Recognition of coarse-grained protein tertiary structure

Area of Science:

• Computational Biology
• Structural Bioinformatics
• Protein Folding

Background:

• Protein structure prediction is crucial for understanding function.
• Current models struggle to accurately predict global protein structures from local conformations.
• Discrepancies exist between locally precise and globally accurate protein backbone descriptions.

Purpose of the Study:

• To investigate the differences between local and global fitting in protein backbone modeling.
• To develop an algorithm for accurate global protein structure prediction from local conformational data.
• To improve the translation of local structural information into a global protein structure.

Main Methods:

• Modeling protein backbones using C(alpha) atom locations and discrete (phi, psi) conformational states.
• Employing a statistical scoring scheme and threading techniques for local structure analysis.
• Developing a two-step algorithm that incorporates neighboring residue conformational information.

Main Results:

• Local best-fit conformations are recognizable but do not directly determine global structure from sequence alone.
• Incorporating neighboring residue conformal states enables accurate translation of local to global structure.
• The developed algorithm achieves up to 95% recognition of native-state structures within 2.5 A root mean square deviation.

Conclusions:

• Accurate global protein structure prediction requires integrating local conformational data with information from neighboring residues.
• The proposed two-step algorithm offers a significant advancement in predicting protein native-state structures.
• This method enhances the ability to determine global protein architecture from sequence and local structural constraints.