Computational prediction of heme-binding residues by exploiting residue interaction network

Area of Science:

• Computational Biology
• Structural Bioinformatics
• Protein Engineering

Background:

• Accurate identification of heme-binding residues is essential for understanding and engineering heme proteins.
• Existing methods often rely solely on sequence or 3D structural information, limiting prediction accuracy.

Purpose of the Study:

• To develop a novel computational method for predicting heme-binding residues.
• To leverage topological properties of residue interaction networks for improved prediction.

Main Methods:

• Generated residue interaction networks from 3D protein structures.
• Analyzed topological features (degree, closeness, betweenness, clustering coefficient) of residues.
• Developed HemeNet, a support vector machine (SVM) predictor integrating network features with sequence and structural data.

Main Results:

• Key heme-binding residues exhibit distinct network topological properties (e.g., higher degree, closeness, betweenness; lower clustering coefficient).
• Incorporating network-based features significantly enhanced the prediction performance of HemeNet.
• HemeNet accurately predicted binding residues in both apo (heme-free) and holo (heme-bound) protein states.

Conclusions:

• Topological analysis of residue interaction networks provides valuable insights into heme-binding site identification.
• HemeNet offers a robust and improved approach for predicting heme-binding residues, aiding in protein design and functional studies.
• The HemeNet web server is available for public use.