May 19, 2004
| Last month, systems biology pioneer Entelos received a milestone payment from Organon for delivering 30 prioritized novel targets for the treatment of rheumatoid arthritis. For some time now, Ingenuity Systems has trumpeted its collaboration with Millennium Pharmaceuticals, and, at the recent Bio-IT World Conference + Expo, Millennium's associate director of informatics, Greg Tucker-Kellogg, declared, "I can say without hesitation that Ingenuity's Pathway Analysis has surfaced important findings that we simply would not have found."
Slowly, deal by deal, evidence of progress is mounting in the ill-defined world of systems biology.
Yes, debate still rages over what constitutes systems biology. Is it data mining writ large? Must it be quantitatively predictive? Is top-down modeling (starting with an endpoint) better than bottom-up exploration?
Says Michael French, chief business officer at Entelos, "We struggled with whether or not we're really a systems biology company. Really, we were just standing there when systems biology swept us up. From a lot of people's standpoint, you put a mathematician across the table from a biologist and you got a systems biology approach."
The most comprehensive description is one put forward by Lee Hood, founder of The Institute for Systems Biology. French ticks off Hood's six tenets: "Systems biology should be hypothesis-driven; use global data (e.g., genomics); be quantitative; be integrative; be iterative; and be dynamic."
No company does all of that. Not Entelos. Not Ingenuity. Many needed technologies aren't ready. Nevertheless, the progress made by Entelos and Ingenuity suggests systems biology is advancing as an approach to drug development. What's more, the approaches taken by Entelos and Ingenuity reflect two main channels in systems biology.
Ingenuity's combination of a proprietary knowledge base (IKB) and a Web-based pathway analysis (IPA) tool has the feel of vast, complex data mining. IKB is filled with roughly 1.5 million findings, the majority of which were extracted by an army of consulting scientists who dig deep into texts and focus heavily on quantitative findings (figures, methods, and materials). Data are rigorously curated and structured. IPA permits researchers to mine the data to identify key pathways based on sophisticated statistical scoring.
"For some customers, it's a choice of whether they want to read the literature for prior knowledge or do research. We'd like to make those things not mutually exclusive. What we offer is a way to integrate and visualize all of the information in peer-review literature," says Megan Laurance, product research scientist at Ingenuity.
"Customers supply a list of genes they've compiled that represent the state of a particular biological system," Laurance says. "IPA computes these to pathways on the fly, and the output is a series of networks focused on the genes and proteins that you informed the system were most significant to you."
Entelos starts with the answer (e.g., a remediated disease state) and takes a top-down approach to modeling a pathway leading to that state. It combines public and client knowledge in the process. The models are deterministic and somewhat quantitative (e.g., they can show dose-dependent differences in response) versus correlative. Entelos calls these PhysioLabs and holds patents on key hierarchical modeling and parameter management techniques as well as the models themselves.
"The purpose is to rapidly look at many what-if questions and get feedback on the questions," French says.
To prove its rheumatoid arthritis model worked, Organon required Entelos "to take a clinical protocol, lie it on top of our model of the human arthritic joint, and demonstrate that by applying our model, we could simulate a human clinical behavior. The endpoint in that [model] development was when we could accurately simulate synovial hyperplasia and cartilage degradation over a 10-year period based on human studies," French says.
Clearly, many bottlenecks remain, not the least of which is hype. Just as clearly, systems biology is already helping researchers see both the forest and the trees — albeit in limited ways. The best is yet to come.