AutoGrow4: an open-source genetic algorithm for de novo drug design and lead optimization

Area of Science:

• Computational Chemistry
• Cheminformatics
• Drug Discovery

Background:

• Computer-aided drug discovery (CADD) is crucial for identifying novel therapeutic agents.
• Existing CADD tools are often limited by predefined virtual libraries.
• Optimization of lead compounds requires efficient and flexible molecular generation methods.

Purpose of the Study:

• To introduce AutoGrow4, an enhanced open-source program for semi-automated drug discovery.
• To demonstrate AutoGrow4's capability in de novo ligand design and lead optimization.
• To showcase AutoGrow4's application in targeting the poly(ADP-ribose) polymerase 1 (PARP-1) protein.

Main Methods:

• Utilizes a genetic algorithm for on-demand ligand evolution, independent of virtual libraries.
• Incorporates advanced cheminformatics, docking program compatibility, chemical filters, and multithreading.
• Applies AutoGrow4 to the catalytic domain of PARP-1 for de novo design and lead optimization.

Main Results:

• AutoGrow4 successfully generated novel, drug-like compounds with superior predicted binding affinities compared to FDA-approved PARP-1 inhibitors.
• The generated compounds exhibited binding modes similar to known inhibitors, even when initiated with random molecules.
• The program demonstrated increased speed, stability, and modularity over previous versions.

Conclusions:

• AutoGrow4 is a powerful and versatile tool for generating novel drug candidates and optimizing existing ligands.
• The software facilitates efficient drug discovery by evolving molecules tailored to specific protein targets.
• AutoGrow4 represents a significant advancement in semi-automated computer-aided drug discovery, freely available for research use.