CGLFold: a contact-assisted de novo protein structure prediction using global exploration and loop perturbation...

Area of Science:

• Computational Biology
• Structural Biology
• Bioinformatics

Background:

• Protein structure prediction is crucial for understanding protein function.
• Protein loops, connecting secondary structures, significantly influence overall protein topology.
• Improving loop modeling accuracy is key to enhancing protein structure prediction.

Purpose of the Study:

• To investigate if a loop-specific sampling strategy can improve protein structure prediction accuracy.
• To develop and evaluate a novel de novo protein structure prediction method combining global exploration and loop perturbation.

Main Methods:

• A novel method, CGLFold, combining global exploration (fragment recombination/assembly) and loop perturbation (local refinement) was developed.
• A loop-specific local perturbation model, solved using a differential evolution algorithm, was designed.
• Filtered contact information guided a conformation selection model for sampling.

Main Results:

• CGLFold demonstrated improved structure diversity and conformational update success rates.
• The loop-specific perturbation gradually enhanced conformation accuracy.
• CGLFold achieved notable success rates on benchmark proteins and CASP targets (CASP13, CASP12).

Conclusions:

• The proposed loop-specific sampling strategy effectively improves protein structure prediction accuracy.
• CGLFold offers a promising approach for de novo protein structure prediction by integrating global and local conformational sampling.
• The method shows potential for advancing structural biology research.