An improved DNA-binding hot spot residues prediction method by exploring interfacial neighbor properties

Area of Science:

• Computational biology
• Structural bioinformatics
• Molecular interactions

Background:

• DNA-binding hot spots are crucial for protein-DNA binding energy but represent a small interface fraction.
• Experimental identification of hot spots is laborious and expensive, necessitating efficient computational methods.

Purpose of the Study:

• To develop an improved computational method for predicting DNA-binding hot spots.
• To enhance prediction accuracy by integrating novel features and employing rigorous feature selection.

Main Methods:

• Developed inpPDH, a computational method incorporating hybrid features (traditional and interfacial neighbor properties).
• Implemented a two-step feature selection strategy to identify optimal features.
• Utilized support vector machine (SVM) for predictor construction with 7 selected features.

Main Results:

• The inpPDH method demonstrated significantly higher prediction accuracy compared to existing literature methods on a benchmark dataset.
• A user-friendly web server for inpPDH was established and made publicly available.
• Identified a comprehensive feature subset, including interfacial neighbor features, crucial for hot spot identification.

Conclusions:

• The inpPDH model provides an accurate prediction of hot spot residues in protein-DNA interfaces from complex structures.
• Interfacial neighbor features, combined with traditional features, form a powerful set for predicting hot spot residues.
• The developed method and identified features offer a more efficient approach to understanding protein-DNA interactions.