GitHub | GitLab mirror | Preprint

Peptide-Protein Docking¶

CABS-dock is a well-established method for flexible peptide–protein docking. It uses the CABS coarse-grained model to perform global peptide docking without prior knowledge of the binding site, while allowing full flexibility of the peptide and limited, near-native flexibility of the protein receptor. This makes it suitable for docking cases where the peptide binding site is unknown, the peptide is highly flexible, or receptor flexibility may affect binding.

The method was introduced as the CABS-dock web server (Nucleic Acids Research 2015), later extended into a standalone package (Bioinformatics 2019), and further developed for information-driven docking using contact restraints (Briefings in Bioinformatics 2019). Its broader applications and developments were summarized in the CABS-dock applications review (Protein Science 2020). Recent extensions include a cyclic peptide docking protocol (JCTC 2025), where CABS-dock is combined with cyclic restraints and refinement procedures.

In CABS-flex standalone 3, the docking workflow provides a command-line interface for peptide docking, receptor flexibility control, information-driven docking with restraints, clustering, scoring, and all-atom reconstruction of representative binding poses.

For more information about the underlying model, see CABS Model. For receptor restraint settings, see Flexibility Modes. For peptide structure prediction, see Peptide Modeling.

Basic Run¶

⬆ Back to top

Minimal command¶

CABSdock -i <receptor_pdb> -p <peptide_sequence>

Example:

CABSdock -i 1p38 -p ALVLP

This command will: 1. Placement: Randomly place multiple copies of the peptide around the receptor. 2. Simulation: Perform Monte Carlo sampling of the peptide's position, orientation, and internal conformation, as well as receptor flexibility. 3. Scoring: Filter and cluster the resulting binding poses. 4. Reconstruction: Provide the top 10 representative binding poses in all-atom representation.

Advanced Docking Options¶

⬆ Back to top

Peptide Input & Structure¶

• --peptide <sequence>: Dock a peptide with a specific sequence.
• --peptide-ss <ss_string>: Provide predicted secondary structure (e.g., CHHHHHHC) to improve docking accuracy.
• --add-peptide <pdb_file>: Use a specific starting structure for the peptide.

Docking Search Control¶

• --separation <dist>: Initial distance between peptide and receptor (default: 20Å).
• --exclude <selection>: Prevent the peptide from docking to specific regions (e.g., hidden pockets or known non-binding sites).

Simulation Depth & Temperature¶

For de novo docking, CABS uses simulated annealing and multiple independent simulation runs (replicas) to broadly explore the binding landscape.

• Default Temperature: 2.0 1.0 (simulated annealing).
• Default Replicas: 10 independent simulation runs.
• Default Cycles: 50 annealing cycles and 50 Monte Carlo cycles.

For a detailed explanation of simulation hierarchy, temperature controls, and replica snapshot calculations, see the Sampling and Temperature guide.

Receptor Flexibility (-g)¶

By default, the receptor is kept relatively rigid to maintain its fold. For cases where significant conformational changes are expected upon binding, use:

--protein-restraints flexible

This allows the receptor backbone (especially loops) to adapt to the peptide binding.

Information-Driven Docking¶

If you have predicted or experimental contacts between the peptide and receptor, you can guide the docking using manual restraints:

--ca-rest-add <res_rec> <res_pep> <dist> <weight>

For example: --ca-rest-add 45:A 3:PEP 5.0 1.0 (forces residue 3 of the peptide to be near residue 45 of chain A).

Docking Cyclic Peptides¶

CABS-flex standalone 3 supports docking of cyclic peptides (including backbone-cyclized and disulfide-stabilized peptides) directly into the receptor. CABS enforces these constraints during the simulation to ensure the peptide preserves its cyclic topology during the global search.

To dock a cyclic peptide, you submit the cyclization and disulfide restraints using the same options as in the flexibility and folding workflows:

• Backbone Cyclization: Restrain the ends of the peptide chain. You must specify the peptide's chain identifier (e.g., PEP1 or PEP):

--backbone-cyclization PEP1

(Enforces a default virtual C-alpha backbone restraint of 3.8 Å with a force weight of 1.0 between the first and last residues to close the peptide loop).

• Disulfide Bonds: Restrain the side-chains of cysteine residues to form disulfide bridges:

  --disulfide-bonds 3:PEP1 8:PEP1

  (Enforces a default side-chain pseudoatom restraint of 2.0 Å with a force weight of 1.0 between the specified Cysteine residues).

• Custom Cyclization: For non-standard cyclizations (e.g., head-to-sidechain or custom weights), you can manually add backbone and side-chain restraints:

  --ca-rest-add 1:PEP1 10:PEP1 5.0 1.0

The protocol for flexible docking of peptides is described in detail in CABS-based flexible docking of cyclic peptides (JCTC 2025).

Output and analysis¶

⬆ Back to top

Docking results are saved in the working directory (default: CABSdock_output/):

Structural Ensembles (output_pdbs/)¶

• model_1.pdb to model_10.pdb: The top 10 binding poses (all-atom). These are the medoids of the most populated clusters.
• replica_*.pdb: Coarse-grained trajectories of the docking simulation.
• start.pdb: The initial setup showing the starting positions of the peptides.

Data & Plots¶

All plots and data are saved in the plots/ and contact_maps/ directories:

• Energy vs. RMSD:
  • plots/E_RMSD_<chains>_total.svg: Total system energy.
  • plots/E_RMSD_<chains>_interaction.svg: Interaction energy between the peptide and receptor. Look for "funnels" where low interaction energy models cluster at specific RMSD values, indicating a stable binding site.
• RMSF Profile: plots/RMSF_seq.svg. Indicates the flexibility of the receptor residues during the docking process.
• Trajectory RMSD: plots/RMSD_frame_<chains>_replica_*.svg. Tracks the movement of the peptides relative to the starting position or reference.

Raw Data (output_data/)¶

• Cluster Density: output_data/cluster_names.txt. Lists the size and density of each cluster. Highly populated and dense clusters often indicate more probable binding sites.
• RMSD Data: output_data/all_rmsds_*.txt and medoids_rmsds_*.txt.
• Alignment: output_data/reference_alignment_*.csv (if a reference was provided).

Understanding PDB Columns¶

Just like in the flexibility workflow, PDB files use standard columns for metadata: * Occupancy (occ): Secondary structure numeric codes. * B-factor (bfac): In model_*.pdb files, this may store RMSF values of the receptor if configured.

Scoring and Clustering¶

CABS-dock uses structural clustering to identify the most probable binding sites. The 10 representative models returned are the centroids of the most populated clusters, ranked by their binding energy.

Related pages¶

⬆ Back to top

• Restraints: Detailed guide on guiding docking with manual restraints.
• Peptide Modeling: How to model cyclic or constrained peptides before docking.
• Examples: Docking case studies, including p53-MDM2.
• References: CABS-dock standalone (Bioinformatics 2019).

← Peptide Modeling | ⬆ Back to top | Next: Examples