GeinDock Suite
Home › Ligand docking

Ligand docking

The default protocol: a rigid receptor and a flexible ligand. What the search explores, what it holds fixed, and how to read what comes back.

Ligand docking is the default protocol and the one every other mode is built on. The receptor is held rigid, the ligand is free to rotate about its rotatable bonds, and the search looks for the placement inside the box that the scoring function likes best.

The panel heading reads Ligand Docking with the note Rigid receptor, flexible ligand, and step 4 is Settings.

What the search actually explores

Three things vary during the search, and it is worth being clear about which they are, because everything the protocol can and cannot do follows from the list.

Varies Held fixed
Where the ligand sits inside the box, in all three directions.Every receptor atom. Nothing on the protein moves.
How the ligand is oriented.The ligand's bond lengths and bond angles.
The torsion angle of each of the ligand's rotatable bonds.The ligand's ring conformations and its stereochemistry.
Ring conformation and stereochemistry come from your input, unchanged. The search does not flip a chair to a boat and it does not invert a stereocentre. Both come from the three dimensional structure you supplied in step 2 and are carried through. If the bound conformation of your compound needs a different ring pucker, the search cannot find it, so supply the conformer you mean.

Running it

The whole procedure is the six steps of the Docking Panel, in order. There is nothing mode specific to do.

  1. Prepare the receptor. Step 1. Choose the chain that carries your site, strip the waters and heteroatoms, and let it protonate at pH 7.4.
  2. Prepare the ligands. Step 2. Check the atom counts and the validity column before you move on.
  3. Place the box. Step 3. Use the crystal ligand if the structure had one, and read the clearance note before you commit.
  4. Prove the box. Validation. Once per target, this tells you whether the configuration can reproduce a pose that is known to be right.
  5. Choose a precision. Step 4. Swift for a first look or a screen, Rigorous for anything you intend to report.
  6. Run it and read the poses. Step 5 and the pose table. Sort by score, then look at whether the top poses agree with one another.
A finished ligand docking run
A finished run. Nine poses for one ligand, sorted by score, with the best pose drawn in the site and its RMSD to the other poses reported beside it.

What the columns mean for this mode

The pose table carries columns for every mode, and the ones that do not apply stay empty. In ligand docking the ones that matter are these.

Column What it tells you
Score (kcal/mol)The scoring function's estimate of binding affinity. More negative is better. It is an estimate from an empirical function, not a measurement, so use it to rank compounds against each other rather than to predict a number.
RMSD l.b. (Å)How far this pose is from the best pose, measured conservatively. The best pose is 0 by definition.
RMSD u.b. (Å)The same comparison measured strictly. A large gap between the two bounds means the pose is similar in shape but differently oriented, which is often a symmetric molecule flipped end for end.
ModeThe rank of the pose within its ligand, 1 being the best scoring.
Read the RMSD columns as a consistency check. If modes 2 and 3 sit within about 2 Å of mode 1, the search converged on one answer from several directions and the pose is worth believing. If the best few poses are several Ångström apart and score within a few tenths of each other, the search has found several unrelated placements it cannot choose between, and the top one is not reliable. Raising exhaustiveness is the first thing to try.

When ligand docking is the wrong tool

The rigid receptor is the assumption that limits this mode, and there are three situations where it breaks down.