Protein–protein docking can produce a long list of candidate poses, including many that are implausible or irrelevant to the biological question. The challenge is not simply to run a search: it is to decide how much of the search space is worth exploring. If you already have clues about the binding site, Hex in SAMSON lets you use that information to narrow the search. The key is to constrain the search deliberately—without excluding orientations you still need to consider.
Start with a usable system
Before opening Hex, prepare both protein structures. Remove atoms with alternate locations, water, monatomic ions, and ligands if they are present. Add hydrogens if needed, but do not replace hydrogens that already reflect a chosen protonation state. If residues or heavy atoms are missing, the PDBFixer extension can help repair the structure. These checks matter because docking results are only as useful as the structures being searched.
Load the proteins into SAMSON, open Hex from Home > Apps > Biology > Hex, and assign one structure as the receptor and the other as the ligand. If you expect to repeat the calculation with different settings, save the ligand’s starting conformation first using Edit > Conformation.
Use prior knowledge to focus the search
Hex can search orientations across the full sphere, but that may be unnecessary when the likely interface is known. In the Hex interface, set the sampling method to Range angles, then open Advanced parameters. The receptor and ligand range angles define spherical cones around the intermolecular axis connecting the proteins’ origins. A value of 180° leaves the angular search unconstrained; a smaller value limits which orientations are sampled.
If you know approximately where the binding site is, first move and orient the ligand near that site using a SAMSON Move editor. Then set appropriate range angles. The tutorial’s example uses 45° for both receptor and ligand. Treat that as an example, not a universal setting: the right range depends on how confidently you know the interface and how well the starting orientation represents it.
A practical caution: narrowing the cones can reduce the search domain and search time, but it can also exclude a relevant pose if the starting orientation is wrong or the interface is uncertain. When you lack useful site information, retaining a broad search is safer than imposing an unsupported constraint. You can also limit the ligand’s twist rotation around the intermolecular axis when that is useful for the search.
Balance speed and detail
Hex offers a two-stage search: a lower-resolution Steric scan followed by a higher-resolution Final search. The first stage rapidly filters orientations; the final stage evaluates the remaining candidates more finely. The documentation suggests that Steric scan orders of 16–18 are sufficient in most cases, and describes combinations such as N=16 for the scan and N=25 for the final search as practical choices. A higher final-search order can be appropriate for high-resolution crystal structures when binding-related conformational changes are expected to be small; more flexible or model-built structures may call for a lower order.
These settings are trade-offs, not a substitute for judging the structural evidence. Likewise, selecting a correlation type that includes electrostatics may help when the proteins have complementary formal charges, but the documentation advises against electrostatics for docking DNA or RNA molecules.
Review poses rather than trusting a score alone
After the run, inspect the results by cluster. You can view the best-scoring mode per cluster or all modes, and select a row to restore its ligand conformation. Hex can also report steric bumps for candidate poses; use the results table to filter solutions with clashes. Molecular-mechanics energies or minimization are available as post-processing options, but the Hex documentation describes this feature as alpha-stage and cautions against using its rigid-body refinement when large conformational changes are expected.
Before choosing a candidate for further modeling, inspect the pose visually and assess receptor–ligand contacts. SAMSON tools and extensions can help examine contact areas and hydrogen bonds. A favorable docking result is a hypothesis to investigate—not, by itself, proof of a binding mode.
For the complete workflow and parameter details, see the Protein docking with Hex tutorial.
Note: SAMSON and all SAMSON Extensions are free for non-commercial use. Get SAMSON.