Prediction of protein secondary structure by mining structural fragment database.

Area of Science:

• * Bioinformatics
• * Computational Biology
• * Structural Biology

Background:

• * Accurate prediction of protein secondary structure is crucial for understanding protein function and design.
• * Existing methods face challenges in regions of low sequence similarity.

Purpose of the Study:

• * To develop and evaluate a new method for predicting protein secondary structure.
• * To improve prediction accuracy by integrating fragment mining with established prediction algorithms.

Main Methods:

• * Utilized multiple sequence alignment (via BLAST) against the Protein Data Bank (PDB) for fragment identification.
• * Explored various weighting schemes and normalized scores for secondary structure prediction (alpha-helix, beta-sheet, coil).
• * Applied and compared artificial intelligence techniques: Decision Trees (DT), Neural Networks (NN), and Support Vector Machines (SVM).

Main Results:

• * Support Vector Machines (SVM) achieved the highest prediction accuracy among the tested AI techniques.
• * Combining fragment mining with GOR V improved prediction accuracy, particularly in regions with low sequence similarity.

Conclusions:

• * The developed fragment mining approach, enhanced by SVM, offers a robust method for protein secondary structure prediction.
• * Integration with GOR V provides a synergistic improvement for challenging low-sequence-similarity regions.