Protein–protein docking

What is protein–protein docking?

Protein–protein docking is a computational method for arranging two protein structures into candidate bound complexes. It searches their relative orientation and ranks interfaces using geometry, interaction scores, supplied restraints or learned structural patterns.

This is a larger interface problem than placing a small molecule in a pocket. A receptor and partner may contact through broad, relatively flat surfaces, and loops or domains can change shape on association. Global docking explores many surface regions; restrained docking focuses the search using residue-level evidence.

The result is a set of interface models, not proof that the proteins interact in the relevant cell or state. Missing loops, wrong oligomeric assembly and uncertain domain orientation can all create false interfaces. A good-looking model may also be incompatible with a membrane, glycan or other biological partner.

When to use protein–protein docking

Docking is useful when both partners have defensible structures and the interface remains uncertain:

• Interaction assays suggest two proteins associate but give no orientation.
• Mutagenesis or cross-links identify possible interface residues.
• You want to compare alternative complex models for experiment design.
• A known complex can benchmark a method on a related system.

Strengths and limitations

Protein–protein docking turns sparse interface evidence into testable residue contacts. Complementary global, restrained and learned methods can show which orientations depend on an assumed interface and which recur without it.

The search is difficult when binding requires large backbone or domain movement. Scores from different programs are not calibrated interaction energies, and a compact interface can be a packing artifact. Reference-based quality metrics are useful only when a native complex is genuinely available.

How protein–protein docking works

The ProteinIQ comparison workflow runs three methods from the same two PDB inputs.

Tool	Search strategy	Main constraint
LightDock	Swarms distributed around the receptor surface	Near-rigid search with optional limited flexibility; residue restraints change the search when supplied
EquiDock	Learned rigid-body placement	Does not remodel either protein
AF2Dock	Learned complex modeling from two structures	Returns model-specific ranked complexes

Restraints and global search

LightDock searches broadly by default and can use residue restraints when evidence exists. Keep any imposed contacts separate from contacts that emerged without guidance.

EquiDock predicts a rigid transformation, so it is most informative when the input structures are close to their bound conformations. Review any learned model against the original partners for clashes and unrealistic interface changes.

DiffDock-PP is another rigid-body option outside this workflow. It holds the receptor fixed, samples placements of the partner protein with a diffusion model and ranks the poses with a confidence model.

Applications of protein–protein docking

Docking can suggest which residues of a signaling pair, enzyme complex or regulatory assembly might contact each other. A candidate interface can guide targeted substitutions, cross-link measurements or structural experiments. The value lies in a falsifiable contact pattern, not in a high score alone.

It can also compare competing orientations for an interaction already supported experimentally. If methods disagree, inspect whether a supplied restraint, missing segment or alternate protein conformation explains the split. Agreement is stronger when the methods use independent evidence and the proposed assembly fits biological context.

How to do protein–protein docking online

1. Prepare both partners. Choose biologically relevant chains, check oligomeric state and repair justified missing atoms with PDBFixer.
2. Record interface evidence. List cross-links, mutations or known contact residues separately from predictions; convert supported contacts to LightDock restraints when appropriate.
3. Run the comparison workflow. Connect the two PDB files to LightDock, EquiDock and AF2Dock.
4. Review method settings. Check whether LightDock is restrained or running a broad surface search.
5. Compare candidate interfaces. Group models by relative orientation and inspect buried surface, clashes and biologically impossible contacts.
6. Benchmark when possible. If a native complex exists, use DockQ to compare models with that reference; otherwise do not report a reference-based accuracy score.

How to interpret protein–protein docking results

Method-native rankings describe each program's preferred models under its assumptions. EquiDock's output rank is not a confidence probability. DockQ summarizes agreement with a known native complex on a 0 to 1 scale; without that reference it cannot validate a novel interface.

Check model geometry and biological context before comparing rankings. A predicted interface can prioritize mutagenesis or structural follow-up, but it does not confirm interaction, affinity, cellular function or safety.

How the protein–protein docking workflow works

Run LightDock, EquiDock, and AF2Dock in parallel and retain each method’s models for comparison.

1. Prepare both partners. Upload separate PDB structures, select the correct chains and biological states, repair common structural problems, and remove irrelevant crystallographic contacts.
2. Map interface evidence. Record known contact residues, cross-links, mutational effects, conservation, or predicted interaction sites. Distinguish direct restraints from lower-confidence hints.
3. Choose global or restrained search. Use global methods when the interface is unknown and information-driven methods when defensible restraints can reduce the search space.
4. Generate complex models. Generate models with LightDock, EquiDock, and AF2Dock while preserving each method’s clusters, rankings, and native structures.
5. Compare interfaces. Inspect recurring contacts, buried surface geometry, clashes, restraint satisfaction, and method agreement. Use DockQ only when a native reference complex is available.

Inputs and outputs

Check formats before running, then inspect and download the result from every workflow step.

Inputs

Structural inputs

PDBSDFSMILES

Two prepared PDB structures, optionally accompanied by residue-level interface restraints.

Method context

Binding-site evidence, restraints, receptor-state provenance, known ligands, or reference complexes when available.

Outputs

Docked structures

PDBPDBQTSDF

Ranked complex models, method-native scores, interface contacts, and validation metrics when a reference is supplied.

Review evidence

Method-native rankings, confidence, logs, interaction context, failures, and files for reproducible follow-up.