SUBSEQ.BIO
DOCS-DIFFDOCK-PP

DiffDock-PP

Predict rigid-body docking poses for one protein receptor and one moving protein partner.

Overview

  • Protein-protein pose hypotheses when large backbone rearrangements are not expected.
  • Binder, antibody-antigen, enzyme-inhibitor, or protein complex screening.
  • Starting poses for downstream relaxation, interface scoring, or visual inspection.

Modes

ModeInput shapeWhen to use it
pair_docking
Pair Docking Default
Uses one selected file/source when file parameters are present. Dock one receptor-ligand protein pair and rank generated poses with the confidence model.

Canonical Job Configuration

These are the fields exposed by the default job configuration for diffdock-pp. They are also returned by GET /api/v1/program/params?program=diffdock-pp and submitted as the params JSON object to POST /api/v1/job/submit.

ParameterTypeModesWhat it does
receptor_structure
Receptor PDB
Structure file All modes Select the receptor PDB. DiffDock-PP expects DB5-style filenames sharing a prefix, such as 1A2K_r_b.pdb with 1A2K_l_b.pdb.
Required; Files: .pdb
ligand_structure
Ligand PDB
Structure file All modes Select the ligand PDB. Use the matching _l_b.pdb suffix, or _l_u.pdb when paired with an unbound receptor.
Required; Files: .pdb
candidate_poses
Candidate Poses
Integer All modes How many candidate poses to sample before confidence ranking. More poses take longer.
Default: 20; Range: 1-100
Advanced configuration fields
ParameterTypeModesWhat it does
random_seed
Random Seed
Integer All modes Use the same seed to reproduce sampling for the same inputs.
Default: 0; Range: 0-999999999
save_trajectory
Save Trajectory
Yes/no All modes Write reverse-diffusion PDB trajectory files under /outputs/visualization.
Default: false

Outputs And Metrics

  • Raw DiffDock-PP prediction bundle for downstream analysis.
  • PDB, CSV, JSON, or related artifacts when generated by the run.
  • The confidence model ranks candidate poses; it is not binding affinity or free energy.
  • With a known reference complex, evaluate poses using C-RMSD, I-RMSD, DockQ, clash checks, and interface-contact recovery.

Common Examples

  • Bound-style redocking: receptor 1A2K_r_b.pdb, ligand 1A2K_l_b.pdb, 20 candidate poses.
  • Unbound docking screen: matching *_r_u.pdb and *_l_u.pdb files with 40 candidate poses.

Example API params

{
  "mode": "pair_docking",
  "receptor_structure": "target_r_u.pdb",
  "ligand_structure": "target_l_u.pdb",
  "candidate_poses": 20,
  "random_seed": 12345
}

Caveats

  • This is binary protein-protein docking, not small-molecule docking.
  • Rigid docking is weaker when binding requires induced fit, loop remodeling, or missing structural context.
  • Inspect multiple top candidates before downstream design.

Advanced Submit

Advanced submit is still available for direct program arguments through POST /api/v1/job/submit-advanced. Prefer canonical configuration unless you need exact low-level arguments or are reproducing a known command line.

  • Advanced submit exposes direct DiffDock-PP inference options for specialized pair naming, sampling, and output controls.
  • The receptor and ligand filenames should share a pair prefix and compatible bound/unbound suffixes.
curl -X POST https://subseq.bio/api/v1/job/submit \
  -H "Authorization: Bearer <api_key>" \
  -F program=diffdock-pp \
  -F 'params={"mode":"pair_docking","receptor_pdb":"target_r_u.pdb","ligand_pdb":"target_l_u.pdb","candidate_poses":20,"random_seed":12345}'

Further Reading