SGE Cluster Docking

SGE Cluster Information

• 'sgehead.compbio.ucsf.edu' is the submit machine for the Sun Grid Engine (SGE) cluster. wilco is also authorized to submit jobs.
• 'sgemaster.compbio.ucsf.edu' is the admin machine for the SGE cluster.
• There are around 250 cluster nodes providing 600 total cores to run jobs in the sge queue as of May, 2009, named like 'node-1-1' through 'node-3-36' where the first number is the rack # and the second is the slot # in that rack.

SGE Commands

• qsub: submit jobs
  
• qstat: check job status
  
• qdel: remove jobs
  
• qhost: check cluster status
  
• man sge_intro: start of manpage documentation
  

Typical Docking Workflow

• Generate spheres and grids - See Using MakeDOCK for more information, including how to prepare the receptor and ligand

ssh sgehead.compbio.ucsf.edu   # ssh to SGE submit machine
mkdir example                  # make docking directory
cd example                     # change to docking directory
cp <somedir>/rec.pdb .         # copy or create rec.pdb
cp <somedir>/xtal-lig.mol2 .   # copy or create xtal-lig.mol2 (or even xtal-lig.pdb)
startdockblaster4              # create spheres and grids                 
                               # Check output for WARNING messages, correct as needed

• Submit docking run

cp calibrate/INDOCK.1.A INDOCK # copy or create INDOCK
md4db.csh bysubset 2 50        # create directories for docking run with 50 chunks
                               # 2 indicates we want the fragment-like subset of ZINC (See http://zinc.docking.org/subset1)
cd run.2                       # chdir into run.2 directory
startdockbks3 .                # submit database chunks to SGE cluster

For information on which ZINC