International Journal of Computing
High throughput biological experiments are critical for their role in systems biology – the ability to survey the state of cellular mechanisms on the broad scale opens possibilities for the scientific researcher to understand how multiple components come together, and what goes wrong in disease states. However, the data returned from these experiments is massive and heterogeneous, and requires intuitive and clever computational algorithms for analysis. The correlation network model has been proposed as a tool for modeling and analysis of this high throughput data; structures within the model identified by graph theory have been found to represent key players in major cellular pathways. Previous work has found that network filtering using graph theoretic structural concepts can reduce noise and strengthen biological signals in these networks. However, the process of filtering biological network using such filters is computationally intensive and the filtered networks remain large. In this research, we develop a parallel template for these network filters to improve runtime, and use this high performance environment to show that parallelization does not affect network structure or biological function of that structure.
Cooper, Kathryn Dempsey; Ufimtsev, Vladimir; Bhowmick, Sanjukta; and Ali, Hesham, "A Parallel Template for Implementing Filters for Biological Correlation Networks" (2013). Interdisciplinary Informatics Faculty Publications. 28.