#!/bin/sh -

# this script will run a search program on a sequence input file or on each
# file in a file of filenames

# to customise this script see the function called run_one_prog below


RCS_HEADER="$Header: //tmp/pathsoft/artemis/etc/run_fasta.sanger.linux,v 1.1 2008-01-24 10:48:29 tjc Exp $"

PROG=`echo $RCS_HEADER | sed 's/.*run_\(.*\),v.*/\1/'`


if [ $# = 4 -a x$1 = x-onefile ]
then
    shift
    ONEFILE=t
    DATABASE=$3 export DATABASE
else
    if [ $# = 2 ]
    then
        DATABASE=$2 export DATABASE
    else
        echo usage: $0 -onefile input_file output_file database 1>&2
        echo    or: $0 file_of_filenames database 1>&2
        exit 1
    fi
fi


#
# Exchange DB name for fasta reference
#
if [ "$DATABASE" = "%uniprot" ]; then
  DATABASE="%U";
elif [ "$DATABASE" = "%uniprot_archaea" ]; then
  DATABASE="%A";
elif [ "$DATABASE" = "%uniprot_bacteria" ]; then
  DATABASE="%B";
elif [ "$DATABASE" = "%uniprot_eukaryota" ]; then
  DATABASE="%E";
elif [ "$DATABASE" = "%uniprot_viruses" ]; then
  DATABASE="%V";
elif [ "$DATABASE" = "%uniprot_rest" ]; then
  DATABASE="%R";
elif [ "$DATABASE" = "%malaria" ]; then
  DATABASE="%M";
elif [ "$DATABASE" = "%kineto_aa" ]; then
  DATABASE="%K";
fi

# expand any ~ or environment variables
EXPANDED_DATABASE=`echo "echo $DATABASE" | /bin/csh -f`

### change this function to suit your site:

run_one_prog () {
    INPUT_FILE=$1
    OUTPUT_FILE=$2
    DATABASE=$3

### strip out directory from command line

    IN=`echo $INPUT_FILE | sed -n -e "s|$PWD//||p"`
 
    if test "$IN" != "" && test -f $IN; then
      INPUT_FILE="$IN"
    fi
   
    echo "\n\nIN=$IN\nPWD=$PWD\nINPUT_FILE=$INPUT_FILE\n\n"
    ### change these lines:

    ### get sequence size
    seq_size=`infoseq "$INPUT_FILE" -length -only -auto | awk '{ sum += $1 } END { print sum }'`

    FASTLIBS=/nfs/disk222/yeastpub/bio-soft/fasta/pubseqgbs export FASTLIBS
    EXEC=/software/pathogen/external/applications/fasta/fasta35

    echo "about to start $EXEC with input from $INPUT_FILE and output to" 1>&2
    echo "$OUTPUT_FILE using database $DATABASE" 1>&2

    # add/change the flags to suit your site:
    COMMAND="$EXEC -B -S -q -b 100 -H $INPUT_FILE $DATABASE ktup 2"

    echo "command line: $COMMAND" 1>&2

#    lsrun -R 'select[blast && mem > 500] rusage[r1m=1:mem=500]' -v $COMMAND 2>&1 > $OUTPUT_FILE | 

     if [ "$seq_size" -lt 50000 ]
     then
       bsub -q normal -n 1 -R 'select[blast && mem > 500] rusage[r1m=1:mem=500]' -I $COMMAND 2>&1 > $OUTPUT_FILE |
         tee ${PROG}_errors.new 1>&2
     else
       bsub -q "longblastq" -n 1 -R 'select[blast && mem > 500] rusage[r1m=1:mem=500]' -I $COMMAND 2>&1 > $OUTPUT_FILE |
         tee ${PROG}_errors.new 1>&2
     fi

    #### end of changes


    # Artemis can read compressed files
    gzip -9 $OUTPUT_FILE &

    if [ -s ${PROG}_errors.new ]
    then
        ( echo ERROR running $PROG: ; echo; 
          echo ===================================================
          cat ${PROG}_errors.new ) >> $OUTPUT_FILE
        cat ${PROG}_errors.new >> ${PROG}_errors
    fi
}

PERL_PROG_1='

local *BSUB;

my $file     = $ARGV[0];
my $database = $ARGV[1];
my $pwd      = $ARGV[2];
chomp $file;
chomp $database;

$ENV{'FASTLIBS'} = "/nfs/disk222/yeastpub/bio-soft/fasta/pubseqgbs_test";


open(BSUB, "| bsub -q normal -o fasta_errors -n 1 -R \"select[blast && mem > 500] rusage[r1m=1:mem=500]\" -K") or die "could not open bsub pipe : $!";
open(LIST_FILE,$file);

$EXEC="/nfs/disk222/yeastpub/bio-soft/fasta/fasta33_t";
$EXEC="/software/pathogen/external/applications/fasta/fasta35";

while(my $inFile = <LIST_FILE>)
{
  chomp($inFile);

  if($inFile =~ m/^($pwd)(.*)/)
  {
    my $inFile_tmp = $2;
    while($inFile_tmp =~ m/^(\/)(.*)/)
    {
      $inFile_tmp = $2;
    }

    if( -e $inFile_tmp )
    {
      $inFile = $inFile_tmp;
    }
  }
 
  if($inFile =~ m/^(\S{100})/)
  {
    if($inFile =~ m/^(\S{90,})(fasta\/\S+)/)
    {
      my $inFile_tmp = $1;

      if( -e $inFile_tmp )
      {
        print BSUB "cd $inFile_tmp\n";
        $inFile = $2;
      }
    }
  }

  print BSUB "$EXEC -B -S -q -b 100 -H $inFile $database ktup 2 > $inFile\.out\n";
  print BSUB "gzip -9 $inFile\.out\n";
}
close BSUB or die "--Could not submit job : $!";
close LIST_FILE;
'



PERL_PROG='

local *BSUB;

my $file     = $ARGV[0];
my $database = $ARGV[1];
my $pwd      = $ARGV[2];
chomp $file;
chomp $database;

$ENV{'FASTLIBS'} = "/nfs/disk222/yeastpub/bio-soft/fasta/pubseqgbs_test";
open(LIST_FILE,$file);

$EXEC="/nfs/disk222/yeastpub/bio-soft/fasta/fasta33";
$EXEC="/software/pathogen/external/applications/fasta/fasta35";
$NEW_WDIR=".";
$NUM_JOBS=0;

while(my $inFile = <LIST_FILE>)
{ 
  $NUM_JOBS++;
  chomp($inFile);

  if($inFile =~ m/^($pwd)(.*)/)
  {
    my $inFile_tmp = $2;
    while($inFile_tmp =~ m/^(\/)(.*)/)
    {
      $inFile_tmp = $2;
    }

    if( -e $inFile_tmp )
    {
      $inFile = $inFile_tmp;
    }
  }
 

  if($inFile =~ m/^(\S{100})/)
  {
    if($inFile =~ m/^(\S{90,})(fasta\/\S+)/)
    {
      my $inFile_tmp = $1;

      if( -e $inFile_tmp )
      {
        $NEW_WDIR=$inFile_tmp;
        $inFile = $2;
      }
    }
  }
  # find number of leading zero

  if($inFile =~ m/^(.*)(fasta\/)([^\/]*)(0{4})(\d{1})$/)
  {
    push(@jobs10, $inFile);
  }
  elsif($inFile =~ m/^(.*)(fasta\/)([^\/]*)(0{3})(\d{2})$/)
  {
    push(@jobs100, $inFile);
  }
  elsif($inFile =~ m/^(.*)(fasta\/)([^\/]*)(0{2})(\d{3})$/)
  {
    push(@jobs1000, $inFile);
  }
  elsif($inFile =~ m/^(.*)(fasta\/)([^\/]*)(0{1})(\d{4})$/)
  {
    push(@jobs10000, $inFile);
  }
  else
  {
    push(@jobs100000, $inFile);
  }
}
close LIST_FILE;

if(defined @jobs10)
{
  my( $num ) = sprintf( "%04d", 0);
  submit($num, @jobs10);
}

if(defined @jobs100)
{
  my( $num ) = sprintf( "%03d", 0);
  submit($num, @jobs100);
}

if(defined @jobs1000)
{
  my( $num ) = sprintf( "%02d", 0);
  submit($num, @jobs1000);
}

if(defined @jobs10000)
{
  my( $num ) = sprintf( "%01d", 0);
  submit($num, @jobs10000);
}

if(defined @jobs100000)
{
  submit("", @jobs100000);
}

for($count = 0; $count < @bsub_jobs; $count++)
{
  open(BSUB, "| bsub -q normal -R \"select[mem > 1] rusage[mem=1]\" -o /dev/null -e /dev/null -w \"ended($bsub_jobs[$count])\" -K -J \"$bsub_jobs[$count]\_fin\"")  or die "could not open bsub pipe  : $!";
  print BSUB "\"echo finished > /dev/null\" > /dev/null 2> /dev/null";
  close BSUB; # or die "--Could not submit job : $!";
} 
 

sub submit
{
  my ($num, @jobs) = @_;
   
  my $prefix;
  my @starts;
  my @prefixes;

  if($jobs[0] =~ m/^(.*)(\/fasta\/)([^\/]*)(\d{5})/)
  {
    $prefix = $1.$2.$3;
    push(@starts, "$4");
  }
  elsif($jobs[0] =~ m/^(fasta\/)([^\/]*)(\d{5})/)
  {
    $prefix = $1.$2;
    push(@starts, "$3");
  }

  # escape characters
  if($prefix =~ /(\+|\?|\*|\[|\])/)
  {
    $prefix =~ s/\+/\\+/;
    $prefix =~ s/\?/\\?/;
    $prefix =~ s/\*/\\*/;
    $prefix =~ s/\[/\\[/;
    $prefix =~ s/\]/\\]/;
  }
  push(@prefixes, $prefix);

  
  #different entries have different prefixes
  for($count =0; $count < @jobs; $count++)
  {
    if($jobs[$count] !~ m/^$prefix(.*)/)
    {
      if($jobs[$count] =~ m/^(.*)(fasta\/)([^\/]*)(\d{5})/)
      {
        $prefix = $1.$2.$3;

        push(@starts, "$4");

        if($prefix =~ /(\+|\?|\*|\[|\])/)
        {
          $prefix =~ s/\+/\\+/;
          $prefix =~ s/\?/\\?/;
          $prefix =~ s/\*/\\*/;
          $prefix =~ s/\[/\\[/;
          $prefix =~ s/\]/\\]/;
        }

        push(@prefixes, $prefix);
      }
    }
  }

  for($count =0; $count < @prefixes; $count++)
  {
    $prefix = $prefixes[$count];
    $start  = $starts[$count];

    # build the index description that need to be run
    $index="$start-";
    $end="$start";
    for($j =0; $j < @jobs; $j++)
    {
      if($jobs[$j] =~ m/^$prefix(.*)/)
      {
        if($jobs[$j] =~ m/^(.*)(fasta\/)([^\/]*)(\d{5})/)
        {
          if($j == @jobs-1)
          {
            $index = "$index$4";
          }
          elsif($end+1 >= $4)
          {
            $end = "$4";
          }
          else
          {
            $index = "$index$end,$4-"
          }
        }
      }
    }

    if($index =~ m/(\-)$/)
    {
      $index = "$index$end";
    }

    bsub($prefix, $index, $num);
  }
}


# start job arrays
sub bsub
{
  my ($prefix, $index, $num) = @_;

  my $name = $prefix;
 
  if($prefix =~ m/(\/fasta\/)(.*)/)
  {
    $name = "$2";
  }

  my $random = int( rand( 999+1 ) );
  
  push(@bsub_jobs, "$name$random\_fasta");

  my $QUEUE="long";
  if($NUM_JOBS <= 6)
  {
    $QUEUE="normal";
  }

  open(BSUB, "| bsub -q $QUEUE -o /dev/null -n 1 -R \"select[mem > 500] rusage[mem=500]\" -J$name$random\_fasta\"[$index]%16\"")  or die "could not open bsub pipe  : $!";
  print BSUB "cd $NEW_WDIR\n";
  print BSUB "$EXEC -B -S -q -b 100 -H $prefix$num";
  print BSUB "\${LSB_JOBINDEX} $database ktup 2 > $prefix$num";
  print BSUB "\${LSB_JOBINDEX}\.out\n";

  print BSUB "gzip -9f $prefix$num";
  print BSUB "\${LSB_JOBINDEX}\.out\n";

 close BSUB or die "--Could not submit job : $!";
}

'

(echo "#!/bin/sh -"; echo "kill $$") > $PROG.kill

chmod a+x $PROG.kill

HOSTNAME=`hostname`
REMOTE=N

case $HOSTNAME in
    deskpro*)
      REMOTE=Y ;;
    *)
esac

if [ x$ONEFILE = x ]
then
   if [ $REMOTE = "Y" ]; then
      WDIR=`pwd`
      export WDIR
      CMD=`echo $PERL_PROG_1`
      ssh babel "cd $WDIR; perl -w -e '$CMD' \"$1\" \"$EXPANDED_DATABASE\" \"$PWD\""
   else
      perl -w -e "$PERL_PROG" "$1" "$EXPANDED_DATABASE" "$PWD"
   fi
else
    run_one_prog $1 $2 $EXPANDED_DATABASE
fi

exit 0