Cellular modeling is more than just the computational flavor of the month, according to GlaxoSmithKline’s Igor Goryanin. Rather than being just an “academic exercise,” he said, computational systems biology methods are already beginning to have a real impact on at least one pharma’s bottom line.
At a talk at SigSim 2004, a biosimulation workshop that preceded this year’s ISMB/ECCB meeting in July, Goryanin discussed some concrete benefits that his cell simulation and pathway modeling team has already brought to the GSK table. Indeed, in a time of shrinking R&D budgets within the pharmaceutical industry, the very existence of such a group is evidence that it must be doing something right. Goryanin said that his team has delivered proof that cellular modeling is not “esoteric knowledge” — as some of his colleagues think — but “a set of methods and tools to solve practical problems.”
Goryanin cited several key areas in which cellular modeling is already having an impact at GSK, including hypothesis generation, experimental design, and drug precursor production. “Modeling has to go further than bioinformatics,” he said. “The key point in saving money is that you can use these models to help design experiments.”
Some modeling approaches, such as reconstructing metabolic networks, are catching on in computational biology, but Goryanin stressed the importance of going a step further, toward kinetic modeling. Metabolic networks may be useful for identifying potential targets, he said, but these models don’t account for the dynamics of biological systems — a crucial factor in determining the biochemical behavior of potential therapeutics. The linear pathways depicted in biology textbooks often don’t account for the “feedback loops” that are often responsible for drug failures, he said.
Nevertheless, GSK has seen some success from metabolic modeling. In one example, Goryanin said his team created a metabolic model for tuberculosis that it then compared to a metabolic model for E. coli to predict analogous enzymes in TB that might serve as potential drug targets.
Another concrete example of the potential payoff for cellular modeling was also related to E. coli, which is used in the fermentation process for a small molecule that serves as a precursor for one of the company’s drug products. GSK wanted to lower the cost of the process, so Goryanin’s team was assigned the task of engineering a new strain of the bacterium that would increase the yield. Using its computational model, the group predicted nine mutations that would do just that, and the outcome of the project was a yield increase of 23 percent, Goryanin said.
In the drug development area, GSK has turned to computational modeling to determine why patients with mutations in the RAS pathway are less likely to respond to a certain GSK drug. Goryanin said that his team plans to publish a paper on this project “soon.”
Goryanin identified a number of challenges that the cellular modeling field must confront before gaining a larger foothold in pharmaceutical R&D. For one thing, he said, a “new generation” of pathway and network databases will be necessary to build a foundation for more complex biological models. This leads directly to a challenge that already plagues the field — the integration of those databases in a way that can lead to “the utilization of all biological knowledge.”
Looking forward, Goryanin said he sees a great deal of potential in combining cellular modeling with molecular modeling and in silico docking methods, as well as with pharmacokinetic/pharmacodynamic modeling approaches (see story on p. 1 for a look at two companies doing this now). Such collaborative tasks could serve as outreach exercises to help expand the reach of this emerging area, which still requires a bit of evangelizing, he said.
“It’s important to show the value of this technology to our colleagues,” he said. “We have to convert them to believe us.”