ACP-MHCNN

ACP-MHCNN predicts anticancer peptides by employing a multi-headed convolutional neural network that integrates sequence, physicochemical, and evolutionary features for computational identification of anticancer peptides.

Key Features:

• Multi-Headed Deep Convolutional Neural Network Architecture: Employs a multi-headed deep convolutional neural network to extract and integrate discriminative features from multiple numerical peptide representations.
• Feature Extraction: Leverages sequence, physicochemical, and evolutionary-based numerical representations to capture discriminative peptide features while maintaining a restrained parameter overhead.
• Performance Superiority: Validated using cross-validation and independent datasets, demonstrating improvements over prior models in accuracy (+6.3%), sensitivity (+8.6%), specificity (+3.7%), precision (+4.0%), and Matthews Correlation Coefficient (+0.20).

Scientific Applications:

• Anticancer peptide identification: Provides computational screening to identify candidate anticancer peptides and accelerate discovery of potential anticancer therapeutics.

Methodology:

Implements deep learning with a multi-headed convolutional neural network to integrate sequence, physicochemical, and evolutionary numerical peptide representations and uses cross-validation and independent-dataset validation.