BPBAac

BPBAac predicts bacterial type III secreted (T3S) proteins by using N-terminal position-specific amino acid composition (Aac) extracted with a Bi-profile Bayes model and an SVM classifier to identify T3S effectors for studies of host–pathogen interactions and bacterial pathogenicity.

Key Features:

• Distinctive N-terminal Aac: Utilizes a N-terminal position-specific amino acid composition (Aac) feature present in approximately 50% of T3S proteins that tolerates shifts in position.
• Bi-profile Bayes Model: Applies a Bi-profile Bayes model to extract position-specific Aac features from protein N termini.
• Support Vector Machine (SVM) Classifier: Trains an SVM classifier on Bi-profile Bayes–derived Aac features to discriminate T3S proteins from non-T3S proteins.
• High Accuracy and Robustness: Demonstrated performance in 5-fold cross-validation with average sensitivity ≈90.97% and average selectivity ≈97.42%, and maintained robustness with limited training data.

Scientific Applications:

• Microbial pathogenesis and host–pathogen interaction studies: Prediction of T3S proteins across bacterial species to support investigations of microbial pathogenicity and host–pathogen interactions.
• Genome-wide effector discovery: Genome-wide prediction of putative T3S effectors, exemplified by analyses in Ralstonia solanacearum.
• Effector candidate identification: Identification of putative T3S effector candidates to explore molecular mechanisms of bacterial infection and inform disease-control strategies.

Methodology:

Extract position-specific N-terminal Aac features using a Bi-profile Bayes model; train an SVM classifier on these features; evaluate performance by 5-fold cross-validation and test robustness with limited training data.