Sequence-based enzyme catalytic domain prediction using clustering and aggregated mutual information content.

Area of Science:

• Bioinformatics
• Computational Biology
• Enzymology

Background:

• Experimental methods for enzyme active site identification are costly and time-consuming.
• High-throughput in silico methods are needed for analyzing vast protein sequence data.
• Accurate prediction of catalytic residues is crucial for enzyme function prediction.

Purpose of the Study:

• To develop a novel, accurate, and computationally efficient in silico method for predicting enzyme active sites.
• To address the challenges of varying sequence similarity in enzyme datasets.
• To facilitate enzyme function prediction in large-scale genomic studies.

Main Methods:

• Enzyme sequences are clustered based on functional categories (EC labels).
• Sequence graphs are constructed, utilizing graph properties like biconnected components and articulation points to identify common sequence segments.
• An information-theoretic approach, the aggregated column related scoring scheme, is applied to aligned subsequences in common regions to identify potential active sites.

Main Results:

• The combined clustering and aggregated information content scoring method successfully identified known catalytic sites in Escherichia coli K12 enzymes.
• The method demonstrated accuracy in predicting potential active sites.
• The approach is computationally efficient, with graph properties computed in linear time relative to graph edges.

Conclusions:

• The proposed sequence-based method offers an accurate and efficient solution for identifying enzyme active sites.
• This computational approach can significantly aid in the analysis of enzyme sequences from numerous genome projects.
• The method provides a valuable tool for advancing enzyme function prediction and understanding.