CISE News & Events
Song, Kahveci, and Ranka Win Best Student Paper Award at ACM-BCB
Aug 30, 2010
UF CISE PhD student Bin Song and her co-supervisors Associate Professor Tamer Kahveci and Professor Sanjay Ranka received the Best Student Paper Award for their research paper titled “Enzymatic target identification with dynamic states” at the ACM International Conference On Bioinformatics and Computational Biology (ACM-BCB), 2010. The BCB conference was held at the Niagara Falls, NY on August 2nd - 4th, 2010.
In their paper, they developed a novel approach for predicting the most desirable enzyme knockouts in a given metabolic network. Metabolic networks consists of a complex network of reactions that transform chemical compounds. The reactions are catalyzed by biomolecules, called enzymes. By knocking out a set of enzymes, it is possible to change the production rate of some of the compounds in the metabolism. This problem has many critical applications, many of them in biotechnology such as biomedicine and bioenergy. For instance it is possible to modify the metabolic networks of bacteria that generate ethanol in their metabolism and increase the ethanol production rate significantly.
The fundamental contribution of this paper to metabolic engineering is that it models the metabolism in a biologically realistic way. When a metabolism is altered, its state — in other words, the set of fluxes of all of its reactions — changes dynamically until it reaches to a steady state. This paper considers the entire trajectory of fluxes, called the dynamic state, in order to decide the best set of enzyme knockouts. To solve this problem, the paper addresses two key issues. First, it answers the question of how two compare whether two dynamic states are similar. The second issue arises from the nature of the dynamic state. Finding the dynamic state after knocking out a set of enzymes can be computationally very expensive depending on the remaining network. Furthermore, this needs to be repeated for many possible enzyme knockouts. The paper develops smart algorithms to avoid this computation whenever possible or to cut down its cost when it can not be entirely eliminated. The resulting algorithm generates hundreds of enzyme knockout predictions at an impressive speed ranging from several seconds to hours. It takes anywhere from days to months to perform small number of enzyme knockouts in wet lab. As Tamer Kahveci says “Considering the entire dynamic state is crucial in making successful enzyme knockouts. What happens between the initial and the final state of the metabolism may be a matter of life or death for the organism.”
