AHTPs

AHTPs predicts anti-hypertensive peptides from peptide sequences using sequence-derived feature encodings and deep learning to identify candidate therapeutics for hypertension.

Key Features:

• Deep Learning Architecture: Integrates convolutional neural network (CNN) and gated recurrent unit (GRU), with the CNN reducing feature dimensionality and the GRU filtering sequential information via reset and update gates.
• Feature Extraction Methods: Uses Kmer, DDE (Deviation Between Dipeptide Frequency and Expected Mean), EBGW (Encoding Based on Grouped Weight), EGAAC (Enhanced Grouped Amino Acid Composition), and DBPF (Dipeptide Binary Profile and Frequency) to encode peptide sequences.
• Output Activation: Employs a Sigmoid activation function in the output layer for binary classification of anti-hypertensive peptides.
• Model Performance: Evaluated by tenfold cross-validation, reporting accuracies of 96.23% and 99.10%, respectively.

Scientific Applications:

• Peptide discovery and validation: Accelerates identification and validation of anti-hypertensive peptides as candidate therapeutics for hypertension.
• In silico screening: Enables efficient screening of peptide sequences to prioritize candidates for experimental validation, reducing time and resource requirements.

Methodology:

Extract features using Kmer, DDE, EBGW, EGAAC, and DBPF; input these features into a CNN to reduce dimensionality; process the CNN output through a GRU to capture sequential dependencies; apply a Sigmoid activation in the output layer for binary classification.