Genetic algorithm-guided design of garlic-derived ligands targeting the TEN domain of telomerase reverse...

Area of Science:

• Computational chemistry
• Drug discovery
• Bioinformatics

Background:

• Human telomerase reverse transcriptase (hTERT) is crucial for cancer cell immortalization.
• hTERT is a significant therapeutic target for developing anticancer drugs.
• Discovering novel hTERT inhibitors is essential for advancing cancer treatment.

Purpose of the Study:

• To develop and validate a computational pipeline for designing and screening garlic-derived compounds as potential hTERT inhibitors.
• To utilize genetic algorithms (GAs) and machine learning (ML) for efficient ligand discovery.
• To identify novel bioactive scaffolds with favorable drug-like and pharmacokinetic properties.

Main Methods:

• A computational pipeline integrating GAs and ML was developed.
• Garlic phytochemicals served as the initial chemical space, iteratively evolved using mutation and fragment expansion.
• A multi-parameter fitness function guided optimization, incorporating drug-likeness rules (Lipinski, Veber, Ghose), QED, and other constraints.
• A RandomForest classifier pre-screened compounds for predicted telomerase activity before molecular docking (CB-Dock2).
• Pharmacokinetic and toxicity properties were assessed using SwissADME.

Main Results:

• Out of 125 generated ligands, 14 met drug-likeness and predicted activity criteria.
• The top-scoring compound showed high binding affinity (-10.5 kcal/mol) but was excluded due to ADMET violations.
• Remaining candidates exhibited favorable physicochemical properties, solubility, lipophilicity, and predicted oral bioavailability.
• The integrated GA-ML approach improved screening efficiency by prioritizing biologically relevant candidates.

Conclusions:

• Genetic algorithms are effective in generating diverse and pharmacologically relevant molecular scaffolds.
• Integrating ML-based activity prediction with GAs enhances the efficiency and selectivity of early-stage ligand discovery.
• This computational strategy offers a cost-effective alternative to traditional docking workflows for identifying potential anticancer drug leads.