Survival of the Fittest?

As sequencing for the human genome wraps up — next month’s 50th DNA anniversary will double as a celebration for the completed sequence — Kucherlapati and directors of more than half a dozen midsize genome centers across the country are looking at an uncertain future. Among them are well-known HGP contributors Cold Spring Harbor, University of Oklahoma, Stanford, and the University of Washington. Unlike dozens of smaller labs around the US that generate GenBank contributions with maybe one sequencing instrument and a couple of researchers, midsize facilities were cultivated for high-throughput sequencing. Some say they have scaled up enough over the years to compete with the cost efficiencies of the biggest production-class sequencing centers.

But post-HGP, it’s becoming clear that NHGRI will award most of its grants for major organism sequencing to the Big Three: Baylor, Whitehead, and WashU. (Separately, DOE continues to fund massive sequencing at its Joint Genome Institute.) At those behemoths, banks upon banks of sequencers spew out, assembly-line style, 30 to 45 million lanes of sequence per year on annual budgets in the neighborhood of $50 million.

With fewer than 20 sequencing instruments apiece, midsize centers operate on anywhere from $2 million to $10 million annually, and, at least in their heyday, could run more than 1 million lanes per year. They’re unusual creatures on the scientific landscape — too big for host universities to support, too small to win major sequencing grants.

During the past few years, NHGRI has approached each midsize center with the same message: We don’t need you; there’s enough sequencing capacity in the triumvirate. The institute is allocating less and less for basic sequencing. It reduced expenditures from $129 million in 2000 to $86 million in 2001. If the midsizers want to keep sequencing, they’ll have to find a compelling niche to fill. Without one, their grants will disappear.

No one’s arguing that NHGRI should pump money into these medium-sized centers simply for the sake of keeping them afloat. But people throughout the community contend that their loss would be a blow to the future of genomics, eliminating the best sources of technology innovation, training, competition for larger centers, and large-scale genomic resources that can be used for medium-scale projects or individual researchers.

Even unexpected voices have gone on record in favor of saving the publicly funded centers. “It’s a mistake that they cut out the Cold Spring Harbors and the small institutions, labs like the one in Oklahoma. A lot of sequencing innovation has come out of these laboratories,” argues Craig Venter. “You need to have a research environment that drives inventiveness.”

As the genomics community continues its rapid evolution, it’s unclear whether these centers will make the Darwinian cut. One thing is certain: only the adaptable will survive.

Mapping the History

The narrowing of the ranks among genomics centers isn’t really a new phenomenon, many scientists point out. “There’s already been a lot of turnover,” says Maynard Olson of the University of Washington. “It’s kind of a harsh system.”

A whole set of labs was already phased out several years ago in the conversion from mapping to sequencing. Ron Davis, director of Stanford’s Genome Technology Center, recalls of the mapping centers, “Some of them were kind of forced out, and not necessarily correctly. … Very few were forced out because they were doing a bad job.”

The HGP began with a priority to fund as many centers as possible. “When we started sequencing, you have to remember, it was not at all clear how it was going to be done,” says Elke Jordan, former NHGRI deputy director. “Our advisory council to the genome institute was concerned about narrowing the field too much to known approaches and wanted to be sure that someone with a bright new idea would still have an opportunity to get into the act. That’s how the … midsize centers got into it.”

Some argue that NHGRI didn’t put enough resources into technology development. “One of the disappointing things about the Human Genome Project is that it didn’t put as much effort into new technology,” says Glen Evans, whose UT Southwestern mapping center got sequencing grants in the early rounds, but was cut off by 1999. “There were all kinds of good ideas, but there weren’t enough funds to bring anything to fruition.”

The HGP carried on and the expected breakthrough technology never broke through. Without a dramatically different approach to sequencing, Jordan says, “the smaller centers were never able to overtake the big ones in terms of efficiency or cost.”

Cost of It All

NHGRI shed the reluctance to whittle down its outposts, and midsizers were sent letters as early as 1999 warning that they’d have to find another way to make themselves valuable. “We have enough capacity in our large centers for the sequencing we need to have done,” says Jane Peterson, NHGRI’s associate director of extramural research. “They [had] to define their role in a unique way. … Not only something that is unique, but is cost-competitive, efficient, and [they] have to have a track record in it.” NHGRI, she adds, made a “conscious decision” to fund fewer genome centers this year than last year.

The reason for that was, ostensibly, that the bigger centers are more cost effective, giving NHGRI better bang for its buck. “When you’re using the volume that [our centers] are using, you get pricing that’s different from what a smaller center’s going to get,” says Graham Scott, director of sequencing instrumentation and robotics at Baylor’s Human Genome Sequencing Center. “That’s just a reality of business.”

Dick McCombie, head of Cold Spring Harbor’s midsize genome center, backs up Scott’s view: “There’s very little doubt in my mind that we’re a bit more expensive than [the major centers],” he says.

But not all medium-size center directors buy that argument: “I’m skeptical, actually, that there are really large economies of scale above 1 to 2 million reads a year,” says Olson. “Large centers are not doing anything technically that’s much different.”

Others say they’ve seen no evidence of a cost differential. Stanford’s Davis says NHGRI told the midsize centers they were more expensive, but never released data to prove it. And Venter says, “Nobody has calculated what the minimal size you need [is] for the maximal efficiency.”

Either way, NHGRI’s warning to the midsizers was hard to ignore. “There was an RFA out that they were going to expand sequencing centers, and you had to have so many bases already completed so you could compete. All the medium-sized ones couldn’t compete,” says Davis. “It was a done deal.”

It might seem strange, given the warnings and obvious attrition, that there are still midsize sequencing centers in jeopardy. “It wasn’t that they didn’t see it coming,” Davis explains. “They weren’t in a place to do much about it.” By then, centers were heavily reliant on HGP grants and had few alternatives. “It was kind of like seeing a famine coming, but you’re hungry, so are you going to stockpile food?” he says.

“This has actually always been a question of how the effort and the project should be distributed — among what size centers and even how many big centers,” says Cold Spring’s McCombie. “It’s unfortunately not the kind of thing where you can actually determine what is best.” He also points out that, like it or lump it, NHGRI had to make allocation choices very quickly. “Realistically, if those decisions had not been made when they were, the human draft would not have been completed when it was.”

Still, Davis, whose center has essentially phased out all sequencing work, clearly opposes NHGRI’s policy. In part, that’s because he believes that narrowing the mapping centers all those years ago may have been a mistake. “In retrospect,” he says, “it would have been good to have those people involved in the Human Genome Project.” Whatever the next phase of genomics is, Davis doesn’t want to lose his peers along the way.

Center Strategy

The centers have responded to the ongoing shakeout differently, and with varying degrees of success. At one extreme is Glen Evans, who disbanded his center once his sequencing grant ran out and the university lost interest in the unfunded project. He persuaded NIH to allow him to transfer the balance of the grants — primarily, the instruments — to a commercial venture he founded, Egea Biosciences, which focuses on DNA synthesis.

By contrast, Elke Jordan points to Eric Green’s lab as exemplary at carving out a niche: Green’s comparative genomics work has become so important that he refers to his group as “a scouting party” responsible for testing new genomes to see which merit sequencing. A mapper by origin, he began with chromosome 7. His center later took that chromosome’s sequence and looked at the orthologous regions in mouse and rat. Now his 38-person team, with annual funding of $7-8 million, previews genomes. “It’s completely complementary to what the big genome centers are doing,” says Green, who has — at least for now — secured his place in the sequencing domain.

Bruce Roe, director of the Advanced Center for Genome Technology at the University of Oklahoma, has wanted to do similar work with his own favorite region, chromosome 22, but so far hasn’t piqued the interest of NHGRI. Last year, “the competing renewal for my genome center didn’t get a high enough priority score for funding,” he says. He resubmitted grants in January, following the institute’s recommendation that he cut his requested funding of $4-5 million per year to $1.5-2 million. For now, he’s surviving on half a million dollars in bridge funding from NHGRI, which covers his students’ work and not much else. His staff, reduced through attrition from 83 to 61, has taken 20 percent pay cuts, and Roe gave up his grant-funded summer salary.

Similarly, McCombie’s lab suffered layoffs as its funding shrank. Last year, McCombie had to let go three people, leaving 18 at the center, which has frenziedly submitted grants with a focus on experimental verification of computational gene prediction, in addition to plant projects in rice, corn, and sorghum. Cold Spring Harbor still gets a little money from NHGRI, McCombie says, but most comes from NSF, NCI, and USDA. The shift doesn’t bother him, though: “We actually want to be making this evolution,” he says, hoping to find research that builds on the center’s robotics and sequencing expertise.

Some who saw their sequencing days numbered took advantage of a relatively new NHGRI program aimed at smaller centers willing to switch gears. Centers of Excellence in Genomic Science, including Maynard Olson’s group and Rick Myers’ Stanford Human Genome Center, will each get about $3 million annually for five years. Olson focuses on finishing methods and keeps up collaborations with other genome centers. He contends that specialization is one place where smaller centers beat out the bigger ones: “The very large centers are actually quite weak technically in finishing and annotation.”

Myers, head of one of two genome centers at Stanford, has a functional genomics CEGS grant and another distinction that has made all the difference: a collaboration with JGI dating back to 1999. Most of the center’s efforts are in polishing JGI data, so “we actually have finished more than 10 percent of the human genome,” he says. The DOE recently funded the center for another five years, giving Myers a peace of mind most of his counterparts can’t enjoy. (So does his $10 million annual budget.) It also keeps him in sequencing, to the envy of his peers. “The plan for the next five years is to sequence our brains out,” he says.

Most directors agree that diversity is key to survival. Lynn Doucette-Stamm heads up the midsize sequencing center based at Genome Therapeutics, which gets about $8-10 million annually from NHGRI. “One of our approaches was not to have all our eggs in one basket,” she says. GTC wound up working on the rat genome and launched fee-for-service sequencing to keep its facility occupied. “Once you’ve built a midsize center you don’t want to just shut it down.”

Other Options

The centers pursue other avenues, as well. An obvious one is getting grants from other agencies, such as NCI or Allergy and Infectious Diseases. “It’s a failed logic that NHGRI is the only place to go for funding,” says WashU’s Elaine Mardis, urging the midsizers to be more creative in seeking out grants.

But Olson points out that it doesn’t necessarily matter who’s offering the grant — “there is a tendency for the work to gravitate to the same big centers.” Agencies such as NSF, though, have been essential in giving smaller centers microbial sequencing projects.

Paul Richardson, head of functional genomics and genomics technologies at JGI, says partnering with bigger centers could pay off for the medium centers. For instance, even genomes that are sequenced at the major centers could fall under the purview of smaller centers for post-sequencing work. “The data can be analyzed wherever,” he says.

A consortium approach has also been considered. Roe’s lab is currently working on alfalfa relative Medicago, and one of the grants he submitted this year is for a consortium of smaller groups to continue that research, divvying up a proposed $12 million from NSF.

Ron Davis also tried to get the midsize centers to join forces a couple of years ago, but admits that nothing happened because “I’m not very organized.” He points to the HapMap as an example of a project that would have been well suited to the smaller centers working together.

Good Things Come…

There’s no doubt that midsize centers have made contributions to the genomics community. One of the earliest was in teaching the art. “[They have] had very key roles in training people who’ve gone on to work at the larger genome centers,” says Mardis, who, along with WashU director Rick Wilson, came out of Bruce Roe’s lab. Roe’s talent for training is legendary: at conferences, people continually thank him for the postdocs they’ve hired out of his lab.

Kucherlapati calls midsize centers “fountains of technology,” noting that many of them were created with tech development mandates, and that losing centers will mean losing ideas. Ron Davis, whose center gave the industry an oligo synthesizer, shakers, colony pickers, DNA prep instruments, and the SNP detection technology that’s the basis for ParAllele, says he had three or four sequencing technology projects in the pipeline when he realized that his center wouldn’t be using that kind of technology in the future. “We just pulled the plug on all the [sequencing-related] instruments that were in development,” he says.

The midsize centers have also driven competition and innovation, proponents say. “There is virtually zero innovation presently in large-scale genome analysis,” Olson contends. Innovation is stifled by the need to reduce costs across the board — eliminating any pilot projects “that are more expensive than the cheapest read per base pair,” he says.

The competition argument is a sticky one. Mardis calls it a “slap in the face” to her group, which works to drive costs down. And Elke Jordan sees competition all around: each major center is trying to outwit the others, from Baylor to JGI to Sanger and more.

Regardless of their relation to big centers, the midsizers have an undeniable role to play within their own communities, says Kucherlapati. “I have long argued that having the genome sequencing done at a few large centers is not in the best interest of the scientific community. Being able to do large-scale sequencing is very important for all biology and should be in the reach of all scientists.”

“I think midsize centers provide a nice conduit between large-scale genomics and an individual investigator,” Eric Green says. How do you put a value on centers that can devote their significant resources to handling individual researchers’ projects? he asks.

Those smaller assignments comprise a demand that major centers aren’t equipped to meet. “They can’t take on little projects,” Stanford’s Rick Myers says. “They shouldn’t. That kills them.”

Elaine Mardis cites resequencing work as one good example: “Team up with a top-notch clinician and go after a specific disease.” These projects might run on half a million dollars — an amount most major centers wouldn’t touch, but “fully enough to sustain” smaller centers, she says.

Midsize facilities also represent a significant investment that shouldn’t be walked away from, Kucherlapati says. “These centers clearly provide tremendously unique expertise. It’s not available easily nor can it be acquired in other places.”

Vertical Horizon?

Rail as they may, midsize centers haven’t convinced everyone that they’re a necessity. “It was a firm belief of the people in the genomics community from the beginning that we shouldn’t be in the business of establishing centers” that couldn’t be shut down, says Elke Jordan. “We should not really view the demise of a center as a loss.”

On some level, even Roe might accept that. “It may be that the life of these centers is finite,” he says, though he’s obviously hoping not to have to face that. As he watches his ranks dwindle and his grants get rejected, he tries to remain upbeat. “You bounce back and you write another grant,” he says. “And part of the answer is, I put [my concerns] in a box and I’m not thinking about it.”

Olson, for one, doesn’t want to watch the field shrink even more, and he argues that it’s NHGRI’s responsibility to protect the genomics community from a sequencing oligo-poly. “During the peak phase of the Human Genome Project there was a real soup-to-nuts attitude about genome centers. … [But] all industries become less vertically integrated as they mature, because smaller organizations specialize on particular aspects.”

Olson likens the big centers to Fortune 100 corporations “Those very large centers will defend their vertically integrated model,” he says. Moving away from that model will “only come with pressure from funding agencies. Neither AT&T nor IBM voluntarily disassembled,” he notes.

In the meantime, Ron Davis advises his counterparts at midsize centers, look at your other expertise, develop that, and forget sequencing. “There are too many things against you, I don’t care how good you are,” he says. “You could be more cost-effective and more innovative and still not win.”

If NHGRI no longer has a need for midsize sequencing facilities, it’s because the major centers — Baylor, Washington University, and Whitehead — have ramped up production tremendously in the past few years. Last fall, GT sat down with Baylor’s Graham Scott and WashU’s Elaine Mardis, both of whom describe themselves as genomic instrument developers, to hear what’s in store for the supersized centers.