A real valued Genetic Algorithm for generating native like structure of small globular protein.

Area of Science:

• Computational biology
• Bioinformatics
• Protein structure prediction

Background:

• Protein structure prediction (PSP) is a significant challenge in computational biology.
• Accurately predicting a protein's 3D native conformation from its amino acid sequence is crucial for understanding its function.

Purpose of the Study:

• To develop and validate a novel computational method for protein structure prediction.
• To simulate the PSP problem using a real-valued Genetic Algorithm (GA).

Main Methods:

• A real-valued Genetic Algorithm (GA) was employed, incorporating evolutionary operators and a fitness vector.
• The fitness vector integrates knowledge-based biophysical filters: persistence length, radius of gyration, packing fraction, hydrophobicity ratio, and irregularity index.
• Specific weights were applied to convert biophysical measures into a single real value for fitness evaluation.

Main Results:

• The GA demonstrated rapid convergence across six tested globular proteins (17-61 residues).
• The predicted protein structures achieved a backbone RMSD of 3-6Å compared to native structures.
• The method proved effective for proteins with 2-4 helices and strands.

Conclusions:

• The real-valued GA is an effective tool for accelerating protein structure prediction.
• The integrated biophysical fitness vector enhances the accuracy of predicted protein conformations.
• This approach offers a promising solution to the long-standing PSP challenge.