Reverse-engineering human olfactory perception from chemical features of odor molecules

Keller, A.; Gerkin, R. C.; Guan, Y.; Dhurandhar, A.; Turu, G.; Szalai, B.; Mainland, J. D.; Ihara, Y.; Yu, C. W.; Wolfinger, R.; Vens, C.; Schietgat, L.; De Grave, K.; Norel, R.; DREAM Olfaction Prediction Challenge Consortium,; Stolovitzky, G.; Cecchi, G.; Vosshall, L. B.; Meyer, P.

doi:10.1101/082495

bioRxiv · 10.1101/082495

Reverse-engineering human olfactory perception from chemical features of odor molecules

Keller, A.·Gerkin, R. C.·Guan, Y.·Dhurandhar, A.·Turu, G.·Szalai, B.·Mainland, J. D.·Ihara, Y.·Yu, C. W.·Wolfinger, R.·Vens, C.·Schietgat, L.·De Grave, K.·Norel, R.·DREAM Olfaction Prediction Challenge Consortium,·Stolovitzky, G.·Cecchi, G.·Vosshall, L. B.·Meyer, P.

View original source ↗Export metadata ↓

Abstract

Despite 25 years of progress in understanding the molecular mechanisms of olfaction, it is still not possible to predict whether a given molecule will have a perceived odor, or what olfactory percept it will produce. To address this stimulus-percept problem for olfaction, we organized the crowd-sourced DREAM Olfaction Prediction Challenge. Working from a large olfactory psychophysical dataset, teams developed machine learning algorithms to predict sensory attributes of molecules based on their chemoinformatic features. The resulting models predicted odor intensity and pleasantness with high accuracy, and also successfully predicted eight semantic descriptors (\"garlic\", \"fish\", \"sweet\", \"fruit\", \"burnt\", \"spices\", \"flower\", \"sour\"). Regularized linear models performed nearly as well as random-forest-based approaches, with a predictive accuracy that closely approaches a key theoretical limit. The models presented here make it possible to predict the perceptual qualities of virtually any molecule with an impressive degree of accuracy to reverse-engineer the smell of a molecule.\n\nOne Sentence SummaryResults of a crowdsourcing competition show that it is possible to accurately predict and reverse-engineer the smell of a molecule.

Source connections

Published article ↗

Explore related subjects

neuroscience Preprint

Keep this discovery

Explore connections, maps & timelines

BibTeX RIS

Keller, A., Gerkin, R. C., Guan, Y., Dhurandhar, A., Turu, G., Szalai, B., Mainland, J. D., Ihara, Y., Yu, C. W., Wolfinger, R., Vens, C., Schietgat, L., De Grave, K., Norel, R., DREAM Olfaction Prediction Challenge Consortium,, Stolovitzky, G., Cecchi, G., Vosshall, L. B., Meyer, P.. 2016-10-21. Reverse-engineering human olfactory perception from chemical features of odor molecules. https://doi.org/10.1101/082495

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related preprints

bioRxiv2026

Sex differences in fear extinction are conditional, not categorical

Threat memories are dynamic, and their behavioral expression can change with subsequent experience and the passage of time. Extinction training can reduce conditioned fear responses, yet these responses may re-emerge over time through spontaneous recovery. Although biological sex has been implicated as modulatory factor in fear and extinction memory processes, studies have reported inconsistent findings. Here, we investigated how differences in extinction training and the timing of memory assessment influence extinction memory in male and female mice. Using contextual fear conditioning, we assessed fear acquisition and threat retention, characterized within-session and across-session extinction during repeated training, and compared shorter (3-minute) and longer (6-minute) daily extinction sessions. We subsequently evaluated extinction memory retention and spontaneous recovery at a later time point. Estrous-cycle stage in females was monitored daily throughout the experimental protocol to examine its potential association with fear and extinction behavior. We found that males and females demonstrated comparable fear acquisition and threat memory, as well as similar extinction progression across days. During longer extinction sessions, but not shorter, males showed greater within-session reductions in freezing, whereas females exhibited modestly higher overall freezing. We found no significant sex difference in the magnitude of spontaneous recovery. There was no significant association between estrous-cycle stage and freezing behavior. Together, our findings suggest that sex-dependent differences in contextual fear extinction are not uniformly expressed, but rather depend on the experimental conditions under which extinction is assessed

neuroscience↗

bioRxiv2026

Endothelin-1 from Tongue Glands Gates 1 Sodium Taste and Intake

Sodium is an essential nutrient. To survive sodium deficiency, animals dramatically increase their appetite for salt. However, whether and how sodium depletion impacts the taste of salt to drive sodium intake is unknown. Here we show that tongue gland-derived endothelin-1 (EDN1) regulates the amiloride-sensitive, sodium-selective salt taste. Edn1 is specifically expressed in the glands embedded within the tongue. Under normal conditions, EDN1 suppresses salt taste by acting on its receptor EDNRB on the projections of salt-sensing taste neurons innervating taste buds. When internal sodium levels drop, EDN1 expression decreases, lifting this inhibition and enhancing responses to salt. Applying EDN1 directly to the tongue, using a nonionic eutectic-based organogel (Neogel) delivery technology that we establish here, dampens the activity of the salt-sensing taste neurons and reduces salt intake and preference. Conversely, genetic deletion of Ednrb in taste neurons mimics the effects of sodium deficiency, enhancing salt-sensing neuron activity and increasing salt consumption. These findings reveal a peripheral mechanism that regulates salt taste and intake in response to internal sodium levels, highlighting a direct role of the taste pathway in maintaining nutrient homeostasis.

neuroscience↗

bioRxiv2026

In-vivo MRI of noradrenergic and dopaminergic nuclei predicts future cognition and Alzheimers disease progression

Noradrenergic and dopaminergic nuclei modulate cognition, and post-mortem data indicate they are affected early in the neurodegenerative cascade of Alzheimers disease. We derived in-vivo locus coeruleus (LC) and substantia nigra-ventral tegmental area (SN-VTA) MRI integrity measures from 739 ADNI participants, examined MRI-histology correspondence in individuals with both in-vivo imaging and post-mortem data, and tested whether these measures reflect early MRI-indexed degeneration predicting later cognitive decline. LC and SN-VTA contrast predicted cognitive and behavioral function 10 years later, despite only weak associations with current function. LC- and SN-VTA-associated brain-wide atrophy patterns aligned with PET-derived norepinephrine-transporter and dopamine-receptor maps, respectively. A one-standard-deviation decrease in baseline LC contrast was associated with 84% higher five-year odds of MCI-to-AD progression, and predictive effects of the catecholaminergic nuclei were mediated by their corresponding atrophy patterns. The SN-VTA pattern was associated with less amyloid PET-related memory decline. Together, these findings establish LC and SN-VTA MRI contrast as markers of early neurodegeneration that forecast cognitive decline years before clinical onset.

neuroscience↗