Protein Sequence Comparison and DNA-binding Protein Identification with Generalized PseAAC and Graphical...

Area of Science:

• Bioinformatics
• Computational Biology
• Proteomics

Background:

• The exponential growth of protein sequence data necessitates advanced computational tools.
• Existing methods for protein sequence analysis face challenges in efficiency and accuracy.

Purpose of the Study:

• To develop an efficient computational approach for encoding protein sequences.
• To extract hidden information from protein sequences for analysis and comparison.
• To create a numerical descriptor for protein sequences.

Main Methods:

• Converted protein primary sequences into three-letter sequences based on physicochemical properties.
• Constructed a generalized pseudo amino acid composition (PseAAC) model.
• Developed a support vector machine (SVM) model using generalized PseAAC for DNA-binding protein identification.

Main Results:

• Successfully compared sequence similarities among beta-globin and coronavirus spike proteins, with clusters aligning to known taxonomic groups.
• The generalized PseAAC-based SVM model demonstrated superior performance in identifying DNA-binding proteins compared to four existing methods (DNAbinder, DNA-Prot, iDNA-Prot, enDNA-Prot).
• Achieved significant improvements in accuracy (ACC), Matthew's correlation coefficient (MCC), and F1-score (F1M) across various datasets.

Conclusions:

• The generalized PseAAC model is highly effective for protein sequence comparison and analysis.
• The developed method shows significant promise and competitiveness in the task of identifying DNA-binding proteins.
• This approach offers a valuable tool for bioinformatics research and applications.