StarDrop users who have licensed the Surflex eSim3D module can freely download prepared virtual screening collections for use in StarDrop.
eMolecules‘ commercially available screening collection (>17 million compounds) has been prepared for use in ligand-based virtual screening.
During preparation, each of the compounds has had conformations generated and stored in a *.sfdb file, which is a file format native to Surflex eSim3D. When a virtual screen is launched from StarDrop, the user points to one of these SFDB files; each conformation is aligned and scored against the reference compound, which is held in a fixed conformation.
For users wishing to run ligand-based virtual screening within StarDrop, we have made the SFDBs available for download, however, please note that the full collection file is very large and will take time to download. The full collection has been randomly split into subsets representing 1/10, 1/100, and 1/1000 of the full collection; these smaller files are also available below.
These collections will be updated regularly.
To start a virtual screening experiment, click the arrow button menu 
Run Virtual Screen.
Several options are available to load the 3D input structure(s) to use for virtual screening:
Open a file containing the 3D coordinates by clicking the 
button.
Download a PDB containing the molecule in a bioactive conformation by clicking the 
button.
Use an entire bunding hypothesis or obtain a molecule in a specific conformation from a binding hypothesis result by clicking the 
button.
Use a compound in the 3D Viewer by selecting all atoms of the molecule and clicking the 
button.
Once the desired input conformations have been selected, click Next.
In the next page of the wizard, you can Choose a virtual screening database.
Load the virtual screen collection by clicking on 
button, locate where you saved the files and confirm it.
For more details, please refer to the StarDrop user guide.
We present a hybrid structure-guided strategy that combines molecular similarity, docking, and multiple-instance learning such that information from protein structures can be used to inform models of structure–activity relationships.
This article discusses logic fallacies in the context of off-target predictive modelling.
We present an approach that uses structural information known prior to a particular cutoff-date to make predictions on ligands whose bounds structures were determined later. The knowledge-guided docking protocol was tested on a set of ten protein targets using a total of 949 ligands.
