Protein–protein docking can produce a long list of candidate poses, even when only a small region of the proteins is biologically plausible. That creates a familiar modeling problem: more computation and more results do not necessarily mean more useful answers. A practical way to make a docking run more focused is to use what you already know about the likely binding site to constrain the search—without treating the resulting poses as proof of a biological interaction.
This workflow uses the Hex Extension in SAMSON. Hex samples possible orientations of a ligand protein around a receptor. If you have no information about where the proteins may bind, a broad search can explore the whole surface. If you have a plausible binding region, you can orient the ligand toward it and limit the angular search to spend less effort on orientations that are unlikely to be relevant.
Prepare before you narrow the search
Start with structures that are ready for docking. The tutorial recommends removing atoms with alternate locations, water, monatomic ions, and other ligands when present, and adding hydrogens if they are missing. If the structure has missing residues or heavy atoms, use a protein-fixing workflow before docking; gaps or incomplete side chains can complicate interpretation of a pose. Be careful not to overwrite hydrogens that already reflect a chosen protonation state.
In SAMSON, open the Hex app from Home > Apps > Biology > Hex. Assign one protein as the receptor and the other as the ligand. Save the ligand’s starting conformation before running the search if you may want to repeat the calculation with different settings.
Use prior knowledge to define a search region
Hex’s Range angles sampling method lets you restrict receptor and ligand orientations. In the advanced parameters, each range angle defines a spherical cone around the intermolecular axis connecting the origins of the two proteins. An angle of 180 degrees applies no angular restriction; a smaller angle samples a narrower region.
When a binding site is already suspected, first move and orient the ligand near that region. Then set the sampling method to Range angles and choose suitable limits. The tutorial demonstrates 45 degrees for both the receptor and ligand. Treat that as an example starting point, not a universal setting: the useful range depends on how well the initial orientation represents the expected encounter. You can also limit the ligand’s twist rotation around the intermolecular axis.
The key trade-off is straightforward: a tighter range can reduce the search domain and search time, but it can also exclude a plausible pose if the initial orientation or site assumption is wrong. If the binding site is unknown, retaining broad angular sampling is safer than imposing an arbitrary narrow cone. Record the assumptions behind any constraint so that a promising result can be interpreted in context.
Choose a search that matches the structure
Hex offers correlation choices such as shape-only and shape plus electrostatics. Electrostatics may be useful when the proteins have complementary formal charges; the documentation advises against using it for DNA or RNA docking. The steric scan is a fast, lower-resolution first stage that keeps a subset of orientations for the higher-resolution final search. This two-stage approach can cover the space more quickly than evaluating every orientation at high resolution.
The Hex tutorial notes that a steric scan order of 16–18 is sufficient in many cases, and describes combinations such as 16 for the scan followed by 25 for the final search. For high-resolution crystal structures where binding-related conformational changes are expected to be small, it recommends considering a final search order of 30; for model-built or more flexible structures, 25 may be more appropriate. Higher-resolution searches can take longer, so match settings to the question and available time rather than increasing every parameter by default.
Read the results as candidates, not conclusions
After docking, inspect the clustered modes in the Results tab. You can view the best-scoring mode per cluster or all modes, restore a conformation by selecting its table row, and create a conformation in the document for further examination. A bumps count can help identify solutions with steric clashes. Molecular-mechanics refinement is described as an alpha-stage option in Hex; its rigid-body procedure may be unsuitable when substantial conformational changes are expected.
Finally, examine candidate poses visually and check their interactions rather than relying on a score alone. SAMSON tools can help visualize structures, measure distances, and inspect ligand–receptor contacts or hydrogen bonds. The useful outcome of a constrained search is not simply a shorter list: it is a set of candidates whose assumptions and limitations you can explain and test.
For the full 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.