Kansas State University
An organism's genome is a functional control system that, inter alia, responds to inputs from the external environment, yielding a sequence of states whose observable features are phenotypes. Our research group (comprising faculty from molecular genetics, computer science, electrical engineering, and theoretical plant modeling) is interested in developing modeling technologies capable of elucidating genotype-to-phenotype relationships in plants. We are currently focused on flowering time control in Arabidopsis thaliana and rice (Oryza sativa) as test case systems in which to develop effective modeling methodologies. We are applying a variety of machine learning tools, including symbolic regression and support vector machines, to the analysis of various example networks. We have developed gene network models capable of predicting A. thaliana flowering time across a range of photothermal environments and mathematical analyses linking critical short day lengths in rice to photoperiod-dependent expression levels of Heading Date 1 (Hd1). Additionally, we have developed new multi-objective simplex-GA hybrid optimization methods for gene network parameter estimation based on a novel concept, Fuzzy Dominance. The method out-performs existing, gold-standard algorithms such as NSGA and SPEA at this particular task.
Back to the top