HEMEsPred: Structure-Based Ligand-Specific Heme Binding Residues Prediction by Using Fast-Adaptive Ensemble Learning...

Area of Science:

• Biochemistry
• Bioinformatics
• Computational Biology

Background:

• Heme is a vital biomolecule present in many organisms.
• Accurate identification of heme binding residues (HEMEs) is critical for disease research and drug development.

Purpose of the Study:

• To develop a novel computational predictor, HEMEsPred, for identifying heme binding residues.
• To enhance prediction accuracy by addressing class imbalance and developing ligand-specific models.

Main Methods:

• Collected sequence- and structure-based features (amino acid composition, motifs, surface preferences, secondary structure).
• Designed a fast-adaptive ensemble learning scheme to handle class imbalance and improve performance.
• Developed ligand-specific models to account for variations in heme ligands.

Main Results:

• The proposed ensemble learning scheme effectively addressed class imbalance.
• Ligand-specific models demonstrated statistical significance and improved prediction.
• HEMEsPred showed robust performance, good generalization, and outperformed existing state-of-the-art predictors on benchmark and independent datasets.

Conclusions:

• HEMEsPred is an effective and robust tool for predicting heme binding residues.
• The developed method offers significant improvements over existing predictors.
• A web server for HEMEsPred is available for academic use.