A biomimetic olfactory recognition system for the discrimination of Chinese liquor aromas

Aroma is an important attribute influencing the perceived quality of Chinese liquors, with each liquor characterized by a unique collection of volatile chemicals. Here, a biomimetic olfactory recognition system combining an optimal panel of 10 mouse odorant receptors with back propagation neural network model was designed to discriminate the aromas of Chinese liquors. Our system shows an excellent predictive capacity with an average accuracy of 96.5% to discriminate liquors of different aroma types, as well as those of different brands and ageing years within the same aroma type. A total of 124 interactions between liquor aroma compounds and odorant receptors were further elucidated to understand odorant coding at the molecular level, including 14 newly deorphaned odorant receptors. Our work represents a proof of concept for combining receptors and machine learning in the discrimination of complex odorant stimuli.