Machine learning creates a massive map of smelly molecules

From the article:

To a human nose, hydrogen sulfide smells like rotten eggs, geranyl acetate like roses. But the problem of guessing how a new chemical will smell without having someone sniff it has long stumped food scientists, perfumers and neuroscientists alike.

Now, in a study published in Science, researchers describe a machine-learning model that does this job. The model, called the Principal Odor Map, predicted smells for 500,000 molecules that have never been synthesized—a task that would take a human 70 years. “Our bandwidth for profiling molecules is orders of magnitude faster,” says Michigan State University food scientist Emily Mayhew, who co-led the study.

If you've read The Emperor of Scent (excerpt here), you'll get an idea of the scarcity of good science on olfaction, and the elusiveness of the relationship between chemical structures and what we perceive. Note that there's still considerable controversy about the validity of the theory of olfaction in that book, and no definitive mechanistic answers yet.

Having any kind of mapping of structure/function relationships is a big deal, and might lead to better research. The ML model is also a likely money-maker, with huge implications for the flavor and fragrance industries.

This is personally relevant simply because our household is heavily interested in cuisine and wine. Spouse is studying for his WSET exams, and he's frequently asking me, "What does this smell like to you?", while waving a glass of wine under my nose. [He's under the mistaken impression that I'm better able to describe scents and flavors accurately.]

Footnote: If you're looking for a fun activity to get through the depths of winter, this Atlas Obscura perfumery course is a great and fairly priced introduction. It's taught remotely, so you can distribute the costs among a group of family or friends, and the kit has enough material for a lot of experimenting.