Integrating Secondary Structures Information into Triangular Spatial Relationships (TSR) for Advanced Protein...

Area of Science:

• Bioinformatics
• Structural Biology
• Computational Biology

Background:

• Conventional protein structural comparison methods often overlook detailed similarity.
• Advanced methods like Triangular Spatial Relationship (TSR) offer finer differentiations but lack secondary structure integration.
• Understanding protein folding patterns requires incorporating secondary structure information.

Purpose of the Study:

• To develop an enhanced protein structure representation by integrating secondary structure elements (SSEs) into the TSR method.
• To improve the accuracy and reliability of protein classification using the novel SSE-TSR approach.
• To assess the impact of SSE integration on TSR-based protein discrimination.

Main Methods:

• Developed the Secondary Structure Element-Triangular Spatial Relationship (SSE-TSR) approach.
• Integrated secondary structure elements (helix, strand, coil) into TSR representations, considering 18 combinations.
• Applied the SSE-TSR method with a neural network model for classification on two large protein datasets (9.2K and 7.8K samples).

Main Results:

• The SSE-TSR approach demonstrated improved accuracy in protein classification on both datasets.
• Dataset 1 accuracy increased from 96.0% to 98.3% with SSE integration.
• Dataset 2 accuracy showed a slight improvement from 99.4% to 99.5%, indicating benefits across varying baseline performances.

Conclusions:

• The SSE-TSR method significantly enhances TSR's discriminatory power for protein classification.
• SSE integration offers substantial benefits for datasets with lower initial accuracies and incremental gains for high-performance datasets.
• SSE-TSR is a valuable bioinformatics tool for improving protein classification, function, and interaction studies.