EnzDP: improved enzyme annotation for metabolic network reconstruction based on domain composition profiles

Area of Science:

• * Computational biology
• * Bioinformatics
• * Systems biology

Background:

• * Accurate enzyme annotation is crucial for reconstructing cellular metabolic networks.
• * Existing methods like structure-based, network-based, and homology-based approaches have limitations in coverage and accuracy.
• * Current homology-based methods often rely solely on sequence similarity, neglecting protein domain information and threshold sensitivity.

Purpose of the Study:

• * To develop a novel, automated enzyme annotation protocol (EnzDP) that improves accuracy and coverage.
• * To utilize functional domain architecture for more robust protein similarity assessment.
• * To enhance the reconstruction of metabolic networks by reducing gaps and increasing enzyme coverage.

Main Methods:

• * Developed Domain-Enzyme Association Scoring (DEAS) to quantify relationships between domain and enzyme families.
• * Implemented a clustering procedure using DEAS scores instead of traditional sequence similarity.
• * Employed stringent classification, optimal threshold settings, and active site verification in the EnzDP protocol.

Main Results:

• * The EnzDP protocol achieved up to 90% coverage of enzyme families in Swiss-Prot.
• * Demonstrated high accuracy of 94.5% via five-fold cross-validation.
• * EnzDP outperformed existing enzyme annotation methods in various testing scenarios.

Conclusions:

• * EnzDP provides a reliable and automated solution for enzyme annotation.
• * The method significantly enhances the quality and completeness of metabolic network reconstruction.
• * EnzDP offers improved accuracy and coverage compared to current state-of-the-art approaches.