The biology of smell is a mystery. AI helps to solve it!

Scientists begin to solve an incredibly complex code that helps us smell.

The smell in the laboratory was new. In business terms, it was persistent: for more than a week the smell was kept on the paper on which it was soaked.

For researcher Alex Wilchko, it was the smell of summer time in Texas: watermelon, or rather, the border where the red pulp turns into a white skin.

"It was a molecule that no one had ever seen before," says Wilchko, manager of Osmo, based in Cambridge, Massachusetts. His team created a compound called 533 as part of its mission to understand and digitize smells. His goal is to develop a system that can detect, predict or create smells - this is not an easy task, as shown by molecule 533. "If you had looked at the structure, you would never have guessed that it smelled like that."

This is one of the problems with understanding the smell: the chemical structure of the molecule says almost nothing about its smell. Two chemicals with a very similar structure can have a completely different smell; and two completely different chemical structures can produce an almost identical smell. And most smells - coffee, camembert, ripe tomatoes - are mixtures of many dozens or hundreds of aromatic molecules, which complicates the task of understanding how chemistry generates olfactory experience.

Another problem is to find out how smells are related to each other. For vision, the spectrum is a simple color palette: red, green, blue and all their intermediate links. Sounds have a frequency and volume, but there are no clear parameters for the smell. What is the place of the smell identified as "frost" in relation to the "sauna"? Making predictions about smell is a real problem, says Joel Mainland, a neurobiologist at the Monell Chemical Senses Center, an independent research institute in Philadelphia, Pennsylvania.

Animals, including humans, have developed an extremely complex decoding system corresponding to a huge set of smell molecules. All sensory information is processed by receptors, and the smell is no different - except for its scale. For light, the human eye has two types of receptor cells; smell - 400. How the signals from these receptors combine to cause a certain perception is unclear. In addition, it is difficult to work with receptor proteins themselves, so how they look and how they function remains for the most part a guess.

However, the situation is beginning to change due to improvements in structural biology, data analysis and artificial intelligence (AI). Many scientists hope that deciphering the olfator code will help them understand how animals use this important feeling to find food or partners and how it affects memory, emotions, stress, appetite and much more.

Others are trying to digitize smells to create new technologies: devices that diagnose smell-based diseases; better and safer insect repellents; and affordable or more effective aromatic molecules for the $30 billion flavoring and flavoring market. At least 20 start-up firms are trying to create electronic noses for use in health and public safety.

All this leads to a surge in research in the field of smell biology, says Sandeep Robert Dutta, a neurobiologist at Harvard Medical School in Boston, Massachusetts. "The smell has a moment," he says.