Site finder

Introduction

Prepare your own pocket exploration simulation to identify the most promising pockets in your protein and obtain putative binding sites of the ligand.

Check out our paper with a real-life application: Monte Carlo simulations using PELE to identify a protein–protein inhibitor binding site and pose

Inputs

  • protein-ligand PDB file

  • YAML file with parameters

Default parameters

Parameters for stage 1:

  • iterations: 50

  • pele steps: 12

  • number of structures: number of available CPUs - 1 (cannot be changed)

Parameters for stage 2:

  • iterations: 10

  • pele steps: 50

  • number of structures: number of available cpus / 6 (cannot be changed)

Recommendations

  1. Expected computational time is around 24 h.

  2. Initial position of the small molecule is irrelevant, since it will be extracted and randomly placed all around the protein.

  3. We recommend using at least 60 CPUs for the this package.

1. Complex Preparation

Prepare the system with Maestro (Protein Preparation Wizard): you must at least protonate the protein, but we also recommend removing any crystallization artefacts and water molecules as well as filling the missing loops and side chains, if possible.

Ensure the ligand has:

  • unique chain ID

  • no atom names with spaces or single letters

  • any residue name except UNK

2. Input Preparation

Prepare the input file input.yml:

system: 'docking2grid6n4b_thc.pdb'
chain: 'L'
resname: 'THC'
seed: 12345
#Distance to track along the simulation
atom_dist:
- "A:2:CA" #First atom to make the distance to
- "B:3:CG" #Second atom to make the distance to
site_finder: true

For more optional flags please refer to optional flags.

3. Run simulation

To run the system launch the simulation with the next command:

python -m pele_platform.main input.yml

4. Output

Best pockets ranked by ligand energy:

working_folder/refinement_simulation/results/clusters

Best snapshots ranked by ligand energy:

working_folder/refinement_simulation/results/top_poses/