AUTOMATING ION CHANNEL MODEL DEVELOPMENT

Abstract

As ion channels enjoy ubiquitous importance in living cells, the characterisation of their electrical activity plays a salient role in virtually all physiologically-related fields. Automated patch-clamp electrophysiological platforms have emerged as high-throughput frontrunners for measuring ion channel activity. However, their measurements remain underutilised in multiscale modelling due to 1) large volumes of generated data 2) the lack of a suitable model capable of integrating both structural and functional information, whilst maintaining computational effi ciency and statistical soundness. This thesis investigates the feasibility of developing such a model using the Manifest Interconductance Rank form and Eyring Rate Theory. Unlike previous models, this model is designed to continually evolve, by automatically 'learning' from the ever-increasing data generated from high-throughput technologies. Through simulation and optimisation experiments, crucial modelling constraints have been identified. A novel goodness-of-fit criterion necessary for objective assessment of the evolving non-linear models has also been created.