iAMAP-SCM

iAMAP-SCM predicts antimalarial peptides (AMAPs) from peptide sequence information using an interpretable scoring card model that estimates amino acid and dipeptide propensities for large-scale identification and characterization.

Key Features:

• Interpretable Scoring Card Methodology: Employs an interpretable scoring card method (SCM) that estimates propensities of 20 amino acids and 400 dipeptides in a supervised manner.
• Sequence-only prediction: Predicts AMAP activity using only peptide sequence information.
• Supervised learning framework: Derives SCM propensities through supervised machine-learning training on labeled peptide datasets.
• Predictive performance: Achieves a maximum accuracy of 0.957 and a Matthews correlation coefficient (MCC) of 0.834 on independent test datasets.
• Physicochemical insights: Analyzes SCM-derived propensities and selected physicochemical properties to relate AMAP activity to length, composition, charge, conformation, hydrophobicity, and amphipathicity.

Scientific Applications:

• High-throughput AMAP discovery: Enables large-scale identification and characterization of antimalarial peptide candidates for downstream study.
• Experimental candidate prioritization: Supports experimental scientists by prioritizing peptide candidates for validation and development of peptide-based therapeutics with potentially lower incidence of drug resistance.
• Broader antimicrobial relevance: Applies to broader clinical contexts where antimicrobial peptides are relevant beyond malaria research.

Methodology:

Implements an interpretable scoring card method (SCM) that estimates propensities of 20 amino acids and 400 dipeptides from peptide sequences in a supervised manner, and analyzes SCM-derived propensities alongside selected physicochemical properties; evaluated on independent test datasets reporting accuracy 0.957 and MCC 0.834.