IBPred

IBPred predicts ion binding proteins (IBPs) in phage genomes using a random forest (RF) model trained on features derived from dipeptide composition and physicochemical residue correlations to enable accurate identification of IBPs.

Key Features:

• Input features: Uses protein sequence data and residue physicochemical properties to represent candidate proteins.
• Feature extraction: Derives dipeptide composition and physicochemical correlation between residue pairs to capture sequence and physicochemical information.
• Feature selection: Applies analysis of variance (ANOVA) to select informative features and reduce redundancy.
• Classifier: Employs a random forest (RF)-based model for prediction of IBPs.
• Performance assessment: Compares the RF model with other classification methods and demonstrates superior accuracy.

Scientific Applications:

• Phage proteome annotation: Identifying ion binding proteins within phage genomes to aid functional annotation of phage proteins.
• Ion-binding mechanism studies: Providing predicted IBP candidates to support investigations into molecular ion-binding mechanisms.
• Comparative phage analyses: Enabling comparative evaluation of IBP prevalence and characteristics across phage datasets.

Methodology:

Computational pipeline uses protein sequence information and residue physicochemical properties to extract dipeptide composition and physicochemical residue-pair correlations, applies ANOVA for feature selection, and builds a random forest (RF) classifier that is compared with other classification methods.