Eagon Research Group

    Our research lab uses a number of programs to perform in silico screening assays (the testing of molecules with a computer program).  To do this, we first build a model of a target protein using either known crystal structures from NCBI or using homology models.  We then create a virtual library of molecules from either publicly available databases (such as the ZINC database) or using in-house software such as Spartan.  Although it is possible to test hundreds of millions of compounds, we typically try to screen less than 10 million compounds in order to keep the processing time reasonable.  We generally filter the large databases to more reasonable numbers by selecting for smaller molecules and those with good ADMET properties.   The programs we usually use to preform these include Autodock, Vina, and FRED.  Once the jobs are processed, the programs will indicate which molecules are suspected to be potent.  We then will synthesize the compounds and test them in a living system to see if the virtual screen results are correct.  When possible, we try to test several programs with real biological data prior to beginning a virtual screen to minimize false hits.

• Open Babel
  
• Autodock Vina
• Smina
• iDock
  
• ChimeraX
  
• Cygwin (for a Windows machine)
• Notepad++
• ChemSketch
• ACD NMR Processor
  

1. A PDBQT file of the protein target named foo_DOCK.pdbqt (where "foo" is usually the PDB code)
2. The iDock program
3. A directory named Ligands that contains all of the input ligand files in PDBQT format.  All input ligands must have a name beginning with the word "ligand" like this:  ligand02_1554.pdbqt.  These ligands also all need to have a "Name" in the file.  This is most commonly a ZINC ID code, or something similar.  Our script will look for this to help reference the compounds later.
   
4. A proper configuration file called idock.conf with the ligand folder, protein target, and coordinates set up.  In our example, we have placed all ligand files in a directory called "Ligands," and our protein target is called 2vt4_DOCK.pdbqt.
• The configuration file above is set to use 6 cores and with default values.  Time per ligand can vary a lot based on system resources and the size of input files, but iDock is very fast and most ligands will be processed in less than a minute.  We have found little improvement by increasing the default values based on a benchmarking study we did here.
  
5. The screen_idock_v4.2.sh script for a Linux system or the screen_idock_Win_v4.2.sh script for Windows.
   

Once everything is ready, simply execute the script in the background and let the job run.

1. A PDBQT file of the protein target named foo_DOCK.pdbqt
2. The Vina program (which is contained in a directory called Vina)
3. A directory named Ligands that contains all of the input ligand files in PDBQT format.  All input ligands must be named with the word "ligand" in front as with the iDock screen (e.g. ligand02_1554.pdbqt).
   
4. A properly formatted conf.txt file for Vina to read.  You can see an example here where the "#" entries need to be filled in (see the Docking the Classroom page for a full guide).
• Running timing tests are a good idea.  Generally, 6-12 CPUs are best depending on your setup.  Much larger numbers can actually make the program slower.
• We find the default settings with 10 conformations works well (see our example file above).
  
• As mentioned before, we don't see much improvement in docking accuracy when we increase the exhaustiveness setting, so we leave it at the default.
5. The screen_Vina_v4.2.sh script for a Linux system, or the screen_Vina_Win_v4.2.sh script for a Windows machine.