About

This website showcases PropMolFlow as published in "PropMolFlow: Property-guided Molecule Generation with Geometry-Complete Flow Matching". Authors: Cheng Zeng, Jirui Jin, Connor Ambrose, George Karypis, Mark Transtrum, Ellad B. Tadmor, Richard G. Hennig, Adrian Roitberg, Stefano Martiniani, Mingjie Liu.

This website was developed by the Liu Group at UF Chemistry and built on their PropMolFlow repository. PropMolFlow uses geometry-complete SE(3)-equivariant flow matching for the conditional generation of molecules.

Select the desired property and in-distribution value below, and PropMolFlow will generate a set of 20 interactable molecules. Hover over a molecule to see its GVP regressor-predicted properties. Click the gray border to select the molecule for download.

Set Properties

Download

Generated molecules will appear here