PharmaDB first came with DS 3.5 back in 2012, and with its 139’482 structure-based pharmacophores it was the largest collection of its kind.
Developed in collaboration with the group of Prof. Didier Rognan at the University of Strasbourg (France), it was derived from 7,687 protein–ligand x-ray structures in their annotated database of druggable binding sites from the Protein DataBank, scPDB. The process was described in the article "Protein-Ligand-Based Pharmacophores: Generation and Utility Assessment in Computational Ligand Profiling", J.Chem.Inf.Model. 52(4) 943–955 (2012).
Both PharmaDB and scPDB continued to progress over time. The last version of the scPDB was released in December 2017, just after DS 2018 had come out, and soon customers started asking us to bring PharmaDB up to speed. This has taken a little bit longer than anyone would have liked, but last November it arrived: both the full (DSC-28672) and the example (DSC-30842) version of PharmaDB have been updated, as well as the matching scPDB database of matching 3D ligand conformations (DSC-23549).
The new version included with BIOVIA Discovery Studio 2020 is larger than ever: for each bound ligand in 13'597 Xray structures up to 2×10 hypotheses were generated, bringing the number of pharmacophores that can be screened with the Ligand Profiler up to 247'044.

But before going on your own target-fishing expedition, be aware that going through that many pharmacophores will take some time, even on modern hardware. Also, to save disk space for people that don’t need the full version, only a (very) small example subset gets installed by default. The rest of PharmaDB comes in compressed form and first needs to be unpacked, before it can be used (see the DS Help page Managing Discovery Studio > Installing Discovery Studio > Installing PharmaDB).