Docking

Pre-Requisites For Docking

To dock a compound, first select a compound from the Selector Compound tab by pressing the focus arrow next to the compound name in the panel on the left side of the screen. You must have a compound in the workspace in order to begin docking. Docking is a "system interrupt" task, meaning users will not be able to interact with the workspace once docking begins until the job has finished. It is therefore important to be sure the system is ready to dock before beginning. Docking in progress will be shown in the Docking menu and golden rod gears will spin in the upper right until the docking is complete.

Next, press the Modeling Tasks button in the right hand menu and select Dock from the list of available modeling tasks. 

You can also click the Dock option from the Ellipse Dropdown Menu of the specific compound. More details in the Menu Section.

On the left side of the screen, the docking menu will appear where a docking program can be chosen as well as other parameters. 

AutoDock Vina

AutoDock Vina samples many locations and orientations of a compound (binding poses) within a designated bounding box region of the protein. These poses are then scored and the best one is returned. The bounding box may be determined by a crystal ligand, hot spot, or selected atoms centroid position using the Box Center dropdown menu.

Under the advanced parameters expansion menu you may adjust the Box edge size, Docking speed, and number of poses you would like returned. By default the Box size will be adjusted to fit the ligand, but may be expanded manually up to 40 angstroms. Please contact the support team if you need a bigger box.

The docking speed sets the exhaustiveness of the pose sampling, trading speed for accuracy. Normal is generally accurate enough for most docking applications. Up to 10 poses may be returned to the Compound tab after docking has completed.

More about AutoDock Vina: AutoDock Vina Publication

DiffDock

DiffDock is a new machine-learning-based docking algorithm that leverages a generative model to explore the conformational space of small molecules, enabling the prediction of their binding poses to protein targets. This model performs global protein docking and will start the compound outside of the protein regardless of user positioning. 

This type of docking is especially useful when multiple binding sites should be considered, or the active site is unknown. It is important to note that this algorithm does not consider waters. The results of this docking simulation will often be more diverse than those of a local docking scheme like AutoDock Vina. Docking with DiffDock is usually less than 10 minutes in the largest compound cases. This docking is not meant for protein-protein docking. 

Under the Advanced Parameters the user may select to have up to 10 of the best poses returned to the Compound menu after the docking run completes. The default DiffDock settings match those with which the model was trained and so changing whether hydrogens are kept during docking or if the initial ligand conformation is retained or stripped and remade with RDKit's 2D to 3D functionalities are entirely experimental and should be used with caution.

More about DiffDock: DiffDock Publication