Latest Machine Learning Research Uncovers a Hidden Order in Scents

Alex Wiltschko is an olfactory neuroscientist for Google Research’s Brain Team. As a youth, his fascination with fragrances. He recently employed machine learning to analyze their oldest and least known sense of smell. Their discoveries considerably increased scientists’ capacity to determine a molecule’s scent from its structure.

Over 800 chemicals reach your nose when you smell coffee. Our brains create the general impression of coffee from this chemical image. Scientists are just starting to understand how many 400 receptors in our noses can interact with a specific molecule to detect its chemical composition. Building models that could infer a molecule’s smell from its structure was a point of competition for teams.

Even the finest models of smell can only account for some things. Sometimes, little changes to a molecule’s chemical makeup result in a completely different smell. In other cases, substantial structural alterations have little or no effect on the smell.

A Smell’s Metabolic Organization

Wiltschko and his team investigated the demands that evolution might have placed on our senses to explain these unusual occurrences. Over millions of years, each sense has been fine-tuned to pick up the most salient variety of stimuli.  

“The core metabolic engine inside every living thing is the one thing that’s been constant over evolutionary time, at least from a very long time ago. The term “metabolism” refers to a collection of chemical processes, such as the Krebs cycle, glycolysis, urea cycle, and several others, which are catalyzed by cellular enzymes and change one molecule into another in living things. These well-established reaction pathways outline the connections between the naturally occurring substances that enter our nostrils. Substances with similar smells are related biologically and chemically.

His team required a map of the metabolic processes that take place in nature to test the theory. These natural chemical interactions and the enzymes that cause them were fortunately already described in a sizable database that metabolomics researchers had built. Using this information, the researchers could determine how many enzymatic processes would be required to change one odorous molecule into another.

They also required a computer model that could calculate how different odorous molecules smell to humans for comparison purposes. The research team has been working to improve a neural network model known as the primary odor map based on the outcomes of the 2015 DREAM competition. This map resembles a cloud of 5,000 points, each representing a different smell molecule. The points for molecules with similar smells group together, whereas those with significantly dissimilar smells are spread apart. Only cutting-edge computer technologies can comprehend the structure of the cloud since it is considerably more than 3D—it has 256 dimensions of information.

Within the two data sources, the researchers searched for corresponding correlations. They examined 50 pairings of molecules and discovered that, despite having quite different structures, compounds closer to the metabolic map tended to be more relative to the fragrance map.

Researchers discovered that while the forecasts were still not exact, they were still better than any prior model that had been able to predict solely on chemical structure. That wasn’t necessary at all, he said. “Two biologically similar compounds, one enzyme catalysis step apart, they could smell like rotten eggs and roses.”  The scientists also discovered that molecules with a natural relationship, such as the many chemical components of an orange, tend to smell more alike than ones without one.

Nature-Attuned Chemistry

“Researchers said that the olfactory system is designed to pick up on different [chemical] coincidences. Therefore, metabolism controls the likelihood of coincidences. This suggests that the metabolic process by which a molecule was created in the natural world matters to our noses in addition to its chemical structure.

“The olfactory system is adjusted for the cosmos it perceives, which consists of these molecular structures. and a big part of that is how these molecules are created. They commended the smartness of the plan to classify odors further utilizing metabolism. Since a molecule’s metabolic origin is already tightly tied to its structure, the metabolism-based map only improves a little on structural models, but it does contribute some more information.

Instead of single molecules, the Researcher said that the smells of mixes would represent the next frontier in olfactory neuroscience. Consider the hundreds of chemicals that are emitted by your coffee mug. In real life, they hardly ever breathe in just one chemical at a time. There needs to be more information on odorant mixes for researchers to create a model similar to the one employed in the current study for pure compounds. To fully comprehend our sense of smell, it will be necessary to look at how chemical combinations combine to create complex smells.

Check out the paper and reference article. All Credit For This Research Goes To Researchers on This Project. Also, don’t forget to join our Reddit page and discord channel, where we share the latest AI research news, cool AI projects, and more.