MLBP

MLBP predicts multifunctional biological activities from peptide sequences to identify bioactive peptides with activities including anti-cancer, anti-diabetic, anti-hypertensive, anti-inflammatory, and antimicrobial effects.

Key Features:

• Multi-Label Prediction Capability: MLBP simultaneously identifies multiple biological activities within a single peptide sequence rather than predicting a single function.
• Deep Learning Framework: The model employs convolutional neural networks (CNNs) and bidirectional gated recurrent units (Bi-GRUs) for feature extraction and prediction.
• Sequence-Based Input: Sequence vectors are transformed by an embedding layer into dense continuous feature vectors that are input to the deep learning architecture.
• Performance Metrics: Validated with 5-fold cross-validation, MLBP achieved training accuracy 0.695 and absolute true rate 0.685, and test accuracy 0.709 and absolute true rate 0.697, representing approximately 5.0% accuracy and 4.7% absolute true rate improvement over suboptimum methods.

Scientific Applications:

• Peptide discovery and characterization: Used in bioinformatics, pharmacology, and medicinal chemistry to discover and characterize multi-functional peptides from sequence data.
• Drug discovery acceleration: Enables identification of novel therapeutic candidates with broad-spectrum biological activities to accelerate drug discovery processes.
• Peptide therapeutics insight: Provides insights into the potential multi-functional nature of peptides to facilitate understanding of peptide-based therapeutics.

Methodology:

Sequence vectors are transformed via an embedding layer and processed by a deep learning model combining convolutional neural networks (CNNs) and bidirectional gated recurrent units (Bi-GRUs); performance was evaluated using 5-fold cross-validation.