iDACP

iDACP identifies subtypes of anticancer peptides (ACPs) using sequential and physicochemical properties, including N- and C-terminal amino acid compositions and the distribution of positively charged residues, to systematically classify ACPs into three major subtypes.

Key Features:

• Subtype classification: Classification is grounded on N- and C-terminal amino acid compositions and the distribution of positively charged residues to define three major ACP subtypes.
• Feature representation: Uses hybrid feature sets combining sequence-based features with physicochemical properties.
• Machine learning architecture: Implements a two-step machine learning model that enhances predictive accuracy through hybrid feature integration.
• Discrimination capability: Distinguishes between anticancer peptides (ACPs) and non-ACPs.
• Performance (cross-validation): Sensitivity 86.75%, specificity 85.75%, accuracy 86.08%, Matthews Correlation Coefficient (MCC) 0.703.
• Performance (independent test): Sensitivity 77.6%, specificity 94.74%, accuracy 88.99%, MCC 0.75.

Scientific Applications:

• ACP subtype discovery: Systematic classification of anticancer peptides into distinct subtypes for comparative studies.
• Peptide characterization: Analysis of sequential features and physicochemical properties to investigate ACP characteristics.
• Anticancer peptide screening: Computational discrimination between ACPs and non-ACPs to support candidate selection for experimental validation and oncology research.

Methodology:

Analysis of N- and C-terminal amino acid compositions and distribution of positively charged residues, construction of hybrid sequence-based and physicochemical feature sets, and application of a two-step machine learning model.