PI: Mehmet Koyutürk, Case Western Reserve University
An important challenge in life sciences, emerging from various successful genome sequencing efforts, is the characterization of molecular mechanisms that underlie phenotypic differences (variation in observable characteristics of biological systems). These variations include developmental stages, disease, and evolution. Molecular sequence and expression datasets have been used effectively to identify frequently mutated genes and the relative abundance/lack of their products in cells carrying the phenotype of interest. While the role of contextual information, manifested in networks of biomolecular interactions is widely acknowledged, lack of high-quality data has impeded the development and use of network-based analyses. Recent high-throughput methods for network inference hold tremendous potential for network-based phenotype analysis. However, existing computational methods are in relative infancy. The primary objective of this project is the development of computational models and algorithms that enable integration of disparate datasets (gene sequences, gene expression, protein expression, protein interactions) for phenotypic characterization.
Network-based modeling of phenotypic differences gives rise to deep intellectual questions relating to computational abstraction, algorithm design, and biological validation. These challenges are exacerbated by the incomplete, static, and noisy nature of available network data. The proposed research aims to overcome these difficulties through innovative use of combinatorial, probabilistic, and algebraic methods, including the following: (i) combinatorial modeling of coordinate changes in the expression of multiple interacting genes, (ii) probabilistic modeling of the crosstalk in biomolecular networks, for statistically sound evaluation of functional association between multiple genes and proteins, and (iii) algebraic modeling of information flow in the cell, to make indirect inferences on the effects of multiple genetic perturbations. The resulting computational tools will be integrated into frequently accessed application frameworks, including Cytoscape, R, and Matlab, used extensively by domain scientists at Case Center for Proteomics and Bioinformatics, for testing, validation, and calibration, and subsequently disseminated to the broader scientific community.
The proposed career plan also incorporates major educational and outreach initiatives that build on the interdisciplinary nature of proposed research. These initiatives include (i) novel instructional techniques based on active learning, (ii) continued development of a Computer Science education platform, TELESCOPE, for the purpose of illustrating basic algorithmic and problem solving principles, (iii) incorporation of TELESCOPE into outreach efforts, targeting K-12 schools, CWRUs TRIO programs for low income pre-college students, and CWRUs Equinox summer program, (iv) programs for student involvement (particularly minority and female) in outreach efforts, (v) development of new interdisciplinary courses and degree programs at undergraduate and graduate levels, and (vi) continued involvement of undergraduate students in research projects.
