Accurate sequence-based prediction of catalytic residues

Area of Science:

• Biochemistry
• Bioinformatics
• Computational Biology

Background:

• Predicting catalytic residues is crucial for understanding enzyme function.
• Existing methods often rely on structural data, limiting their applicability.
• There is a need for accurate, sequence-based prediction tools.

Purpose of the Study:

• To develop and validate a novel sequence-based catalytic residue predictor.
• To achieve prediction quality comparable to structure-based methods.
• To outperform existing sequence-based prediction approaches.

Main Methods:

• Utilized sequence-based features and PSI-BLAST profiles.
• Employed feature selection to optimize input for a Support Vector Machine (SVM) classifier.
• Incorporated novel features including amino acid propensities, flexibility (Glycine), and conserved sequence motifs.

Main Results:

• CRpred achieved prediction quality comparable to structure-based methods and superior to other sequence-based methods.
• Reported precision of 15-19% and true positive (TP) rates of 48-58%, varying by dataset.
• Demonstrated that catalytic residues are generally more conserved and influenced by local hydrophobicity.

Conclusions:

• CRpred offers a powerful, accessible tool for predicting enzyme catalytic residues.
• The method's novel features enhance prediction accuracy.
• Sequence-based prediction is a viable and effective alternative to structure-based approaches.