| The synapse is the basic unit of computation in the brain. Its ability to adapt its strength on a millisecond timescale is a prerequisite for information processing in neuronal networks. This short-term synaptic plasticity is largely determined by presynaptic processes that regulate neurotransmitter release and involve many different genes. In recent years, most of these genes were studied intensively on the expression-, interaction-, and functional level, but a mechanistic understanding of the underlying gene network is still lacking. The general aim of this proposal is to apply machine learning techniques to construct a dynamic model for the presynaptic gene network underlying short-term plasticity in neuronal synapses. We propose to develop a novel Bayesian perturbation approach, which combines a computational representation of the system with controlled genetic perturbations in a Bayesian framework. A Bayesian formulation of the problem allows evaluation of different levels of information, simultaneous testing of multiple models and design of informative experiments. In the project a large group of alternative hypothesis for vesicle release will be tested. Life scientists and computer scientists work closely together in an iterative scheme of model selection and model validation which eventually converges to a realistic model. This model will provide a quantitative description of computation at the synaptic level which is essential for understanding information processing at the network level. It will greatly facilitate new insights in how the brain functions in health and disease and inspire the fields of artificial intelligence and machine learning in finding new algorithms for information processing. |
