March 14, 2006 | The great sucking sound of bioinformatics groups being shrunk or swept away after disappointing results has never been heard at Genzyme. It would have been a soft whoosh anyway. Genzyme’s core bioinformatics team has only seven members, including one recent hire. For the first few years, it was pretty much Partha Manavalan working alone at the bench and computer trying to improve ligand-target binding affinities.
This slowness — perhaps prudence is a better word — to jump on the bioinformatics bandwagon hasn’t hurt Genzyme. What began roughly 25 years ago as pursuit of protein-based orphan drugs, propelled by a small cadre of scientists, has mushroomed into a $2.7-billion-plus business with more than 8,000 employees focused on six medical markets (lysosomal storage disorders; renal disease; orthopedics; transplant and immune diseases; oncology; and genetics/diagnostics).
Genzyme is a solid success in the storm-tossed world of biotechs. Compound annual revenue growth for the past five years was 29 percent. Last month, the company reported FY05 income of $441 million and projected sales would surpass $3 billion in FY06. Traditional strengths in niche therapeutics (e.g., Gaucher, Fabry, and Pompe diseases) are being supplemented with cancer and diagnostics initiatives. Its historic protein-only focus has expanded to include small molecules.
In reporting year-end results, CEO and chairman Henri A. Termeer indicated that new products would drive future sales, “while Cerezyme (Gaucher) sales in 2006 will represent just 30 percent of overall revenue.” Myozyme, Genzyme’s new drug for Pompe disease, will launch this quarter. Genzyme has had a productive if occasionally bumpy ride, managing growth driven both internally and by acquisition.
Dabbling Turns to Data
Whether by wisdom or necessity, Genzyme didn’t fall prey to industry infatuation with bioinformatics tools, a point made nicely by Clarence Wang, associate director of bioinformatics and chief evangelist: “Computational modeling, and bioinformatics in general, has a particular role here at Genzyme, which is probably not typical of a company of our size. Genzyme tried to dabble in some of these leading-edge approaches but not commit huge amounts of resources to them without any sense of what’s going to come out at the other end.”
|THE TORTOISE WINS THE RACE: Genzyme's Slava Akmaev, |
Clarence Wang, and Partha Manavalan.
Photo by Jonathan Kannair
Eighteen years ago, Manavalan was the only one dabbling. He remembers computers weren’t cheap then: “The CEO used to come in and say, ‘Partha, you have the best toys.’” Digital Equipment’s microvax computer was the gold standard then. Software choices were limited largely to Tripos (small molecule focused, and therefore of less interest) and BioSym Technologies (now part of Accelrys) for proteins. Competing with Genentech on various fronts, Manavalan spent much of his time “mutating” residues to improve protein potency.
The big change, if it can be called that, came in 1997, when Genzyme acquired PharmaGenics Inc. (a genomics-focused cancer therapy company), and with it an energetic Ph.D. trained in synthetic biochemistry, but self-schooled in bioinformatics and IT — Clarence Wang.
“I had the choice of going back into the lab doing more combinatorial synthesis or bioinformatics,” says Wang. He built the team slowly. “I knew the gene expression stuff was going to be a big effort for the beginning years. It was turning the gene expression technology, SAGE, into something that could be marketed as well as contributing to target discovery internally. It worked out pretty much that way.”
Wang was also charged by CSO Alan Smith to find uses for modern computational approaches in a company traditionally not interested in them.
“It  was the point at which we could start to see other applications for molecular modeling besides protein structural engineering. One thing that helped was the acquisition of a very strong small-molecule drug discovery group with GelTex Pharmaceuticals [in 2000]. They did a lot with high-throughput methods and also dabbled in computational modeling,” says Wang.
Wang led the charge cautiously. Desktop tools for managing data, mostly gene expression data at the time, were among first brought in. Tackling genomic data management was a priority.
“We developed a lot of infrastructure ourselves, for example, a database for SAGE, because there really wasn’t anybody with one. We also made some extensions to it to integrate published microarray data. We also did and still do a lot of one-off solutions for three people in some group that need something specific right away but may disappear in six months,” says Wang.
Manavalan remains the only practitioner of molecular modeling. A current project is work to optimize the second generation of Cerezyme. Manavalan’s earlier models of Cerezyme’s cysteine residues’ susceptibility to oxidation, a factor in drug persistence, were later borne out in assays.
Recently, Genzyme mounted a systems-level modeling effort to improve manufacturing to support its drive to develop new antibody therapeutics. Proteins are basically grown in mammalian cells in giant vats. In the past, even industrial-size culturing was a lot of art. Today, high-throughput technologies able to measure cultured media attributes promise to transform the art to science.
“I think the scientist in these groups realized they could do more with the data that they had available from these experiments,” says applied mathematician Slava (Viatcheslav) Akmaev, a staff scientist in the bioinformatics department. “Overall, what I’m doing for them is applying dynamical systems to model biochemical pathways in these reactors.”
A Measured Approach
Bioinformatics has finally and clearly gained a toehold at Genzyme, says Wang, though he jokes, “I think Partha still has some hesitation about whether joining bioinformatics full-time was a smart move or not.”
Plummeting computing costs and a willingness to handle their own IT issues certainly helped the small informatics group sail under the cost radar. “We run independent of IT for the most part,” say Wang. “It can be a challenge. I’ve been able to try to strengthen our relationship. But still, even on things as basic as Oracle licensing, there’s a lot of back and forth about how that really should be handled.” Corporate IT is focused on basic infrastructure and core business applications.
Along the way, Wang brought in tools at a measured pace. Genzyme was an early user of Scitegic’s Pipeline workflow software, before Accelrys acquired Scitegic. Genzyme’s genomics group uses Ingenuity Systems pathway tools and database. And, of course, Wang’s group develops a fair number of custom solutions. Wang reports to Kathy Klinger, senior vice president of genetics and genomics.
Pondering what lies ahead, Wang hopes the positive results his group has produced will make selling the technologies inside easier. He suggests moving from desktop deployments to an enterprise rollout of at least some tools is being considered. The recent addition to his staff is further evidence of corporate commitment.
In the meantime, it’s still Wang’s job to determine which tools will produce results and to prod Genzyme’s researchers to use them. Each fall, Genzyme researchers gather to review and share their work. Last year, the group had grown so large (about 600) that the venue was moved from a local country club to a major Worcester, Mass., facility. The Genzyme research meeting is replete with posters, sessions, and Clarence.
“It’s like sales,” notes Wang. “You get leads, you get a sense that someone is interested in something, and you go pound on doors. Those posters come from scientific groups all the way from early-stage discovery to development. I just go and take notes. Fortunately, since I report to a senior manager [Kathy Klinger], she can open a lot of doors and at least get me names. Then the responsibility is mine to develop it into a project.”