GitHub | GitLab mirror | Preprint

Examples¶

Welcome to the collection of ready-to-use examples for CABS-flex standalone 3. These examples cover the main workflows: protein flexibility, peptide modeling, and peptide–protein docking.

1. Protein Flexibility Examples¶

⬆ Back to top

1.1 Default simulation¶

To run CABS-flex using the default settings (see Modeling Scheme) use the following command:

CABSflex -i PDB/FILE

For example:

CABSflex -i 1hpw

will download 1hpw.pdb file from PDB to cache and start simulation on default settings, while

CABSflex -i 1hpw.pdb

would try to open file 1hpw.pdb from working directory.

1.2 pLDDT-guided simulation¶

For AlphaFold models, you can use pLDDT scores to guide the flexibility (regions with low pLDDT will be more flexible).

If the pLDDT scores are in the B-factor column of the PDB file:

CABSflex -i model_af.pdb -g plddt --protein-plddt pdb

If the pLDDT scores are in a separate JSON file (from AlphaFold):

CABSflex -i model_af.pdb -g plddt --protein-plddt confidence.json

1.3 Protein Flexibility Visualizations¶

The following galleries showcase CABS-flex simulations for various protein targets, grouped by PDB ID. Each section includes a brief description of the protein and visualizations of different structural features.

1C9S: Transcription attenuation protein MtrB¶

Transcription attenuation protein MtrB is a key regulator in Bacillus subtilis. The simulations capture the dynamic behavior of this protein, highlighting its structural flexibility and conformational sampling.

1BL8: Potassium channel KcsA¶

Potassium channel (KcsA) from Streptomyces lividans is a classic model for ion channel studies. These visualizations show the protein's stability and local fluctuations within its functional domains.

1EOV: Aspartyl-tRNA synthetase¶

Free Aspartyl-tRNA synthetase (AspRS) from yeast. The free yeast aspartyl-tRNA synthetase differs from the tRNA(Asp)-complexed enzyme by structural changes in the catalytic site, hinge region, and anticodon-binding domain.

1GOY: Ribonuclease Bi¶

Endoribonuclease Ribonuclease Bi complexed with guanosine-3'-phosphate (3'-GMP). The simulations illustrate the mobility of the enzyme, particularly around the active site and complexed ligand.

1IJZ: Human IL-13¶

Solution Structure of Human IL-13. The ensemble of models reflects the flexible regions of this cytokine, which are essential for its biological signaling and receptor interaction.

2. Peptide Modeling Examples¶

⬆ Back to top

2.1 Linear Peptide Prediction¶

To predict the structure of a linear peptide from its sequence:

CABSflex --peptide ACDEFGHIKLMNPQRSTVWY

2.2 Cyclic Peptide Prediction¶

To model a cyclic peptide with a disulfide bond between two cysteines (e.g., residues 2 and 8):

CABSflex --peptide ACDEFGHCKL --ca-rest-add 2:PEP 8:PEP 5.5 1.0

3. Peptide–Protein Docking Examples¶

⬆ Back to top

3.1 Default Docking¶

To run CABS-dock using the default settings (recommended for inexperienced users):

CABSdock -i protein-pdb-code -p peptide-sequence:peptide-secondary-structure

For example, to dock the HKLVQLLTTT peptide (with predicted secondary structure CHHHHHHHCC) to chain A of PDB structure 2FVJ:

CABSdock -i 2FVJ:A -p HKLVQLLTTT:CHHHHHHHCC

This command will: - Load the conformation of chain A from 2FVJ as the protein structure - Load "HKLVQLLTTT" peptide sequence with the secondary structure assignment "CHHHHHHHCC" - Set default simulation settings (no knowledge about the binding site; almost rigid backbone of the protein receptor; random initial peptide conformations and positions)

To extend the outputs with contact maps, config saving, and simulation files:

CABSdock -i 2FVJ:A -p HKLVQLLTTT:CHHHHHHHCC -M -C -S

3.2 Docking Visualizations¶

The following galleries showcase CABS-dock simulations for various protein-peptide targets, grouped by PDB ID.

1IHJ: N-terminal PDZ domain¶

The Peptide is 5 amino acid long with beta strand. Top 10 predicted peptide pose (dark cyan) vs experimental (magenta) are shown for both local and global docking.

1MVU: C219 monoclonal antibody scFv fragment¶

The Peptide is 13 amino acid long with helix secondary structure. Top 10 predicted peptide pose (dark cyan) vs experimental (magenta) are shown for both local and global docking.

1KL3: Streptavidin complexed with strep-tag II peptide¶

The peptide has an helical structure with 6 amino acid residues. Top 10 predicted peptide pose (dark cyan) vs experimental (magenta) are shown for both local and global docking.

1YCR: MDM2 BOUND TO THE TRANSACTIVATION DOMAIN OF P53¶

Here we show an example case of docking 15 residue peptide SQETFSDLWKLLPEN to MDM2 (1YCR).

The simulations were performed with following command for precise control of options and was run in triplicate.

#Replicate 1; for other replcates change r1 to r2 and r3
CABSdock -i 1iak.pdb:AB -p STDYGILQINSRW -R 1iak.pdb:AB:P -w 1iak_r1 -a 20 -y 50 -M -C -S -o A --renumber-residues-to-original --pdb-bfac-output A --csv-output A --json-output --dssp-output --ss-output --restraints-output --image-file-format svg --log --aa-rebuild T --aa-method cg2all --cg2all-representation calpha-sc --aa-rebuild-workers 24 --generate-chimera-visualizations --generate-pymol-visualizations --generate-notebook --aa-minimize 

The simulation results in 10 replicas and three replicates. The energy v/s RMSD plot is below:

Figure: Energy v/s RMSD plot for 1YCR. The panels shows the interaction energy and total energy of the system for the triplicate simulations. The medoids and the best RMSD pose from the best replica are highlighted in the plot.

Trajectory from the best replica. The video below shows the structural trajectory from the replica containing the lowest RMSD pose on the left panel, synchronized with its corresponding total energy vs RMSD scatter plot on the right. In the structural view, the receptor is shown as a blue cartoon with a semi-transparent surface, the simulated peptide trajectory is colored red, and the experimental reference peptide is displayed in green with 50% transparency for comparison. In the plot, a red tracking dot indicates the active frame, while the background scatter points represent the energy landscape across all replicates.

Top 10 medoids and best model. The docking simulations cluster the poses and prepare representative models. The videos below show the structural analysis of these final predictions undergoing a 360-degree rotation. The left video displays the top 10 cluster medoids superimposed, with all predicted peptides in red and the reference peptide in green. The right video highlights the single absolute lowest RMSD frame from the entire triplicate simulation against the reference structure, using the same color scheme.

← Peptide–Protein Docking | ⬆ Back to top | Next: Case Studies