Quantitative Structure-Activity Relationship Study of Bitter Di-, Tri- and Tetrapeptides Using Integrated Descriptors

Area of Science:

• Food Science
• Computational Chemistry
• Biochemistry

Background:

• Peptide bitterness is a significant factor in food and pharmaceutical applications.
• Predicting peptide taste requires understanding structure-taste relationships.
• Existing models may not fully capture the nuances of bitterness across different peptide lengths.

Purpose of the Study:

• To develop novel quantitative structure-activity relationship (QSAR) models for predicting bitterness in dipeptides, tripeptides, and tetrapeptides.
• To identify key amino acid descriptors that significantly influence peptide bitterness.
• To elucidate the specific structural determinants of bitterness for varying peptide lengths.

Main Methods:

• Integrated 14 amino acid descriptor sets for QSAR model development.
• Utilized a bootstrapping soft shrinkage approach for robust variable selection.
• Evaluated variable importance using selection frequency and standardized regression coefficients.

Main Results:

• Achieved high model performance with R² and Q² values for dipeptides (0.950, 0.941), tripeptides (0.770, 0.742), and tetrapeptides (0.972, 0.956).
• Identified C-terminal hydrophobicity as crucial for dipeptide bitterness.
• Determined that hydrophobicity and electronic properties at specific positions are key for tripeptide and tetrapeptide bitterness.

Conclusions:

• Developed reliable QSAR models for predicting peptide bitterness across different lengths.
• Established a deeper understanding of the structure-bitterness relationships in peptides.
• Provided insights for designing peptides with modulated taste profiles.