Application and assessment of deep learning for the generation of potential NMDA receptor antagonists.
Physical chemistry chemical physics : PCCP January 21, 2021 Katherine J Schultz, Sean M Colby, Yasemin Yesiltepe et al.
Generative deep learning models can design new drug-like compounds that block the N-methyl d-aspartate receptor (NMDAR), a target for neurological diseases like Parkinson's and Alzheimer's, though some existing NMDAR antagonists cause dissociative effects and have been used to create illicit drugs. The study created a library of experimentally validated NMDAR PCP-site antagonists and applied ligand- and structure-based assessment techniques to deep learning-generated compounds. Twelve candidate antagonists not found in existing chemical databases were identified, but synthesis and experimental validation are still needed.