PSAC-PDB: Analysis and classification of protein structures.

Area of Science:

• Structural bioinformatics
• Computational biology
• Biochemistry

Background:

• Protein Data Bank (PDB) is a crucial resource for structural biology.
• Accurate classification of protein structures is essential for understanding biological function.
• Existing methods for protein structure analysis can be computationally intensive.

Purpose of the Study:

• To introduce PSAC-PDB, a novel framework for analyzing and classifying protein structures.
• To evaluate the effectiveness of different features (AA sequences, aligned AA, aligned secondary structure elements, frequent AA patterns) for classification.
• To compare PSAC-PDB performance against state-of-the-art methods.

Main Methods:

• Utilizing a protein structure comparison tool to identify similar structures in the PDB.
• Extracting and analyzing amino acid sequences, aligned amino acids (AAA), aligned secondary structure elements (ASSE), and frequent amino acid (FAA) patterns.
• Employing eleven classifiers and evaluating performance using six metrics.

Main Results:

• Three classifiers demonstrated strong overall performance.
• Frequent amino acid (FAA) patterns proved efficient for classifying protein structures, serving as a viable alternative to full sequences or alignments.
• Aligned amino acids (AAA) yielded superior classification results compared to raw amino acid sequences.
• PSAC-PDB surpassed existing approaches in classifying SARS-CoV-2 genome sequences.

Conclusions:

• PSAC-PDB offers an efficient and effective framework for protein structure classification.
• Frequent amino acid patterns represent a promising feature set for structural bioinformatics.
• The framework demonstrates potential for applications beyond PDB analysis, including genomic sequence classification.