The study focuses on using machine learning to identify pleasant smelling insect repellents by modeling the valence of volatiles for insects and humans. By training models on chemical structures, the researchers successfully predicted the aversive valence of test chemicals with high accuracy. They also prioritized candidate repellents based on their predicted odor qualities for humans. Experimental validation showed that most predicted compounds exhibited strong repellency to mosquitoes and Drosophila flies, indicating a physicochemical basis for odor valence across species. The study highlights the potential of machine learning in accelerating the discovery of novel insect repellents with desirable fragrance profiles.
To Another Language
from source content
biorxiv.org
Key Insights Distilled From
by Kowalewski,J... at www.biorxiv.org 12-26-2023
https://www.biorxiv.org/content/10.1101/2023.12.25.573309v1Deeper Inquiries