Machine Learning for Scent: Learning Generalizable Perceptual Representations of Small Molecules

TL;DR

This paper introduces graph neural networks to predict the relationship between molecular structure and odor, significantly outperforming previous methods and capturing meaningful odor representations for transfer learning.

Contribution

The study demonstrates the effectiveness of graph neural networks for QSOR modeling, providing a novel approach and a new dataset labeled by olfactory experts.

Findings

GNNs outperform prior QSOR methods on the new dataset.

Learned embeddings capture meaningful odor space representations.

Strong transfer learning performance on two challenging tasks.

Abstract

Predicting the relationship between a molecule's structure and its odor remains a difficult, decades-old task. This problem, termed quantitative structure-odor relationship (QSOR) modeling, is an important challenge in chemistry, impacting human nutrition, manufacture of synthetic fragrance, the environment, and sensory neuroscience. We propose the use of graph neural networks for QSOR, and show they significantly out-perform prior methods on a novel data set labeled by olfactory experts. Additional analysis shows that the learned embeddings from graph neural networks capture a meaningful odor space representation of the underlying relationship between structure and odor, as demonstrated by strong performance on two challenging transfer learning tasks. Machine learning has already had a large impact on the senses of sight and sound. Based on these early results with graph neural networks for molecular properties, we hope machine learning can eventually do for olfaction what it has already done for vision and hearing.