CWLy-RF: A novel approach for identifying cell wall lyases based on random forest classifier

Area of Science:

• Biochemistry
• Computational Biology
• Drug Discovery

Background:

• Antibiotic resistance in pathogenic bacteria is a growing global health concern.
• Cell wall lyases show potential as antibacterial agents by degrading bacterial peptidoglycan.
• Current experimental methods for identifying lyases are resource-intensive.

Purpose of the Study:

• To develop a rapid and accurate computational method for identifying cell wall lyases.
• To address the urgent need for efficient tools in the face of rising antibiotic resistance.

Main Methods:

• A random forest (RF) algorithm was employed to build the predictor.
• Mixed-feature representation combined 400D, 188D, and k-spaced amino acid group pairs.
• Information gain was used to select the top 100 features, enhancing predictive ability.

Main Results:

• The CWLy-RF predictor achieved high performance metrics: 96.09% accuracy, 0.993 AUC, 0.922 MCC, 94.92% sensitivity, and 97.32% specificity.
• 10-fold cross-validation confirmed the model's robustness.
• CWLy-RF demonstrated superior performance compared to existing models.

Conclusions:

• CWLy-RF is a highly accurate and efficient computational tool for identifying cell wall lyases.
• This predictor can significantly accelerate the discovery of novel antibacterial agents.
• The tool offers a valuable resource for researchers combating antibiotic resistance.