SEFP: Structure-Based Enzyme Function Prediction.

Area of Science:

• Computational Biology
• Bioinformatics
• Structural Biology

Background:

• Traditional enzyme property determination is time-consuming and expensive.
• Existing computational methods struggle to effectively utilize enzyme structure information.
• There is a need for efficient and accurate computational tools for enzyme function prediction.

Purpose of the Study:

• To introduce SEFP, a novel computational method for enzyme function prediction.
• To leverage enzyme point clouds and advanced feature extraction techniques.
• To improve the accuracy and efficiency of predicting enzyme functions.

Main Methods:

• SEFP utilizes a tailored enzyme point cloud network to analyze the 3D atomic arrangement.
• It integrates hierarchical residue global features via a residue feature adapter.
• The Bio-BCS residue feature encoder employs an attention mechanism for weighted feature extraction.

Main Results:

• SEFP significantly outperforms existing structure-based and sequence-based computational methods.
• Achieved an f1-score of 95.85 on the RSCB enzyme structure dataset.
• Achieved an f1-score of 94.29 on the HECNet dataset, demonstrating broad applicability.

Conclusions:

• SEFP offers a reliable and precise approach for enzyme function prediction.
• The method demonstrates the potential of leveraging enzyme point clouds for enhanced prediction accuracy.
• SEFP represents an advancement in computational biology and bioinformatics tools.