Modules identification in protein structures: the topological and geometrical solutions

Area of Science:

• Structural Biology
• Computational Biology
• Biophysics

Background:

• Identifying modules in protein structures is crucial for understanding protein stability, function, and drug design.
• Protein structures exhibit complex architectures that can be analyzed using network principles.

Purpose of the Study:

• To compare topological (spectral clustering) and geometrical (k-means) methods for identifying protein structure modules.
• To investigate the relevance of protein contact networks and their small-world properties.
• To demonstrate the utility of a topological approach for analyzing protein functional domains and linking sequence to structure.

Main Methods:

• Multiscale analysis of protein architecture.
• Comparison of spectral clustering (topological) and k-means (geometrical) methods.
• Analysis of protein contact networks and their small-world characteristics.
• Development of new representations linking amino acid sequence to topological role.

Main Results:

• Both spectral clustering and k-means methods confirm the relevance of protein contact networks.
• Protein contact graphs exhibit a 'small-world' character, with average shortest path length being key for signal transmission.
• Topological approaches are more effective in identifying protein functional domains.
• New representations linking sequence and topological roles aid in protein structural analysis.
• A case study on azurin highlights the method's ability to capture joint structure-function features.

Conclusions:

• Topological methods, particularly spectral clustering, are superior for identifying functional modules in protein structures.
• Protein contact networks and their small-world properties are fundamental to understanding protein architecture and function.
• The developed methods offer new perspectives for protein structural and functional analysis, with implications for drug design.