## Research Abstracts Online

January - December 2011

Main TOC ...... Next Abstract

### University of Minnesota Twin Cities

College of Science and Engineering

Department
of Computer Science and Engineering

# PI: Daniel L. Boley, Fellow

### Computation and Analysis of Metabolic Pathways

This project to compute metabolic pathways is a joint project between Computer Science and a team in the Biotechnology Institute led by Professor Friedrich Srienc. It is computationally difficult with exponential algorithmic complexity and a high degree of parallelism of a highly non-regular nature. The ultimate goal is to make progress in the area of genetically engineering microorganisms for the purpose of obtaining optimal amount of product metabolites, such as biofuels (ethanol), and with the sufficient growth support (biomass). Once the elementary modes are computed, it is possible to compute the optimal gene knockouts to preserve desired pathways. These genetically engineered microorganisms, if done properly, may produce an optimal amount of ethanol and biomass given the appropriate metabolic intake. For this purpose, previous work has relied on the established theory of Elementary Mode Analysis and the existing, but non-scalable software to compute the complete set of elementary modes for a given metabolic network. The researchers have developed a C++ implementation with support for hybrid parallelization using MPI and OpenMP. They have continually been improving and optimizing the parallel software in terms of utilization of both processors and memory. In current work, the researchers intend to re-implement the algorithm for the computation of elementary flux modes using Global Array Toolkit (implementation of partitioned global addressing space), and port it to a GPU machine using CUDA architecture. With respect to the problem of finding efficient reaction knockouts, they intend to improve the algorithmic design and efficiency of the current parallel implementation. Ultimately they are looking to attain the computation of the complete set of pathways for the genome-scale metabolic networks.

### Group Member

Dimitrije Jevremovic, Graduate Student