Pick a target and see whatever the pipeline produced for it. A machine-learning potency model ranks an 8.4M in-stock library; where a structure exists, the top of that ranking is docked and pose-scored. Two prioritised lists from one pipeline — and, when both exist, the consensus: compounds that came through both.
Buying a hit. Every compound has a live buy link resolved from its structure. It lands on a direct structure search at the source vendor, a ZINC/CartBlanche page listing suppliers and price, or a PubChem page with its chemical-vendor list and CAS number. These are catalog compounds you can order.
How the pieces relate. The docking never saw the whole library, it was handed the top of the ML ranking to dock. The consensus is not two independent methods agreeing; it is the set that survived both stages of one pipeline. Scores are for triage: ML pIC50 is a calibrated prediction with a 90% interval; docking GNINA (CNNaffinity) and Vina are ranking scores, not affinities. The ranking check on the docking view reports how often that score ordered known potency correctly for this target.
Methods & references. Docking with Uni-Dock (Yu Y, Cai C, et al. Uni-Dock: GPU-Accelerated Docking Enables Ultralarge Virtual Screening. J Chem Theory Comput 2023; PMID 37125970), scored with the Vina function (Trott O, Olson AJ. AutoDock Vina: improving the speed and accuracy of docking with a new scoring function, efficient optimization, and multithreading. J Comput Chem 2010;31(2):455–461; PMID 19499576) and rescored with GNINA CNNaffinity (McNutt AT, Francoeur P, Aggarwal R, et al. GNINA 1.0: molecular docking with deep learning. J Cheminform 2021;13(1):43; PMID 34108002).