The scripts and input files that accompany this demo can be found in the demos/protocol_captures directory of the Rosetta weekly releases.

This is a protocol capture, and represents the protocol at a fixed point in time. It may not work with the current version of Rosetta.


Author: Gordon Lemmon

  • Gordon Lemmon, Jens Meiler (2012). Toward ligand docking including explicit interface water molecules. PLoS ONE (submitted).

Small molecule docking predicts the interaction of a small molecule ligand with a protein at atomic-detail accuracy including position and conformation the ligand but also conformational changes of the protein upon ligand binding. While successful in the majority of cases, leading docking algorithms including RosettaLigand fail in some cases to predict the correct protein/ligand complex structure. In this study we show that simultaneous docking of explicit interface water molecules greatly improves Rosetta’s ability to distinguish correct from incorrect ligand poses. This result holds true for both protein-centric water docking wherein waters are located relative to the protein binding site and ligand-centric water docking wherein waters move with the ligand during docking. Protein-centric docking is used to model 99 HIV-1 protease/protease inhibitor structures. We find protease inhibitor placement improving at a ratio of 9:1 when one critical interface water molecule is included in the docking simulation. Ligand-centric docking is applied to 341 structures from the CSAR benchmark of diverse protein/ligand complexes. Across this diverse dataset we see up to 56% recovery of failed docking studies, when waters are included in the docking simulation.

Purpose and algorithm

The simultaneous docking of ligands and water molecules is now possible within Rosetta This work is being submitted to the PLoS ONE Special collection from RosettaCon 2012

Waters can be docked in two ways. The first way is refered to in the paper as protein-centric water docking. Protein centric waters sample the protein binding pocket independent of the ligand. Ligand-centric waters are positioned initially around the ligand. As the ligand translates and rotates about the protein binding site, the waters move with it, maintaining their positions relative to the protein. After low-resolution docking of the ligand. The waters undergo their own independent but smaller movements.

Tools and Input Files


  • Use to identify waters within the protein/ligand interface. The script takes 3 arguments, each specifying a PDB file (protein.pdb, ligand.pdb, water.pdb). This script requires that you have installed pymol as a python library

  • The script ligand_properties_from_bcl requires that you download and install BCL, available from the Meiler Lab website,

Required flags:

-in:file:s <pdb file> # a starting structure upon which docking will be performed. Should contain a protein, a ligand, and one or more waters
-in:file:extra_res_fa # the .params file for your ligand. This is created by providing a .mol file to the script: rosetta_source/src/python/apps/public/
-treat_residues_with_this_chain_as_separate_chemical_entities <1 letter chain from PDB> # Useful for giving waters the same chain and having Rosetta treat them separately.

Optional flags:

-ex1, -ex1aro, and -ex2 # expand the rotamer sets that are sampled during packing.
-in:file:native # allows calculation of comparison metrics between Rosetta models and the correct pose if this is known

Example Rosetta Command Line (Use Rosetta3.5 or revision 48472):

The three XML files, standard.xml, protein_centric.xml and ligand_centric.xml, demonstrate how to dock waters using RosettaLigand. Simply run the following command from the directory where this readme is found:

<path to rosetta_source>/bin/rosetta_scripts.linuxgccrelease @rosetta_inputs/flags.txt -parser:protocol <choice of xml file> -database <path to rosetta_database>

Expected Outputs

This demo only produces one structure. Add "-nstruct 1000" to produce the amount of sampling used in this paper. In our paper we sort the top 1000 by total score, then the top 100 by interface score. The top model by interface score is used as our most likely prediction of ligand pose

To sort models by total score simply:

grep -H  total_score *.pdb | sort -nk 2 | head -n 100 | cut -d ':' -f 1 > top100.txt

To sort by interface score:

grep -H interface_delta_X *.pdb `cat top100.txt` | sort -nk 2 | head -n 1