It would take 10 million years to simulate an E.coli cell with today’s hardware and software, but the University of Alberta is getting ready to pick up the pace.
The school said Monday it had received equipment as well as funding for the $30 million project as part of IBM Canada’s Shared University Research program. The university will get access to colleagues in IBM’s Computational Biology Centre with biological modeling applications expertise, the company said. The exact amount of Big Blue’s financial award was not disclosed.
U of Alberta wants to create a virtual cell, in this case a model of E.Coli bacteria, which it could then use to test potential drugs. The project, called Cyber Cell, is one of the Canadian examples of what IBM executives call a move towards more personalized medicine, or “”targeted treatment.””
Mike Ellison, director of the Institute of Biomolecular Design and lead on the Cyber Cell project at the University of Alberta, said the project could eventually allow health-care providers to take the cellular equivalent of a fingerprint, which could then be tested with various doses of potential medicine. This could speed the rate at which drugs are brought to market and improve the overall effectiveness of treatment, he said.
The task will require massively parallel computer clusters, which are far too slow to track E.coli’s 100 million biomolecules in space and time. However code optimization and specialized architecture from IBM should eventually speed things up 6,000 times, Ellison said, while the company’s forthcoming Blue Gene supercomputer could improve the process even more. Factor in Moore’s Law, which states processor speeds will double every 18 months, and Ellison predicted the school should be able to perform the calculations it needs over a three-day period by 2012.
“”We think it’s very dangerous to allow a lot of assumptions (about the quality of the data) in order to gain in computer speed,”” he said.
Sal Causi, business development executive with IBM Canada’s Life Sciences unit, said modelling, simulation and high-performance computing are becoming a much more important part of drug discovery within academia and among pharmaceutical companies. Big Blue has responded by spending the last seven years to create DiscoveryLink, a middleware application designed to deliver federated access to heterogenous data stored across the enterprise. Causi said DiscoveryLink will connect to any database and allow researchers to ask very specific questions. One he offered was: “”Find a compound similar to X with the following assay results, associated with asthma targets.””
Causi said tools like DiscoveryLink will be vital to disease researchers and drug companies because they help weed through the vast amount of biological data, which he said is doubling every six months. “”There’s no standard in the industry in terms of how you output the data,”” he said. “”If you format the data in one lab, it isn’t necessarily done that way in another lab.””
Though some consortiums have started work on this problem, Ellison agreed that standards are a long way off. “”You’ve got to remember we’re researchers: we’re socially austistic headbangers,”” he said. “”Getting us all in a room to agree on something is a challenge.””