182. INFORMATION THEORETIC-BASED APPROACH FOR DATA-DRIVEN BIOLOGICAL NETWORKS RECONSTRUCTION

Department: Mechanical & Aerospace Engineering
Faculty Advisor(s): Daniel Tartakovsky | Shankar Subramaniam

Primary Student
Name: Farzaneh Farhangmehr
Email: fafarhan@ucsd.edu
Phone: 858-822-0960
Grad Year: 2014

Abstract
The development of high-throughput technologies of biological measurements has resulted in a massive amount of quantitative data which requires us to develop advanced approaches to analyze and interpret this data to understand the underlying processes and networks. Reconstruction of biological networks from measured data of various components is an interesting challenge in systems biology. Systems biology approaches to cellular networks are based on integration of diverse read-outs from cells. With appropriate analyses of the read-outs, we can quantitatively map the inputs to responses of a given phenotype. Generally, a parsimonious model that captures necessary and sufficient predictive information about a complex system is preferred. Mutual information networks, as one of data-driven network reconstruction methods, rely on analyzing the statistical dependencies of interacting components by measuring the mutual information of interactions. Using the information-theoretic approach, we developed a parsimonious input-output model of regulatory protein-cytokine network. Adverse effects of cytokines have been linked to many disease states and conditions. Cytokines profoundly alter the body's response to cellular damage or invasive pathogens and are secreted by a wide range of regulatory components. Identifying these regulatory components can help understand common regulatory modules for various cytokine responses and help differentiate between the causes of their releases. Our model demonstrated the applicability of the mutual information approach to the reconstruction of biological network from data generated by the Alliance for Cellular Signaling (AfCS) through a systematic profiling of signaling responses in RAW 264.7 macrophage consisting of toll and non-toll receptor ligand data sets. Information theoretical approach provided a predictive model of cytokine releases by analyzing the statistical dependencies of signaling proteins and cytokines. This model not only successfully captures the most of known signaling components involved in cytokine releases but also predicts potentially new signaling components involved in releases of cytokines, like Ribosomal S6 kinas for Tumor Necrosis Factor and Ribosomal Protein S6 on Interleukin-10.The results of this study are important for gaining a clear understanding of macrophage activation during the inflammation process. Keywords: network reconstruction, high-throughput data, information theory, phosphoprotein signaling, cytokine, inflammation.

Related Files:

  1. Farzaneh_RExpo_Abstract.pdf

« Back to Posters or Search Results