Bastion6

Bastion6 predicts type VI secreted effectors (T6SEs) in Gram-negative bacteria to identify proteins exported by the type VI secretion system (T6SS) involved in bacterial competition and pathogenesis.

Key Features:

• Feature Extraction and Analysis: Extracts diverse protein sequence features and evaluates them using unsupervised and supervised learning methods.
• Two-Layer SVM-Based Ensemble Model: Integrates extracted features into a two-layer Support Vector Machine (SVM)-based ensemble model with optimized parameters.
• High Predictive Performance: Reported performance metrics include ACC 0.943, F-value 0.946, MCC 0.892, and AUC 0.976.
• Validation with Independent Dataset: Model performance was validated using an independent dataset.
• Application to Novel Effectors: Successfully identified two recently validated T6SE proteins that differ substantially in sequence similarity and cellular function from known effectors.
• Genome-Wide Prediction Capability: Applied genome-wide across 12 bacterial species to analyze 54,212 protein sequences and identify 94 putative T6SE candidates.

Scientific Applications:

• Discovery of Novel T6SEs: Supports discovery of novel type VI secreted effectors (T6SEs) from proteomes.
• Study of Bacterial Competition and Host Interactions: Enables investigation of mechanisms of bacterial competition and host interactions mediated by T6SS effectors.
• Genomic Analysis of Virulence Factors: Facilitates genome-scale identification of bacterial virulence factors for research in pathogenesis and microbial ecology.

Methodology:

Extracts diverse protein sequence features; analyzes them via unsupervised and supervised learning; integrates features into a two-layer SVM-based ensemble model with optimized parameters; validates performance on an independent dataset and applies genome-wide prediction across proteomes.