A novel model for protein sequence similarity analysis based on spectral radius

Area of Science:

• Bioinformatics
• Computational Biology
• Sequence Analysis

Background:

• Advances in sequencing technologies have led to a surge in biological sequence data, necessitating improved analysis methods.
• Protein sequence similarity analysis is crucial for understanding protein function, evolution, and structure.
• Existing methods may have limitations in capturing complex sequence characteristics.

Purpose of the Study:

• To develop a novel method for quantifying protein sequence similarity.
• To introduce a new 2-D graphical representation for protein sequences based on amino acid spectral radii.
• To numerically characterize protein sequences using statistical, dynamic, and complexity features.

Main Methods:

• Calculated spectral radii for all 20 amino acids.
• Developed a novel 2-D graphical representation for protein sequences.
• Extracted three groups of features: statistical, dynamics measurements, and fluctuation complexity.
• Built two similarity models: Gaussian Kernel and Cosine similarity, using the extracted feature vectors.

Main Results:

• Applied the novel method to four datasets, analyzing similarities and dissimilarities.
• Both Gaussian Kernel and Cosine similarity models yielded consistent results.
• The proposed models demonstrated improvements compared to traditional ClustalW analysis.

Conclusions:

• The novel 2-D graphical representation and feature extraction method provide an effective way to analyze protein sequence similarity.
• The developed models offer a promising tool for enhancing protein research in medical and scientific domains.
• This approach may facilitate deeper insights into protein relationships and functions.