By Michael Gross
January 13, 2003
| Inpharmatica Ltd. could be called the mermaid of biotech companies, as it is made of two quite distinct halves. In Charlotte Street, West London, one floor of the building that houses many of the company's bioinformaticians is crammed with as much computing power as it can carry — and the building regulations allow. Just a few miles farther north, in the borough of Camden, a wet lab contains 20 biologists wielding their microliter pipettes.
Inpharmatica sells data products, but it also wants to discover drugs. Similarly, its product catalog consists of two rather different items: Biopendium, a database product that subscribers can purchase and receive in regular updates to use at home, and Pharmacarta, a drug discovery platform that users can access only via an agreed research collaboration with the company.
This game of halves is actually a well thought-out plan. "It's our hybrid business model," explains Patrick Banks, Inpharmatica's chief financial officer.
It all began in 1998 in the lab of Janet Thornton, then-director of the Bloomsbury Centre for Structural Biology, a shared initiative between two London universities — University College and Birkbeck College. Thornton, who now heads the European Bioinformatics Institute in Cambridge, England, spun out the company from her research work in bioinformatics — particularly, the comparison of protein structures. Her algorithms, developed over several years, became the cornerstone of the database product that was launched as Biopendium in 1999.
Finding That Little Bit Extra
Essentially, Biopendium picks up all the publicly available protein sequence data and all 30,000 known structures and compares them to find relationships or similarities, however distant. In 15 minutes or less, the software can check all 2.4 billion precalculated relationships within Biopendium and select the ones relevant to the question at hand. In comparison to publicly available bioinformatics packages such as BLAST or PSIBLAST, Banks says, Biopendium can typically provide an additional 30 percent of annotation. For example, public domain searches for protein kinases yield less than 800 examples; Biopendium, by contrast, comes up with 973.
Computer pharm: Inpharmatica bioinformaticians work diligently at the company's West London office, crammed with the latest computing hardware.
An example of the value of this extra annotation power is a protein called nicastrin, which Peter St. George-Hyslop and colleagues in Toronto have identified as a component of the Alzheimer's disease-related complex of presenilin and the amyloid precursor protein. While this group could not relate nicastrin to any known proteins using standard bioinformatic approaches, Inpharmatica researchers using Biopendium discovered that it is related to a subset of transferrin receptors.
For all its performance and power, Biopendium remains a product designed to be used by biologists rather than computer experts. "A biologist can learn to use it in a couple of hours," says Richard Lingard, the company's director of business development. Inpharmatica expects that the main applications will include high-throughput data analysis (such as microarray or yeast two- hybrid studies) to prioritize research projects, deep genome mining, and literature-driven identification and prioritization of target proteins.
In practical terms, clients can buy their own server and license for the Oracle Corp. database system and then run Biopendium locally. Alternatively, they can now also use the database hosted at the 2,500-CPU compute farm in England to run their Biopendium queries remotely on Inpharmatica's hardware.
Current subscribers include a small circle of biotech and drug companies that pay a subscription fee depending on the number of users within the company and receive a new update of the package every three months. The oldest and biggest customer is Pfizer Inc., which signed up in December 1999. Pfizer stores the program on a central server in Ann Arbor, Mich., letting employees around the world access it via the Web. Last July, Inpharmatica secured its first Japanese customer: Daiichi Pharmaceutical Co.
While Biopendium essentially compares proteins and protein structures to each other, the company's second major product, PharmaCarta, matches proteins to small molecules. It is designed to build on the annotation afforded by Biopendium and move from there to drug discovery. Specifically, it helps to assess the "druggability" of the proteins of interest, identify the most promising drug targets, and then match them up with suitable small-molecule compounds. As the program can cross-check for unwanted interactions with other proteins, side effects that would otherwise have shown up in clinical trials can be identified much earlier and at a fraction of the cost.
|Managing the Clusters
|The availability of massive amounts of computing power has played an important role in the development of Inpharmatica's two products.
Using this product in its own research and in research collaborations, Inpharmatica developed the other half of its business model: a biopharma company. In July 2001, Inpharmatica initiated a research collaboration with Geneva-based biotech company Serono S.A. Inpharmatica delivers intellectual property in the form of proposed drug targets; Serono carries out the biological validation.
A major inhouse project at Inpharmatica's Camden research center concerns the family of nuclear hormone receptors. Of the 48 members of this protein family that have been publicly described, 14 already are targets for drugs currently on the market. Using its proprietary software products, Inpharmatica identified 15 additional putative receptors and has isolated 13 of them. Considering the high proportion of drug targets among the 48 previously identified members of this family, the researchers are optimistic that one or more products can be developed against these novel targets.
In a field moving as fast as proteomics and all the other "-omics," companies must run to stay ahead. The double challenge of preserving Biopendium's 30 percent advantage over public resources and discovering drugs that make it to market will keep Inpharmatica on its toes.
In November 2001, the company launched a joint venture with Celera Genomics Group to develop a special edition of the Biopendium platform incorporating Celera's exclusive genomic data. The first issue of the Celera edition of Biopendium was launched recently.
Proprietary data of other customers can also be combined with Biopendium on request, but, as Inpharmatica Chief Information Officer Pat Leach points out, the company would aim at appending the additional calculations at the end of a normal production run (involving 500 processors over three months) to keep the costs down.
On the sales front, Banks says Inpharmatica hopes to have 10 customers for Biopendium in the next two to three years. A new Biotech edition of Biopendium should make the product accessible to smaller biotech companies, which may need to register only a few users and may prefer to use the version hosted at Inpharmatica rather than buying their own server. The intent, however, is to preserve Biopendium as a premium-priced product that can give Inpharmatica clients (and its own researchers) a crucial advantage in a cutthroat field. So far, the data and discovery business are complementing each other well, adding credibility to Inpharmatica's unusual mermaid approach.
Michael Gross is a science writer in residence at the School of Crystallography, Birkbeck College, University of London. He can be contacted through his Web page at www.michaelgross.co.uk.