De Novo Drug Design of Targeted Chemical Libraries Based on Artificial Intelligence and Pair-Based Multiobjective Optimization

• DOI: 10.1021/acs.jcim.0c00517
• Publication Date: 2020-08-26T14:39:24Z
• Authors: Alberga Domenico; Gambacorta Nicola; Trisciuzzi Daniela; Ciriaco Fulvio; Amoroso Nicola; Nicolotti Orazio
• Keywords: 0301 basic medicine; de novo; 570; General Chemical Engineering; design; Pneumonia, Viral; Neuraminidase; Library and Information Sciences; Antiviral Agents; chembl; Small Molecule Libraries; Betacoronavirus; 03 medical and health sciences; Artificial Intelligence; Humans; Enzyme Inhibitors; Pandemics; Coronavirus 3C Proteases; multiobjective; SARS-CoV-2; drug; COVID-19; pareto; General Chemistry; Computer Science Applications; 3. Good health; Molecular Docking Simulation; Cysteine Endopeptidases; ai; AI; Influenza A virus; Drug Design; Acetylcholinesterase; Cholinesterase Inhibitors; Neural Networks, Computer; Coronavirus Infections

Artificial intelligence and multiobjective optimization represent promising solutions to bridge chemical biological landscapes by addressing the automated de novo design of compounds as a result humanlike creative process. In present study, we conceived novel pair-based approach implemented in an adapted SMILES generative algorithm based on recurrent neural networks for new molecules whose overall features are optimized finding best trade-offs among relevant physicochemical properties (MW, logP, HBA, HBD) additional similarity-based constraints biasing specific targets. this respect, carried out libraries targeting neuraminidase, acetylcholinesterase, main protease severe acute respiratory syndrome coronavirus 2. Several quality metrics were employed assess drug-likeness, feasibility, diversity content, validity. Molecular docking was finally better evaluate scoring posing generated with respect X-ray cognate ligands corresponding molecular counterparts. Our results indicate that artificial allow us capture latent links joining aspects, thus providing easy-to-use options customizable strategies, which especially effective both lead generation optimization. The is freely downloadable at https://github.com/alberdom88/moo-denovo all data available Supporting Information.