Response Prediction of an Insect’s Olfactory Receptor Neuron by Using Structural Parameters of Odorant and Self-Organizing Map

Abstract

Our group has, for years, been studying odor approximation to express a variety of odors using a small number of odor components. The insect’s olfaction is more appropriate for systematic survey for odor approximation than mammal’s one because of its database’s availability. Thus, we propose a prediction method of Drosophila’s olfactory receptor neuron (ORN) response from the structural parameters of an odorant molecule and the reported data of ORN response using self-organizing map (SOM). We obtained the structural parameters of odorant molecules by using Dragon (Talete s.r.l.). Then, the SOM maps the structural parameters of an odorant molecule onto an OR response to the corresponding odor. For improving accuracy, we selected important parameters based on Akaike information criteria. Then, the simulation suggests that the prediction can be performed only using a few tens of parameters.