New tools and specification languages for biophysically detailed neuronal network modelling

Increasingly detailed data are being gathered on the molecular, electrical and anatomical properties of neuronal systems both in vitro and in vivo. These range from the kinetic properties and distribution of ion channels, synaptic plasticity mechanisms, electrical activity in neurons, and detailed anatomical connectivity within neuronal microcircuits from connectomics data. Publications describing these experimental results often set them in the context of higher level network behaviour. Biophysically detailed computational modelling provides a framework for consolidating these data, for quantifying the assumptions about underlying biological mechanisms, and for ensuring consistency in the explanation of the phenomena across scales. Such multiscale biophysically detailed models are not currently in wide- spread use by the experimental neuroscience community however. Reasons for this include the relative inaccessibility of software for creating these models, the range of specialised scripting languages used by the available simulators, and the difficulty in creating and managing large scale network simulations. This thesis describes new solutions to facilitate the creation, simulation, analysis and reuse of biophysically detailed neuronal models. The graphical application neuroConstruct allows detailed cell and network models to be built in 3D, and run on multiple simulation platforms without detailed programming knowledge. NeuroML is a simulator independent language for describing models containing detailed neuronal morphologies, ion channels, synapses, and 3D network connectivity. New solutions have also been developed for creating and analysing network models at much closer to biological scale on high performance computing platforms. A number of detailed neocortical, cerebellar and hippocampal models have been converted for use with these tools. The tools and models I have developed have already started to be used for original scientific research. It is hoped that this work will lead to a more solid foundation for creating, validating, simulating and sharing ever more realistic models of neurons and networks