PubChem

Millions of compound structures and descriptive datasets can be freely downloaded via FTP.

PubChem contains multiple substance descriptions and small molecules with fewer than 100 atoms and 1,000 bonds.

[2] PubChem was released in 2004 as a component of the Molecular Libraries Program (MLP) of the NIH.

As of 5 November 2020 (number of BioAssays is unchanged): Searching the databases is possible for a broad range of properties including chemical structure, name fragments, chemical formula, molecular weight, XLogP, and hydrogen bond donor and acceptor count.

PubChem contains its own online molecule editor with SMILES/SMARTS and InChI support that allows the import and export of all common chemical file formats to search for structures and fragments.