Emily J. Mayhew

The paper shows that odor mixtures can be predicted linearly from component odorant profiles, challenging the assumption of nonlinear interactions.

The paper develops a Principal Odor Map using graph neural networks that matches human-level odor description performance and generalizes to diverse odor prediction tasks.

The paper develops a model to predict odorous molecules based on chemical properties and estimates the number of possible odorous compounds to be over 30 billion.