Do I need to prepare molecules from the ZINC database for docking? If so, how should I download and prepare the structures?

ZINC offers prepared, 3D databases that used Epik for state generation, so those structures can be used directly for virtual screening (e.g., with Glide). However, for best results we still recommend LigPrep over the preparation process used for ZINC.

If you want to prepare ZINC yourself with LigPrep, we recommend using the ZINC-supplied script to download the "Single" reference (pH 7) SDF files for your chosen subset. LigPrep can prepare structures from SMILES input, too, which makes for a quicker download, but longer preparation time for the 1D to 3D conversion.

Using the ZINC website, locate the subset that you would like to download (for instance the "lead-like" or "drug-like" subsets). Once you choose the subset, there are additional choices based on pH. The majority of the subset is a single representative form at pH 7. Additional "incremental" subsets are available based on pH ranges.

Next, run LigPrep with Epik to expand ionization states and refine the 3D structures. We recommend a pH tolerance of +/-2 pH units, for instance pH 7.4 +/-2.

We recommend using the "Add metal binding states" LigPrep option. This feature adds additional, targeted deprotonated states to the collection for possible binding to metal atoms in the active site. When the option "Add Epik state penalties to docking score" is selected for Glide docking and the deprotonated heteroatom binds to the metal, then the metal-specific state penalty is added to the docking score instead of the regular Epik state penalty. These extra metal-binding states are simply skipped during docking when there is no metal in the receptor active site.

LigPrep jobs can be split into multiple subjobs and distributed across multiple processors or machines.

Back to Search Results