Virtual Screening and Design with Machine Intelligence Applied to Pim‐1 Kinase Inhibitors. Issue 9 (9th July 2020)

• Record Type: Journal Article
• Authors: Schneider, Petra; Welin, Martin; Svensson, Bo; Walse, Björn; Schneider, Gisbert
• Is Part Of: Molecular informatics. Volume 39:Issue 9(2020)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2020-07-09
• Subjects: artificial intelligence -- crystal structure -- de novo design -- drug discovery -- neural network; Cheminformatics -- Periodicals; QSAR (Biochemistry) -- Periodicals; Structure-activity relationships (Biochemistry) -- Periodicals; Drugs -- Structure-activity relationships -- Periodicals; 615.19
• DOI: 10.1002/minf.202000109
• Languages: English
• ISSNs: 1868-1743

Abstract

Abstract: Ligand‐based virtual screening of large compound collections, combined with fast bioactivity determination, facilitate the discovery of bioactive molecules with desired properties. Here, chemical similarity based machine learning and label‐free differential scanning fluorimetry were used to rapidly identify new ligands of the anticancer target Pim‐1 kinase. The three‐dimensional crystal structure complex of human Pim‐1 with ligand bound revealed an ATP‐competitive binding mode. Generative de novo design with a recurrent neural network additionally suggested innovative molecular scaffolds. Results corroborate the validity of the chemical similarity principle for rapid ligand prototyping, suggesting the complementarity of similarity‐based and generative computational approaches. Abstract : Fast computational compound screening led to the discovery of a novel inhibitor of human Pim‐1 kinase, an important anticancer drug target. Chemical ensemble‐based machine‐learning and de novo molecular design with artificial intelligence resulted in innovative chemical scaffolds. This approach might provide a universal template for efficient hit and lead discovery in medicinal and biological chemistry.