Protein structural similarity search by Ramachandran codes.

Area of Science:

• Structural bioinformatics
• Computational biology
• Genomics

Background:

• Exponential growth in protein structural data necessitates efficient similarity search tools.
• Existing methods for reducing 3D protein structures to 1D strings often sacrifice accuracy for speed.
• There is a need for improved linear encoding methodologies and search tools for rapid retrieval of structural homologs.

Purpose of the Study:

• To enhance linear encoding methodology for protein structure similarity searches.
• To develop efficient search tools capable of rapidly retrieving structural homologs from large databases.
• To improve the speed and accuracy of protein structure similarity searching.

Main Methods:

• Proposed SARST (Structural similarity search Aided by Ramachandran Sequential Transformation), a novel linear encoding method.
• Transformed protein structures into text strings using Ramachandran map and nearest-neighbor clustering.
• Employed a regenerative approach to generate substitution matrices for sequence alignment methods.

Main Results:

• SARST achieves accuracy comparable to Combinatorial Extension (CE).
• SARST is over 243,000 times faster than existing methods, searching 34,000 proteins in 0.34 seconds on a 3.2-GHz CPU.
• Provided statistically meaningful expectation values for assessing retrieved information.
• Implemented SARST as a web service and a cross-platform Java program.

Conclusions:

• SARST efficiently distinguishes high from low structural similarities and retrieves homologous structures.
• Linear encoding methodology offers a foundation for efficient protein structural similarity search tools.
• SARST is applicable to automated, high-throughput functional annotations and predictions for large protein structure datasets.