Press "Enter" to skip to content

The biology of smell and how AI is helping to solve the mystery

Of all the senses human beings have that allow us to understand the world, the sense of smell is the hardest for scientists to understand. All sensory information is captured by receptors. Our eyes have two types of receptor cells that allow us to perceive light. Our ears have small hair cells that allow us to pick up the frequencies of sound. However, our noses have over 400 receptor cells that allow us to recognize trillions of different smells. These receptors bind to odorant molecules to provide olfactory information. How these receptors work and their structures are unknown and much work needs to be done to determine how the body takes in and processes smell. Scientists at various research institutes have made advances in biology, data science, and artificial intelligence (AI) to better understand how smell works and the role it plays in memory, stress, appetite, and other factors. The hope is that these advancements can be applied to clinical settings. 

The problem with understanding smell is that it does not exist on a fixed spectrum like our other senses. “With vision, the spectrum is a simple color palette: red, green, blue and all their swirling intermediates. Sounds have a frequency and a volume, but for smell there are no obvious parameters”, says Kerri Smith of Nature science journal. There is no way to tell what molecules in the air are binding to what receptors in the nose. People are also capable of gaining vastly different opinions about what they smell and how strongly. One solution scientists have considered to help with this problem is by looking at the molecular structure of compounds to see if there is a correlation between structure and smell. As Smith explains this approach is flawed because, “the chemical structure of a molecule tells you almost nothing about its odor. Two chemicals with very similar structures can smell wildly different; and two wildly different chemical structures can produce an almost identical odor.” 

Last year, Alex Wiltschko, head of a research institute called Osmo, located in Massachusetts, collaborated with researchers at Monell Chemical Senses Center to develop an AI model to show the mechanism of molecules binding to smell receptors. “Their program was trained by feeding the model thousands of descriptions of molecular structures from fragrance catalogs, along with smell labels for each — terms such as ‘beefy’ or ‘floral’”, says Smith. They compared the AI system with human noses and trained 15 professionals to rate hundreds of smells with 55 different labels such as smoky, sweet, and waxy. This program takes into account the chemical structure and components of the molecules and correlates them to the smell descriptions given by the experts. Those involved in its development recognize the subjectivity of a smell description as a factor that could affect the validity of their program.

Students at the University of California, San Francisco have created the first 3D visual representation of how molecules bind to cell receptors. Aashish Manglik, associate professor of pharmaceutical chemistry, and his lab, “used a type of imaging called cryo-electron microscopy (cryo-EM), that allows researchers to see atomic structure and study the molecular shapes of proteins”, writes Robin Marks. Manglik collaborated with Dr. Hiroaki Matsunami of Duke University to locate, “an odorant receptor that was abundant in both the body and the nose, thinking it might be easier to make artificially”. They chose the receptor OR51E2, as this is capable of detecting water-soluble odorants. The collaborating teams were able to use their imaging to map the interaction between the chemical propionate, which gives swiss cheese its rich and nutty aroma, and OR51E2. Understanding this relationship gives us insight into how we use smell to tell when food is bad. 

Other applications for the visualization of smell include detecting harmful chemicals in the air and detecting harmful reagents in the body to detect disease. More receptors and their molecules are being studied so we can quantify how smell works as well as its relationship with our other senses, our emotions, and our memories.

Photo Courtesy of www.uclahealth.org