AutoGrow4

AutoGrow4 evolves predicted ligands using a genetic algorithm to generate and optimize novel drug-like molecules for computer-aided drug discovery.

Key Features:

• Genetic algorithm: Uses a genetic algorithm to iteratively evolve ligand structures.
• Dynamic ligand evolution: Generates and modifies candidate ligands de novo rather than relying on fixed compound lists.
• Elimination of pre-enumerated virtual libraries: Operates without dependence on pre-enumerated virtual libraries of compounds.
• Novel molecule generation and ligand optimization: Produces entirely novel drug-like molecules and optimizes existing ligands.
• Compatibility with docking programs: Supports integration with various docking programs for scoring predicted binding poses and affinities.
• Customizable chemical filters: Applies user-definable chemical filters to constrain chemical space.
• Multithreading: Provides multithreading options to parallelize computations.
• Diverse selection methods: Implements multiple selection strategies for choosing candidates across generations.
• Predicted binding affinities and binding modes: Outputs predicted binding affinities and reproduces binding modes comparable to known inhibitors.
• Performance and modularity: Implements computational optimizations that yield enhanced speed, stability, and modularity.

Scientific Applications:

• PARP-1 inhibitor design: Applied to the catalytic domain of poly(ADP-ribose) polymerase 1 (PARP-1) to generate compounds with predicted binding affinities surpassing those of FDA-approved PARP-1 inhibitors and with binding modes resembling known inhibitors, including cases initiated from random small molecules.

Methodology:

Employs a genetic algorithm to evolve ligands without pre-enumerated virtual libraries, interfaces with docking programs for scoring, applies customizable chemical filters, and supports multithreading and diverse selection methods.