QSAR-Guided Generative Framework for the Discovery of Synthetically Viable Odorants

TL;DR

The paper tackles odorant discovery within vast chemical spaces under data scarcity by proposing a QSAR-guided variational autoencoder (VAE) that learns SMILES grammar from large general databases (ChemBL) while steering the latent space with an odor-predictive QSAR head. The integrated model generates syntactically valid, novel odorants with practical synthetic feasibility, validated against an external ground truth (Unique Good Scents) and supported by synthesis planning (AiZynthFinder), thermodynamic stability (GFN2-xTB), and ADMET safety profiling. Key findings include 100% validity, 74.4% uncharted scaffold hops, and latent-space organization by odor relevance (low Fréchet ChemNet Distance to odorant manifolds; strong agreement between QSAR and VAE odor predictions). The approach delivers an industrially actionable, data-efficient path to de novo odorants, bridging molecular generation with real-world synthesis and safety considerations for fragrance and flavor applications.

Abstract

The discovery of novel odorant molecules is key for the fragrance and flavor industries, yet efficiently navigating the vast chemical space to identify structures with desirable olfactory properties remains a significant challenge. Generative artificial intelligence offers a promising approach for \textit{de novo} molecular design but typically requires large sets of molecules to learn from. To address this problem, we present a framework combining a variational autoencoder (VAE) with a quantitative structure-activity relationship (QSAR) model to generate novel odorants from limited training sets of odor molecules. The self-supervised learning capabilities of the VAE allow it to learn SMILES grammar from ChemBL database, while its training objective is augmented with a loss term derived from an external QSAR model to structure the latent representation according to odor probability. While the VAE demonstrated high internal consistency in learning the QSAR supervision signal, validation against an external, unseen ground truth dataset (Unique Good Scents) confirms the model generates syntactically valid structures (100\% validity achieved via rejection sampling) and 94.8\% unique structures. The latent space is effectively structured by odor likelihood, evidenced by a Fréchet ChemNet Distance (FCD) of $\approx$ 6.96 between generated molecules and known odorants, compared to $\approx$ 21.6 for the ChemBL baseline. Structural analysis via Bemis-Murcko scaffolds reveals that 74.4\% of candidates possess novel core frameworks distinct from the training data, indicating the model performs extensive chemical space exploration beyond simple derivatization of known odorants. Generated candidates display physicochemical properties ....

Figures (9)

• Figure 1: Variational Autoencoder framework concept. Abbreviations: CNN, convolutional neural network; GRU, gated recurrent unit; FC, fully connected (dense); KL, Kullback-Leibler divergence; Recon Loss, reconstruction loss (categorical cross-entropy); Prop Loss, property prediction loss (mean squared error, as described in text). The diagram illustrates the flow: input SMILES from ChemBL are processed by an Encoder (CNN based) mapping to a continuous latent space. Latent vectors $z$ are input to both a Decoder (GRU based) for reconstructing SMILES sequences and an Odor Property Prediction head (FC based) predicting odor probability $P(\text{odor})$. The combined training objective minimizes the Recon Loss, the Prop Loss, and the KL divergence regularization term.
• Figure 2: QSAR Odor molecule prediction results. a) Schematic showing the QSAR model pipeline where logistic regression was trained on 1,924 experimentally validated and/or literature reported molecules and subsequently used to label ChemBL molecules with a target odor probability, b) UMAP visualization of training set in space of molecular descriptors showing the odor probability, c) classifier performance for logistic model, and d) Distribution of predicted probabilities across the training set, and e) residual plots for the logistic regression model.
• Figure 3: a) UMAP dimensionality reduction and visualization of the molecular descriptor space showing 5 x $10^5$ VAE training set ChemBL molecules (green) alongside the original set of odor molecules (red) and non-odor molecules (blue) and the same UMAP representation showing the predicted probabilities across all ChemBL molecules used for VAE training as predicted by the QSAR model (right). Inset histogram for the frequency of odor probabilities in the ChemBL set used to train the VAE, b) t-SNE dimensionality reduction of 1 x $10^4$ randomly sampled ChemBL molecules encoded into the latent space (green) together with validated odor (red) and non-odor molecules (blue), c) Validated odor molecule probabilities as predicted by the QSAR model, $P(\text{qsar})$, compared against VAE odor property prediction, $P(\text{odor})$. Shaded region shows where both models agree on odor molecule prediction ($P(\text{odor}) > 0.5$ and $P(\text{qsar}) > 0.5$).
• Figure 4: Generated molecules from four example seed odor molecules. a & b) Two example seed molecules (seed is distance 0.0, top left) with sampling of the latent space over increasing distance from the seed in latent space (units of $\sigma_{\phi}(z|X)$). Where available Pubchem catalogued compounds show their corresponding IUPAC name and CID reference, whereas potentially novel compounds show the corresponding SMILES and CID: N/A. SA: synthesis accessibility score. c) traversing the latent space between two odor seed molecules $t = 0$ is start seed and $t = 1$ is the end seed in 11 discrete steps. Attempts are the number of probabilistic decode cycles required to obtain a valid structure at that point of the latent space.
• Figure 5: Generated molecules' uniqueness. a) Jaccard similarity, and b) Frechet ChemNET distance across data-sets.