Odour coding in a virtual Cybernose

We will study the sensitivity of the receptors to pure chemicals and against natural backgrounds. We will then assemble a virtual Cybernose with various combinations of these receptors and analyse its potential in decoding specific odour problems. To do this we will develop software tools that can answer specific questions. For instance can a Cybernose with DmelOr43b, DanaOr43b and DmelOr22a channels detect explosives in the context of airport security? Which is the minimal set of receptors that can reliably detect Brett off-flavours in various wine samples? We will also study algorithms that decode the output from such a set of detectors and determine how reliable they are at detecting specific volatiles.
We will determine a reliable way of estimating the activity of odorants on each of these receptors. Measuring dose response curves of Drosophila receptors has established that the relation can be modelled as a Hill curve characterised by a coefficient for the slope and an EC50 representing affinity/
Each receptor will be characterised as a vector in an n-dimensional odour space where n is the number of odorants tested and the direction of the vector represents odour selectively or tuning width while the length of the vector indicates sensitivity.
We will characterise odour space using molecular descriptors and explore possible correlation between receptor tunning and physico-chemical properties of the odour molecules