mk.pl, dock.pm, and go.pl

# mk.pl
#
# Makes a set of classads for CONDOR.  Each classad
# defines a subset of ligand files to be docked.
# User specifies gene name, start index, sub set size, max startindex.
# May generate hundreds of classads.
#
# Author :  Peter Hebden,  phebden@gmail.com

$#ARGV == 3 || die "Usage: $0  GN start size maxStart";

$GN        = $ARGV[0];   # Gene name
$START     = $ARGV[1];   # starting with ligand file index number
$SIZE      = $ARGV[2];   # number of ligands in this batch
$MAX_START = $ARGV[3];   # maximum index of first ligand in batch;  
                         # last batch size may be truncated if not that many ligands 

for($i = $START; $i <= $MAX_START; $i += $SIZE)
{
  open(FD, ">d_${GN}_${i}.classad") || die "$!"; # save classad to file

  $end = $i + $SIZE - 1;
  print "\n making ads for GN=$GN $i to $end \n\n";    # following lines will be a classad
  print FD "
  universe     = vanilla
  Executable   = /usr/bin/perl
  Log          = perl.log

  RESERVED_SWAP = 0
      
  Error   = err.\$(Process)
  Input   = dock_${GN}_${i}_${end}.pl
  Output  = out.\$(Process)
  Log     = foo.log

  Queue";
  close FD;
  open(FD, ">dock_${GN}_${i}_${end}.pl") || die "$!";  # save to file; parameters used by dock.pm

  print FD "use dock;

  \$GN = \"$GN\";

  \$start = \"$i\";
  \$end   = \"$end\";

  runDock();\n";

  close FD;
}





# dock.pm
#
# This module is used by a unique perl script for each classad, 
# and runs Vina to dock ligands with receptor.
#
# Author:  Peter Hebden,  phebden@gmail.com

use Cwd;

sub runDock()
{
  $d = getcwd;

  $vina = "/common/home/YOUR-LOGIN-ID/bin/vina";
  $startingDir = "$d";

  $ligandFolder   = "$startingDir/ligand";
  $receptorFolder = "$startingDir/receptor-$GN";
  $outFolder      = "$startingDir/out-$GN";
  $logFolder      = "$startingDir/log-$GN";

  print "\nstarting dir = $startingDir \n";
  print "ligandFolder   = $ligandFolder \n";
  print "receptorFolder = $receptorFolder \n";
  print "outFolder      = $outFolder \n";
  print "logFolder      = $logFolder \n\n";


  -d $receptorFolder || die "receptor folder does not exist\n"; 
  -d $ligandFolder   || die "ligand folder does not exist\n"; 

  mkdir $logFolder;
  mkdir $outFolder;

  chdir($ligandFolder) or die "Error: failed to chdir to $path $!";

  opendir(DIR, $ligandFolder);
  @files = sort ( grep /pdbqt$/, readdir(DIR) ); 
  closedir(DIR);

  $count = 0;

  print "Docking ... \n";
  $t = localtime();
  print "$t\n";

  $config   = "${receptorFolder}/conf.txt";
  $receptor = "$receptorFolder/$GN.pdbqt";

  -e $config   || die "config file does not exist\n"; 
  -e $receptor || die "receptor file does not exist\n"; 

  print "\n config=$config \n receptor=$receptor \n";
  $max = $#files;

  if( $start < 0 ) { $start = 0; }

  if( $end > $max )  { $end = $max; }

  if ($start > $end ) {  $start = $end;  }

  @files = @files[$start..$end]; #    here we select a slice of ligands from the ligand folder

  foreach $file (@files) 
  {
     if ($file =~ /.*pdbqt$/) 
     {
      print "$file \n";
      $count += 1;
      if ($count % 100 == 1) { print "$count >> $ligandFolder/$file\n";  }  # feedback for when not using nohup
      
      $ligand   = "$ligandFolder/$file";
      $out      = "$outFolder/out_$file";
      $log      = "$logFolder/${file}_log.txt";
      
      if ( ! -e $out ) 
      {
      `$vina --config $config --ligand $ligand --receptor $receptor --out $out --log $log`  
      }
     }   
  }
  $t = localtime();
  print "$t\n";

  print "docked $count ligands\n";
}
1






# go.pl
#
# submits all classads in each docking folder specified on command line.
#
# Author:  Peter Hebden,  phebden@gmail.com

# docking-ID is the folder that contains classad files and Vina in/output files

use Cwd;

$#ARGV == 0 || die "Usage: $0 docking-ID\n";  

$dr = "$ARGV[0]";
print "\n";

@dirs = glob "$dr*";

foreach $d (@dirs) 
{
  chdir $d;
  print "$d \n";
  @files = glob "*.classad";

  foreach $f (@files)
  {
      `condor_submit $f`;     #  submits classad to condor, job waits in queue as Idle until Run.
       print "$f \n";
  }
  chdir "..";
}