Date of Award:

Spring 2017

Document Type:

Thesis

Degree Name:

Master of Science (MS)

Department:

Computer Science

Advisor/Chair:

Nicholas Flann

Abstract

Multiscale computational models integrating sub-cellular, cellular, and multicellular levels can be powerful tools that help researchers replicate, understand, and predict multicellular biological phenomena. To leverage their potential, these models need correct parameter values, which specify cellular physiology and affect multicellular outcomes. This work presents a robust parameter optimization method, utilizing a parallel and distributed genetic-algorithm software package. A genetic algorithm was chosen because of its superiority in fitting complex functions for which mathematical techniques are less suited. Searching for optimal parameters proceeds by comparing the multicellular behavior of a simulated system to that of a real biological system on the basis of features extracted from each which capture high-level, emergent multicellular outcomes. The goal is to find the set of parameters which minimizes discrepancy between the two sets of features. The method is first validated by demonstrating its effectiveness on synthetic data, then it is applied to calibrating a simple mechanical model of biofilm wrinkling, a common type of morphology observed in biofilms. Spatiotemporal convergence of cellular movement derived from experimental observations of different strains of Bacillus subtilis colonies is used as the basis of comparison.

Checksum

a40216158bc27c347ff14547362a0e54

Available for download on Monday, February 07, 2022

Share

COinS