample where you have many species that have nearly identical masses, the mass spectrum will begin to become very complicated and you’ll lose peaks underneath other peaks,” says Clemmer. “If you can do another separation that’s a little bit different in the gas phase before you go into the mass spectrometer, you can see those peaks.”
Clemmer says it’s easiest to think of IMS-TOF as a “dispersive technology rather than as a filtering technology.” Ions aren’t thrown away, they’re moved apart in time, broken apart, and moved into the mass spectrum. “Those ions that you see fragments from at the same time in the high-pressure separation can be correlated. There’s a correlation in time and that can be used to sequence or to gain fragmentation information for ions in parallel. You can’t do that with mass spectrometry alone,” he notes.
Who better to capitalize on the next big thing in mass spec than Noubar Afeyan, founder of PerSeptive Biosystems? Afeyan flew Clemmer to Boston and pitched him on a business plan that featured his invention as one in a stable of “horizon technologies.”
“When I saw how they placed our platform into others I became really excited,” Clemmer says. “The idea that we’d be incorporating them into a systems biology platform to study disease states and healthy states sounded really exciting to me, and they had all the right people in place to do it.”
Clemmer’s first protoype, as well as his first PhD student Steve Valentine, are now installed at Beyond Genomics. While the prototype, which takes up the space of a small office, is set up to perform three-dimensional analysis — liquid chromatography, ion mobility, and time-of-flight — a second-generation machine that Clemmer is still tinkering with will include a fourth dimension: collision. “That’s where you get the fragmentation information that helps with the sequencing,” he explains.
Beyond Genomics CSO Steven Naylor says, “Clemmer’s instrument is going to give us a dimension of resolving capability that is anywhere from a hundred to a thousand times better than just doing conventional online chromatography. It’s in effect a second-generation proteomic analyzer.”
Beading the Odds
The discovery of what has become Illumina’s BeadArray technology was “a classic case of serendipity,” says inventor David Walt, 48, a professor of chemistry at Tufts University and head of Illumina’s scientific advisory board.
The story, set in the mid-’90s in Walt’s fiber-optical chemical sensor development lab, goes like this: Grad student Paul Pantano got an undesired result when trying to make etchings in the cladding on an optical fiber. Instead of etching the cladding, his experiment produced a bunch of little wells in the fiber cores. Interesting, but not what he was after, so he shelved it.
About a year later, another grad student, Karri Michael, put one of Pantano’s unwanted etched fibers into a solution of latex beads and took a scanning-force microscope image of it. She found that a high proportion of the wells had been filled with beads. “It was an interesting observation,” says Walt, “but we had absolutely no idea what to do with it. So it sat around in the lab for another year.”
Walt, meanwhile, already known for using photopolymerization to make sensors out of optical fibers to measure things such as carbon dioxide in the ocean or fuel leaking from storage tanks, had begun trying to put DNA on the ends of fibers. (His area of expertise, he explains, is “attaching things to surfaces.”) Photopolymerization was so cumbersome, however, and the laser beams used to deposit the sequences so restrictive, that he found he could fit only 10 or 20 DNA probes on a fiber.
As the molecular biology research community’s desire to analyze DNA sequence heated up, Walt got the message that 10 or even 100 probes on a fiber wouldn’t be useful to anyone. “People started saying you might need 1,000 [probes on a sensor], and then they started saying 10,000, and the numbers just started growing as sequence information became available.”
Walt’s friend Clark Still, the scientific founder of Pharmacopeia and an organic chemist at Columbia, had already invented a technology for making libraries of compounds by tagging beads and carrying them through mix-and-split synthesis. Says Walt, “It occurred to me that we could use the same kind of approach — using tagged beads to create arrays.” The key for DNA analysis was that the DNA would be attached randomly to the beads, and pulled into the optical fiber wells to be decoded afterwards.
For the first time in his 15-year academic career, Walt says he pulled four of his students off their own projects and set them to work perfecting bead arrays. “Within a few months, we had demonstrated that we could successfully make high-density arrays,” he says.
Soon enough, venture capitalist Larry Bock, who had helped launch a number of biotechs including Pharmacopeia, took a short-term license on the invention. In 1998 he launched Illumina.
Now a 200-person company housed in a state-of-the-art office building in La Jolla, Calif., Illumina has automated the process of manufacturing arrays of BeadArrays. In their earliest incarnation, each Array of Arrays holds 96 fibers glued down to a microtitre plate. Each fiber carries 2,000 unique beads, 50,000 overall, each of which hosts between half a million and one million molecules of DNA. Illumina is working to increase the density by adding beads and expanding the fiber bundles, and Walt’s at work shrinking beads from four microns to a few hundred nanometers, “which would enable you to put literally millions of beads in the same region that thousands now fit,” he says.
Illumina expects to begin selling its arrays for a few thousand dollars apiece by the middle of this year through a partnership with Applied Biosystems. In the short term, the company touts the technology as a multipurpose tool applicable to genotyping, gene expression analysis, and proteomics. Walt envisions the invention eventually being used for high-throughput diagnostics or for routine genotyping of newborn babies.
— Adrienne Burke