Emergent models in a multi-level biochemical network regulating pea flowering
Jacob Stolk, Jim Hanan · 2009
Abstract: The Emergent Models methodology (EM) is an adaptive computational method for discovering models of complex systems in computer simulations (Stolk 2005). EM uses machine learning and optimisation algorithms such as genetic programming. Stolk and Hanan (2007) used EM to discover genetic regulatory network models of branching in Pisum sativum (pea). Here EM is used to discover models of genetic and metabolic networks regulating flowering in pea. These models describe multiple levels and components in the whole plant complex system, including genes, intercellular signals, modules and phenotype. Flowering in pea is determined by genes and mobile signals, mediating environmental influences such as photoperiod. Models of biochemical mechanisms explaining flowering time of pea studied here incorporate modules such as a circadian clock, signal processors and switching mechanisms. Each module is a combination of chemical reactions. Three hierarchical system levels are involved: the top level of the whole plant (phenotype); a middle level of modules; a bottom level of chemical reactions. It was hypothesised that models describing each level could be automatically discovered by genetic programming, given data on the next higher level. Discovered models should predict experimental data on gene expression and flowering time of wild type and several mutant pea plants. The purpose of this research