Python API · stjames models · API example
Screen a ligand library against one protein binding site. Rowan returns the best docking score for each compound and can save the best pose of selected top-scoring compounds for inspection and optional MM/GBSA refinement.
For each SMILES, openconf's rapid conformer search generates starting geometries. Rowan docks only the lowest-MMFF-energy conformer, using a fixed receptor. The docking engine still searches ligand positions and conformations and can generate multiple poses from that starting geometry.
Results retain one best score per input compound; failed compounds have no score and do not stop the library screen. Lower docking scores rank better. The saved subset includes ligand poses, protein–ligand complexes, and PoseBusters geometry and clash checks. These checks run only for saved poses and do not establish binding.
MM/GBSA applies only to the saved subset. It reports a separate score without changing the original docking score or the selection of compounds. If refinement cannot run or fails, Rowan retains the original docking pose without an MM/GBSA score.
Docking and MM/GBSA scores are approximate ranking tools, not measured binding affinities. Inspect saved poses before choosing follow-up compounds. See the docking guide for shared preparation and scoring guidance and more extensive conformer sampling or induced-fit docking.