VmmScore: An umami peptide prediction and receptor matching program based on a deep learning approach.

Area of Science:

• Biochemistry
• Computational Biology
• Sensory Science

Background:

• Peptides possess physiological and medical benefits, including blood pressure and lipid level reduction.
• Understanding umami taste perception is crucial for various applications.
• Identifying specific peptide-receptor interactions for umami taste is complex.

Purpose of the Study:

• To develop a computational strategy for identifying optimal umami receptors for peptides.
• To introduce the VmmScore algorithm, comprising Mlp4Umami and mm-Score modules.
• To streamline the process of umami receptor determination for peptides.

Main Methods:

• Developed VmmScore algorithm with Mlp4Umami (predicts umami taste potential) and mm-Score (machine learning-optimized molecular docking).
• Optimized docking structures, clustered umami peptides, and analyzed docking energies.
• Performed virtual screening of peptides from Lateolabrax japonicus.

Main Results:

• Successfully identified and experimentally verified the umami taste of three peptides from Lateolabrax japonicus.
• Determined the specific receptors corresponding to these verified umami peptides.
• Demonstrated the VmmScore algorithm's efficacy in rapid and cost-effective peptide screening.

Conclusions:

• The VmmScore algorithm provides a strategic machine learning approach for umami receptor determination.
• This research advances the understanding of umami taste perception mechanisms.
• The publicly accessible source code facilitates further research and collaboration.