HADDOCK

FAQ

We will collect here problems and solutions. The following topics are currently available:

• What about missing atoms or point mutations?
• What about ions?
• Domain definition for docking
• Clustering issues
• Increasing the number of flexible segments
• Running in batch mode using a queuing system
• Small ligands docking with HADDOCK

What about missing atoms or point mutations?

Missing atoms will be automatically generated by HADDOCK when generating the topologies and PDB files of the molecules in the begin directory. This is done when running the generate_X.inp CNS scripts (these are originally based on the generate_easy.inp script from the CNS distribution). HADDOCK uses the PDB files defined in the new.html file. In case of missing residues, chain breaks will be introduced.

When setting up the docking from an ensemble of structures it can happen that single point mutants coordinates files are available. These can be used as well provided you do the following:

• edit the PDB file and rename the mutated residue to the proper amino acid name
• keep or rename appropriately the matching side-chain atoms

The missing atoms will be generated automatically. It is important to have at least the backbone atoms defined since their average position will be used as starting point to "grow" the missing atoms. Always check that the sequence of the various PDB files match!

What about ions?

Some proteins contains ions such as for example calcium. Their inclusion might be important for docking purposes, in particular for proper electrostatics! In principle, they should be recognized in the topology generation step provided their name in the PDB file matches the ion names defined in the ion.top file in the toppar directory. To avoid that a N- or C-terminal patch be applied to them, they should also be defined in the topallhdg5.3.pep file (look for the "first IONS" and "last IONS" statements).

Another problem can occur with ions in torsion angle dynamics since they are unconnected single atoms. In the new HADDOCK 2.X version, for the torsion angle dynamics part of the docking protocol (it1), a covalent bond will be automatically defined to the closest ligand atom (only for cations). This is done in the covalions.cns CNS script in the protocols directory; the following cations are currently defined: MG⁺², CA⁺², FE⁺², FE⁺³, NI⁺², CO⁺², CO⁺³, CU⁺¹, CU⁺² and ZN⁺². If your system contains other ions add them to the covalions.cns file (they should however be defined in ion.top).

Domain definition for docking

In general, it is recommended to remove any part of your system such as flexible linkers that are not involved in the interaction with the partner for docking. Keeping these might give trouble in the sorting of solutions. For example, such a linker can make contacts with the partner molecule, resulting in a lower total energy and, in that way, "bad" solutions could still be kept.

Clustering issues

When choosing which of the two molecules will be in the first segid (e.g. "A"), it is recommended to choose the largest and/or most rigid one of the two. This should give better clustering results since in the rmsd calculation for clustering (rmsd.inp CNS script) the structures are first fitted on the semi-flexible segments of the first molecule and then the rmsds are calculated on the semi-flexible segments of the remaining molecules (defined as "ligand interface RMSD").

Defining the largest and best defined (most rigid) molecule first should thus result in a better fitting.

Increasing the number of flexible segments

In the current distribution, the number of flexible segment is set to 10 for the semi-flexible interface and 5 for the fully flexible segments. If needed, these numbers can be increased. The only file that you will need to modify is run.cns.

As an example, say you wish to increase the number of fully flexible segments for molecule A to 10. Locate in run.cns the section concerning the fully flexible segments, e.g.:

Running in batch mode using a queuing system

There are several ways to run HADDOCK in batch mode on cluster. Here are a few examples provides by users.

Running HADDOCK on a cluster without queuing system

a contribution from Melissa Stauffer, Vanderbilt University, USA

Here is a description of the setup on a Linux cluster that has no queuing system installed. It has a head node/firewall with 16 dual-cpu compute nodes behind it. From the head node, processes can be spawned to the compute nodes using rsh. Authentication to compute nodes is done via the /etc/hosts.equiv mechanism - no prompting for passwords. HADDOCK was setup to use each of the 32 compute-node CPUs by creating a simple csh wrapper script called "haddock_wrapper" that contains two lines like this:

cd project_directory
csh -f $1

{===>} queue_1="rsh node1 ~/bin/haddock_wrapper";
{===>} cns_exe_1="cns";
{===>} cpunumber_1=2;

Running HADDOCK on a cluster using the PBS queuing system

a contribution from Nuno Ferreira, Universidade de Coimbra, Portugal

Here is an example on how to run HADDOCK in two nodes with PBS queuing system.

Write a batch executable script

--> masterqueue is the routing queue
--> clustermaster is the head, where you submit the job
--> node1 and node2 are the name of the two requested nodes
--> after editing all the files, submit job with qsub job.pbs

The PBS script (job.pbs) should look like:

#!/bin/bash
#PBS -q masterqueue@clustermaster
#PBS -l nodes=1:node1+1:node2
#PBS -l cput=24:00:00
#PBS -N haddock_test
#PBS -o haddock.out
#PBS -e haddock.err

cd /home/haddock/e2a-hpr/run1
/usr/bin/python $HADDOCK/Haddock/Runhaddock.py

Change your run.cns script

{===>} queue_1="rsh node1 /home/haddock/e2a-hpr/haddock_wrapper";
{===>} cns_exe_1="/home/cns_solve_1.1_ev67/dec-alpha-linux/bin/cns";
{===>} cpunumber_1=1;

{===>} queue_2="rsh node2 /home/haddock/e2a-hpr/haddock_wrapper";
{===>} cns_exe_2="/home/cns_solve_1.1_ev67/dec-alpha-linux/bin/cns";
{===>} cpunumber_2=1;

haddock_wrapper is a script similar to the one described above.

Small ligand docking with HADDOCK

It's possible to dock small ligands using HADDOCK but for that topology and parameter files for the ligand should be provided in CNS format. Several sources exist to find such files:

• the PRODGR server maintained by Daan van Aalten at Dundee University: http://davapc1.bioch.dundee.ac.uk/prodrg
  This server allows you to draw your molecule or paste coordinates and will return topologies and parameter files in various format, including CNS. You should turn on the electrostatic to obtain partial charges. Save the resulting PDB file and the corresponding CNS parameter and topology files to use in HADDOCK.

  Important: The generated parameter file contains a CNS NBONds statement which should be removed prior to use in HADDOCK. Look in the parameter file for:

       NBONds
         CUTNB=7.0 WMIN=1.5
         REPEL=1.0 REXPONENT=4
         IREXPONENT=1 RCONST=16.0
         TOLERANCE=0.5 NBXMOD=5
         CTONNB=5.5 CTOFNB=6.0
       END
  

  and remove or comment it out (by adding ! before each line).

• the HIC-Up database maintained by Gerard Kleywegt at Uppsalla University: http://xray.bmc.uu.se/hicup
  One problem with those files for use in docking is that typically all partial charges are set to zero, meaning that the electrostatic interaction energies will thus be zero unless you change the partial charges. Still, the HIC-Up topology and parameter files provide a good starting point.

For docking small ligand with HADDOCK using custom-made topology and parameter files and perform the following steps:

1. Setup your HADDOCK run in the usual way, i.e. generating the new.html file and running haddock first to generate the run directory structure.
2. Place your custom topology and parameter files in the toppar directory.
3. When modifying the docking parameters in run.cns, specify the proper topology and parameter files for your ligand.
4. To avoid that a N- or C-terminal patch be applied to your ligand, add "first IONS" and "last IONS" statements with the name of your ligand in the topallhdg5.3.pep file in the toppar directory (look for the "first IONS" and "last IONS" statements).

Important1: When starting a run, always check for error messages in the begin directory in the various generate...out files, especially for your ligand.

Please send any suggestions or enquiries to Alexandre Bonvin