A new machine learning and laboratory screening approach could help flavor researchers identify compounds that activate TRPA1, a sensory receptor associated with pungency, irritation and other chemesthetic sensations.
Published October 26, 2026, in Current Research in Food Science and authored by researchers at the China-based Zhengzhou Tobacco Research Institute of CNTC, “Integrated machine learning and in vitro pipeline for TRPA1-active chemesthetic flavor compounds: From predictive screening to molecular mechanisms” combines computational prediction with laboratory validation to streamline the discovery of sensory-active flavor molecules.
Researchers developed an optimized random forest machine learning model to predict which flavor compounds activate TRPA1, a transient receptor potential ion channel involved in detecting chemical irritation and pungent sensations. The approach addresses a key limitation of traditional screening methods: the relatively low throughput of experimental testing.
To validate the predictions, the team developed a TRPA1-expressing cell line and used calcium flux assays to measure receptor activation. Selected candidate compounds demonstrated dose-dependent activation, supporting the model's ability to identify potential TRPA1-active molecules.
The researchers also used molecular docking and molecular dynamics simulations to investigate how the compounds interact with the receptor at the molecular level, identifying key amino acid residues involved in activation.
For flavor developers, the integrated approach offers a potential pathway to screen chemesthetic ingredients more efficiently while better understanding the molecular mechanisms behind their sensory effects. Such insights could support more targeted exploration of pungency, irritation and other mouthfeel-related sensations, helping researchers connect molecular structure with sensory performance.
