Hot spots localization in proteins by optimized short time Ramanujan Fourier transform

Area of Science:

• Biochemistry and Molecular Biology
• Computational Biology
• Bioinformatics

Background:

• Protein-protein interactions are crucial for biological functions.
• Hot spot residues are key determinants of these interactions, with applications in protein engineering and drug discovery.
• Existing experimental and computational methods for hotspot identification are often resource-intensive or require structural data and training.

Purpose of the Study:

• To develop a computational method for identifying protein hot spots using only sequence information.
• To introduce a novel approach based on the Resonant Recognition Model (RRM) utilizing characteristic periods instead of frequencies.
• To provide a model-independent and training-free method for hotspot identification.

Main Methods:

• Utilized the Resonant Recognition Model (RRM) with characteristic periods.
• Employed the Ramanujan Fourier Transform (RFT) to extract characteristic periods from protein family consensus spectra.
• Generated position-period plots using Short Time RFT (ST-RFT) with an optimized Gaussian window, identifying hot spots via signal thresholding.

Main Results:

• Successfully identified hot spots using sequence information alone.
• The proposed method demonstrated improved sensitivity compared to existing RRM-based approaches.
• The optimization of the Gaussian window shape parameter using a concentration measure enhanced ST-RFT performance.

Conclusions:

• The developed sequence-based method offers an efficient and accessible tool for identifying protein hot spots.
• This approach eliminates the need for protein structures or prior training, making it broadly applicable.
• The method holds significant potential for advancing protein engineering and drug discovery efforts.