Protein interaction hotspot identification using sequence-based frequency-derived features.

Area of Science:

• Computational biology
• Bioinformatics
• Structural bioinformatics

Background:

• Accurately identifying protein-hotspot residues is crucial for understanding protein interactions.
• Current methods often rely on complex tertiary structure information, posing limitations.

Purpose of the Study:

• To develop novel sequence-based descriptors for predicting protein-hotspot residues.
• To evaluate the efficacy of these descriptors compared to structure-based methods.

Main Methods:

• Utilized digital signal processing (DSP) techniques to analyze amino acid sequences.
• Developed sequence-derived descriptors for hotspot prediction.
• Employed a random forest classifier for predictive modeling.

Main Results:

• Sequence-derived descriptors achieved 79% accuracy and 75% precision independently.
• Combining sequence descriptors with tertiary structure features improved prediction to 82% accuracy and 80% precision.

Conclusions:

• Digital signal processing-based sequence descriptors are effective for predicting protein-hotspot residues.
• These descriptors offer a viable alternative or complement to structure-based approaches.
• Integrating sequence and structure features enhances prediction performance significantly.