Pred-O3, a web server to predict molecules, olfactory receptors and odor relationships

Abstract

The sense of smell is a biological process involving volatile molecules that interact with proteins called olfactory receptors to transmit a nervous message that allows the recognition of a perceived odor. However, the relationships between odorant molecules, olfactory receptors and odors (O3) are far from being well understood due to the combinatorial olfactory codes and large family of olfactory receptors. This is the reason why, based on 5802 odorant molecules and their annotations to 863 olfactory receptors (human) and 7029 odors and flavors annotations, a web server called Pred-O3 has been designed to provide insights into olfaction. Predictive models based on Artificial Intelligence have been developed allowing to suggest olfactory receptors and odors associated with a new molecule. In addition, based on the encoding of the odorant molecule's structure, physicochemical features related to odors and/or olfactory receptors are proposed. Finally, based on the structural models of the 98 olfactory receptors a systematic docking protocol can be applied and suggest if a molecule can bind or not to an olfactory receptor. Therefore, Pred-O3 is well suited to aid in the design of new odorant molecules and assist in fragrance research and sensory neuroscience. Pred-O3 is accessible at ' https://odor.rpbs.univ-paris-diderot.fr/'.

Graphical Abstract

Figure 1.

Overview of the available tools in Pred-O3. Users can access information in the database by searching using odorants, odors, and olfactory receptors (Search database). Chemical information can be retrieved by searching the database using similarity, measured by a Tanimoto score, or by searching for substructure (Structure search). For new compounds, the server can predict their potential odors and interacting olfactory receptors using two GNN models in Prediction toolbar. The web server is linked to the SeamDock web server, allowing the transfer of odorants and olfactory receptors from the database for docking simulations (Docking). Users have the capability to analyze the overrepresented or underrepresented functional groups in the database for a particular odor or olfactory receptor through a radar plot (Functional groups). The chemical space of compounds associated with a specific odor or interacting with a specific olfactory receptor can be visualized with an UMAP (Chemical space).