Comparison of structure-based and threading-based approaches to protein functional annotation.

Area of Science:

• Bioinformatics
• Computational Biology
• Structural Biology

Background:

• The genomic revolution has generated vast sequence data, necessitating methods for biological function identification.
• Purely sequence-based functional inference faces challenges, particularly at low sequence similarity levels (the "twilight zone").
• Existing structure-based methods often require high-quality crystal structures and struggle with the complex relationship between protein fold and function.

Purpose of the Study:

• To evaluate and enhance structure-based functional annotation techniques.
• To address limitations of purely sequence-based and global structure similarity approaches.
• To improve the accuracy of protein function prediction, especially in the "twilight zone".

Main Methods:

• Extensive benchmarking of structure-based functional annotation aspects, including binding pocket detection, molecular function assignment, and ligand-based virtual screening.
• Utilizing protein threading with a strong sequence profile component.
• Integrating evolutionary information with structural data.

Main Results:

• Protein threading significantly improves structure-based functional annotation quality in the "twilight zone."
• Detecting evolutionarily related proteins substantially reduces false positives from global structure similarity.
• A combined evolution/structure-based approach enhances functional inference accuracy.

Conclusions:

• Integrating evolutionary and structural data offers a powerful strategy for protein function assignment.
• This combined approach significantly contributes to comprehensive proteome annotation.
• The method effectively overcomes limitations of purely sequence or structure-based methods in challenging cases.