Extended Data Fig. 5: SynGFN has substantially accelerated the drug DMTA cycle.
From: SynGFN: learning across chemical space with generative flow-based molecular discovery
(a) Chemical structures of the candidate compounds synthesized and evaluated for biological activity. Each compound is annotated with its Combo Score and IC50 value. The overlays below depict the align-ment of the generated molecules with the reference molecule: gaussian surface overlap is shown with the corresponding Shape Score, and pharmacophore feature alignment is shown with the corresponding Color Score. (b) Comparison between the synthetic routes proposed by SynGFN and the actual synthetic routes used.
