The odour world is vast; research suggests that the human olfactory bulb can distinguish over a trillion odours. By mapping the scent of molecules and digitizing the sense of smell, Alex Wiltschko, Richard Gerkin, and the Google team have accomplished a remarkable feat. Moreover, by doing so, the team opens up the possibility of discovering new odours and molecules. 

The researchers validated their model against new molecules, linked their findings to biological odour mechanisms, and expanded their results to find new approaches to global health challenges. This transformational research's real-world applications are beyond exciting.

Molecules that travel through the air create odours. It is challenging to determine which molecules are responsible for particular scents because there are potentially billions of molecules that can produce an odour. We can resolve this issue with molecular maps. However, the absence of good olfactory "cameras" and "monitors" makes them more challenging to make.

The names of the smells that various molecules are said to evoke, such as "meaty," "floral," or "mint," were combined with thousands of examples of those molecules in a graph neural network (GNN) model that Google AI created in 2019. This approach was necessary to study the correlation between a molecule's structure and its likelihood of having a specific odour label. 

The POM displayed pairs of odours perceived similarly as close-by points with a similar hue. The Google AI researchers show how we can use the map to understand these properties in basic biology, predict the future odour properties of molecules, and address urgent global health issues. Several tests have already run on the map.

Test 1: Testing with molecules that did not correlate with odours

The researchers tested the fundamental model to see if it could accurately predict the smells of novel molecules they didn't include in its development.

Image source: Google AI

Test 2: Linking odour quality to fundamental biology

The researchers tested the odour map to see if it could predict animal odour perception and the underlying brain activity. They discovered that it accurately predicted the behaviour of most of the animals studied by neurobiologists, including mice and insects, and the activity of sensory receptors, neurons, and synapses.

Test 3: To address the global health problem

The odour map opens up new possibilities because it closely relates to animal perception and biology. They chose to retrain the POM to manage one of humanity's most significant issues: the spread of diseases carried by ticks and mosquitoes. The POM can be used to predict animal olfaction in general.

Conclusion

In laboratory studies, many molecules showed repellency when used on humans. Some have proven to be more effective than the current most popular repellents (DETA and picaridin) Furthermore, in experimental testing with brand-new molecules, Filter found that more than a dozen had insect repellency at least as effective as DETA, the main component in most insect repellents. So, for instance, less expensive, longer-lasting, and safe repellents will help reduce malaria.