FEgrow

FEgrow generates reliable initial binding poses for congeneric series of ligands in protein binding pockets and prepares optimized structures for protein–ligand binding free energy predictions.

Key Features:

• User-Defined Ligand Series Construction: Builds congeneric series of ligands from a specified ligand core structure.
• Optimization of Bioactive Conformations: Enumerates and optimizes bioactive conformations of functional groups using hybrid machine learning and molecular mechanics potential energy functions.
• Scoring with gnina: Optionally scores low-energy ligand poses using the gnina convolutional neural network scoring function.
• Preparation for Free Energy Calculations: Produces optimized and scored ligand poses prepared for rigorous protein–ligand binding free energy predictions.

Scientific Applications:

• Standard Dataset Analysis: Applied to ten congeneric series of ligands bound to targets from a high-quality dataset of protein–ligand complexes.
• SARS-CoV-2 Main Protease Inhibitors: Constructed a set of 13 inhibitors from literature and enabled retrospective computation of their relative binding free energies.

Methodology:

Enumerates and optimizes ligand functional-group conformations with hybrid machine learning and molecular mechanics potential energy functions, optionally scores poses with the gnina convolutional neural network scoring function, and prepares optimized/scored poses for protein–ligand binding free energy predictions.