iBitter-SCM

iBitter-SCM predicts and characterizes bitter peptides from amino acid sequences to support identification of peptides relevant to drug development and nutritional research.

Key Features:

• Scoring Card Method (SCM): Uses estimated propensity scores for amino acids and dipeptides to enable direct prediction without reliance on functional domain or structural information.
• Accuracy and Performance: Achieves 84.38% accuracy and a Matthews correlation coefficient of 0.688 on independent datasets.
• Comparative Superiority: Outperforms k-nearest neighbor, naive Bayes, decision tree, and random forest classifiers while offering a simple and interpretable scoring-based approach.

Scientific Applications:

• Drug Development: Identifies bitter peptides to inform design of therapeutic peptides with reduced bitterness.
• Nutritional Research: Profiles peptide bitterness to aid development of more palatable food products and supplements.

Methodology:

Applies the scoring card method (SCM) with estimated amino acid and dipeptide propensity scores and analyzes derived biophysical and biochemical properties; performance was evaluated on independent datasets and compared against k-nearest neighbor, naive Bayes, decision tree, and random forest classifiers.