An unrivaled multidisciplinary nonprofit
The Monell Chemical Senses Center is an unrivaled multidisciplinary nonprofit research institution leading the world to a better understanding of taste and smell through research, training, and communication.Every day, Monell scientists make discoveries about how our bodies detect and respond to chemical molecules in the worlds within and around us. The Center’s work guides policies and inspires products that enable healthier diets, safer environments, and more powerful ways to meet pressing health challenges— from obesity to infectious diseases to Alzheimer’s dementia.
Thanks to 50 years of unrestricted support from the Monell Foundation, Monell Center scientists have built and continue to expand a remarkable legacy of scientific breakthroughs that translate to real-world advances in policy, practice and behavior.
Dreams are the seedlings of realities – James Allen
Dreaming of Smells
Most people say they don’t experience smells in their dreams, but Monell’s Joel Mainland is not like most people. Over the past several years, Mainland has devoted a fair bit of time to dreaming of smells. That’s because the olfactory neurobiologist, along with two members of his lab, spent that time contributing to the DREAM Olfaction Prediction Challenge, working with scientists around the globe to better understand how the brain transforms information from chemical molecules into the perception of a smell.
Recognized as a leader in the fields of olfactory psychophysics (the study of the relationship between physical stimuli and the sensations they evoke) and molecular genetics, Mainland’s research focuses on understanding how the brain translates airborne chemical molecules into what we perceive as odors.
Somewhat surprisingly, how we distinguish odors is a pervasive puzzle that has yet to be solved. When it comes to perception, the senses of sight and hearing are orderly systems based on wave characteristics. Scientists long ago figured out that we can predict color perception by knowing the wavelength of light when it hits our retina: 700 nanomenters (nm) will be blue, 470nm will be red. For hearing, a soundwave frequency of 261 Hz (waves/second) is perceived as middle C.
But olfaction, the sense of smell, is a seemingly unruly chemical sense that responds to an undetermined number of molecules. Furthermore, no one knows how to predict what a given molecule will smell like.
The big knowledge gap consumes Mainland. “We don’t understand how the brain interprets olfactory information. That is a huge hole in our basic knowledge right now,” says Mainland, who speaks in a rapid-fire cadence and comes across as simultaneously relaxed and intense. That’s why Mainland joined with scientists from around the world to collectively try to fill that hole and also why he’s so enthusiastic about the outcome. “Many scientists thought that this was almost an insurmountable problem. But now the DREAM Olfaction Challenge has shown us that actually big chunks of this problem are pretty solvable. So it turns out that it’s not as hard as we previously thought,” he says.
Using Machines That Learn to Dream?
No one knows exactly how many different kinds of molecules humans can smell – it likely numbers in the millions. That said, probably millions more molecules exist that don’t have an associated odor. Wanting to understand the chemical features that predict whether a given molecule does or does not have an associated smell, Mainland turned to a field of artificial intelligence known as machine learning, where computers examine patterns in large data sets so they can ‘learn’ to make predictions.
“My lab was asking, can you use the chemical structure of any given molecule to tell whether it has an odor?” said Mainland. “Then the DREAM Challenge came along, and they were using these same molecular characteristics to ask a much more complex question – that is, what does the molecule smell like?”
DREAM, shorthand for Dialogue for Reverse Engineering Assessments and Methods, goes beyond its acronym to represent the goal of using data sharing and open science to answer increasingly complex questions in biology and translational medicine. Run by researchers from diverse organizations, the Challenges use crowdsourcing to attack big questions and identify those solutions having the greatest impact on human health. For example, one DREAM Challenge sought to identify genetic traits that contribute to cancer cell viability, while another aimed to improve the accuracy of digital mammograms.
The DREAM Olfaction Prediction Challenge sought to advance understanding of how the brain interprets smells by taking advantage of a large set of perceptual data previously collected by scientists at the Rockefeller University in New York City. To construct the dataset, the Rockefeller researchers asked 49 human subjects to smell 476 different molecules and rate each for intensity (how strong), valence (how pleasant), and quality (from a list of 19 different descriptors, including sweet, burnt, fruit, grass, musky, wood).
Challenge participants were given these data, along with a list of 4,884 physical-chemical features for each of the different molecules assessed by the subjects. Their goal was to use the combined perceptual and molecular data to build models that could accurately predict how any given molecule will smell.
Sensing an opportunity to utilize modeling approaches already in place in his lab, Mainland convened a meeting to talk about the challenge with members of his team, including visiting scientist Yusuke Ihara, PhD, and Research Analyst Wendy Yu. “We discussed different machine learning strategies and then jumped in to start working on models,” he recalls.
Attacking the DREAM
A total of 22 teams, including two from Monell, tackled the massive problem, using machine learning procedures to build computer models and test their accuracy. Feedback from the Challenge leaderboards soon told the Monell teams that other teams were outperforming them. “This actually was a big motivation for us to push ourselves to try new things, asking ourselves how we could improve what we were doing,” recalls Mainland.
After the Challenge results were announced in June 2015 (with the Monell researchers placing near the middle of the pack), many of the teams worked together to help improve the winning models. Because many contestants were data scientists with little interest or expertise in olfaction per se, this collaborative phase is where Mainland believes that he and his lab helped to strengthen the models through their expertise in olfactory neuroscience and perception. In one example involving an amino acid called cysteine, they noted that a smell the subjects rated was due to an impurity, and not to cysteine itself — thus weakening the model. The Monell researchers also were able to interpret some rules that the model had developed (“without any training from us,” notes Mainland) to make its predictions.
On the flip side, Mainland’s own research benefited when one of the other teams started describing the various molecules using their ‘fingerprints’, a way of quantifying shared molecular substructures. “These features really improved models for certain descriptors, and we now use them for other modeling efforts in the lab,” he says.
At the end, the collaborative model was able to predict a given molecule’s odor pleasantness and intensity fairly well, with prediction scores of about 7 or 8 on a scale of 1-10 (with 10 being the most accurate). The odor qualities – whether the molecule smelled like descriptors such as grass, bakery, or decayed – were more difficult to predict.
The team and collaborative results from the DREAM Olfaction Prediction Challenge were published in February 2017 in the journal Science.
Following the DREAM
Working on the DREAM Challenge turned out to be a transformative experience for Research Analyst Wendy Yu. Possessing a Master’s degree in Biotechnology, Yu already was immersed in data science as project lead for the Mainland lab’s effort to understand how a molecule’s chemical properties confers an odor. But, working as part of Mainland’s team on the DREAM Challenge allowed her to discover a true passion for machine learning. Following her dream, Yu moved to New York to enroll in an intensive data science bootcamp and now works as a data analytics manager at a pharmaceutical company, where she uses machine learning models to help improve the quality of clinical trials.
Expanding the DREAM
“The models we created were not perfect, but this challenge showed that you can jump straight from molecule to odor perception reasonably well,” says Mainland. “This tells us broad themes exist that we can use to understand olfactory perception.”
With one very large hurdle at least partially cleared, Mainland is seeking funding to allow him to refine the DREAM Challenge’s approach to predicting odor quality. Noting that most of us cannot distinguish between “musky” and “sweaty” or between “sandalwood” and “oak,” he next wants to ask perfumers and flavorists, professional raters trained to recognize and differentiate thousands of odors, to assess a wider range of odor qualities.
Joel Mainland continues to dream big. “Once we understand how the brain is encoding all this information, we can use that knowledge to create a way to digitize odors. Imagine being able to send a scent signal over the internet – that would open up a whole new way for us to communicate with one another,” he says. “Right now, we don’t even have the words to imagine what that might be like.”