%0 Generic
%9 Doctoral Dissertation
%A Gupta, Sanjana
%D 2020
%F pittir:39391
%K cell signaling, single-cell dynamics, heterogeneity, computational modeling, bayesian parameter estimation, lasso, information theory
%T Statistical and Mechanistic Approaches to Study Cell Signaling Dynamics
%U http://d-scholarship-dev.library.pitt.edu/39391/
%X Cells use complex signaling systems to constantly detect environmental changes, relay extracellular information from the cell membrane to the nucleus, and drive cell responses, such as transcription. The ability of each single cell to dynamically respond to changes in its environment is the basis for healthy, functioning, multicellular beings. Diseases often arise from dysregulated signaling, and our ability to manipulate cell responses, that stems from our growing understanding of signaling processes, is often the basis for disease treatments.    Computational approaches can complement experimental studies of cellular systems, allowing us to formalize our growing body of knowledge of cellular biochemistry. Mechanistic modeling provides a natural framework to describe and simulate complex systems with many system components and causal interactions that often lead to non-intuitive emergent behavior, lending itself well to the analysis of signaling systems. Statistical approaches can complement mechanistic modeling by enabling an analysis of complex input-output relationships in the data, providing insight into how cells translate input environmental cues into output responses, even when the underlying mechanisms are only partially understood.    In this thesis, we explore both mechanistic and statistical approaches and address several challenges in modeling signaling processes within a cell, and signaling heterogeneity between cells, using the NF-kB pathway as a model system. First, we evaluate methods to efficiently determine numerical values of model parameters, enabling model simulations that are comparable to experimental data. Second, we develop methods to identify reduced submodels that are sufficient for the data, highlighting simple mechanisms that drive emergent behavior. Third, switching gears to study signaling heterogeneity, we use information-theoretic analyses to evaluate the capabilities of the NF-kB pathway to effectively transduce cytokine dosage information in the presence of biochemical noise. Finally, we develop a framework to calibrate mechanistic models to heterogeneous signaling data, enabling simulation-based analyses of single-cell signaling capabilities.