sanger version

git-svn-id: svn+ssh://svn.internal.sanger.ac.uk/repos/svn/pathsoft/artemis/trunk@2322 ee4ac58c-ac51-4696-9907-e4b3aa274f04

etc/run_blastn.sanger

+#!/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_blastn.sanger,v 1.1 2005-02-21 11:42:18 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
+        echo    or: $0 file_of_filenames database
+        exit 1
+    fi
+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
+
+
+    ### change these lines:
+    EXEC=blastall
+
+    echo "about to start $EXEC with input from $INPUT_FILE and output to"
+    echo "$OUTPUT_FILE using database $DATABASE"
+
+
+    EXTRA_ARGS=
+
+    # add/change the flags to suit your site:
+
+    HOSTNAME=`hostname`
+    REMOTE=N
+    
+    case $HOSTNAME in
+        deskpro*)
+          REMOTE=Y ;;
+        *)
+    esac
+
+    if [ $REMOTE = "Y" ]; then
+      WDIR=`pwd`
+      export WDIR
+      rsh babel "cd $WDIR; lsrun -v blastwrap.pl $EXEC -d $DATABASE -i $INPUT_FILE -p blastn \
+                 $EXTRA_ARGS >! $OUTPUT_FILE"
+    else
+      lsrun -v blastwrap.pl $EXEC -d $DATABASE -i $INPUT_FILE -p blastn \
+        $EXTRA_ARGS 2>&1 > $OUTPUT_FILE |
+        tee ${PROG}_errors.new 1>&2
+    fi
+
+#   lsrun -v blastwrap.pl $EXEC -d $DATABASE -i $INPUT_FILE -p blastn \
+#     $EXTRA_ARGS 2>&1 > $OUTPUT_FILE | 
+#     tee ${PROG}_errors.new 1>&2
+
+    #### 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
+}
+
+(echo "#!/bin/sh -"; echo "kill $$") > $PROG.kill
+
+chmod a+x $PROG.kill
+
+if [ x$ONEFILE = x ]
+then
+    for i in `cat $1`
+    do
+        run_one_prog $i $i.out $EXPANDED_DATABASE
+    done
+
+else
+    run_one_prog $1 $2 $EXPANDED_DATABASE
+fi
+
+exit 0

etc/run_blastp.sanger

+#!/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_blastp.sanger,v 1.1 2005-02-21 11:45:09 tjc Exp $"
+
+PROG=`echo $RCS_HEADER | sed 's/.*run_\(.*\),v.*/\1/'`
+
+#BLASTDB=/data/blastdb
+#BLASTMAT=/data/blastdb/aa
+#export BLASTDB
+#export BLASTMAT
+
+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
+        echo    or: $0 file_of_filenames database
+        exit 1
+    fi
+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
+
+
+    ### change these lines:
+    EXEC=blastall
+
+    echo "about to start $EXEC with input from $INPUT_FILE and output to"
+    echo "$OUTPUT_FILE using database $DATABASE"
+
+
+    EXTRA_ARGS=
+
+    HOSTNAME=`hostname`
+    REMOTE=N
+
+    case $HOSTNAME in
+        deskpro*)
+          REMOTE=Y ;;
+        *)
+    esac
+
+    if [ $REMOTE = "Y" ]; then
+      WDIR=`pwd`
+      export WDIR
+      rsh babel "cd $WDIR; lsrun -R 'select[blast && mem > 500] rusage[r1m=1:mem=500]' -v \
+                           blastwrap.pl $EXEC -d $DATABASE -i $INPUT_FILE -p blastp \
+                           $EXTRA_ARGS >! $OUTPUT_FILE"
+    else
+    # add/change the flags to suit your site:
+      lsrun -R 'select[blast && mem > 500] rusage[r1m=1:mem=500]' -v \
+        blastwrap.pl $EXEC -d $DATABASE -i $INPUT_FILE -p blastp \
+        $EXTRA_ARGS 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
+}
+
+(echo "#!/bin/sh -"; echo "kill $$") > $PROG.kill
+
+chmod a+x $PROG.kill
+
+echo $1
+if [ x$ONEFILE = x ]
+then
+    for i in `cat $1`
+    do
+      echo "run_one_prog $i $i.out $EXPANDED_DATABASE"
+      run_one_prog $i $i.out $EXPANDED_DATABASE
+    done
+
+else
+    run_one_prog $1 $2 $EXPANDED_DATABASE
+fi
+
+exit 0

etc/run_blastx.sanger

+#!/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_blastx.sanger,v 1.1 2005-02-21 11:46:35 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
+        echo    or: $0 file_of_filenames database
+        exit 1
+    fi
+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
+
+
+    ### change these lines:
+    EXEC=blastall
+
+    echo "about to start $EXEC with input from $INPUT_FILE and output to"
+    echo "$OUTPUT_FILE using database $DATABASE"
+
+
+    EXTRA_ARGS=
+
+    # add/change the flags to suit your site:
+
+    HOSTNAME=`hostname`
+    REMOTE=N
+
+    case $HOSTNAME in
+        deskpro*)
+          REMOTE=Y ;;
+        *)
+    esac
+
+    if [ $REMOTE = "Y" ]; then
+      WDIR=`pwd`
+      export WDIR
+      rsh babel "cd $WDIR; lsrun -v blastwrap.pl $EXEC -d $DATABASE -i $INPUT_FILE -p blastx \
+                            $EXTRA_ARGS >! $OUTPUT_FILE"
+    else
+      lsrun -v blastwrap.pl $EXEC -d $DATABASE -i $INPUT_FILE -p blastx \
+          $EXTRA_ARGS 2>&1 > $OUTPUT_FILE |
+          tee ${PROG}_errors.new 1>&2
+    fi
+
+#   lsrun -v blastwrap.pl $EXEC -d $DATABASE -i $INPUT_FILE -p blastx \
+#     $EXTRA_ARGS 2>&1 > $OUTPUT_FILE | 
+#     tee ${PROG}_errors.new 1>&2
+
+    #### 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
+}
+
+(echo "#!/bin/sh -"; echo "kill $$") > $PROG.kill
+
+chmod a+x $PROG.kill
+
+if [ x$ONEFILE = x ]
+then
+    for i in `cat $1`
+    do
+        run_one_prog $i $i.out $EXPANDED_DATABASE
+    done
+
+else
+    run_one_prog $1 $2 $EXPANDED_DATABASE
+fi
+
+exit 0

etc/run_clustalx.sanger

+#!/bin/sh -
+
+# this script will run clustalx on a temporary file containing the concatenated
+# contents of files listed in a file of filenames
+
+
+RCS_HEADER="$Header: //tmp/pathsoft/artemis/etc/run_clustalx.sanger,v 1.1 2005-02-21 11:49:41 tjc Exp $"
+
+PROG=`echo $RCS_HEADER | sed 's/.*run_\(.*\),v.*/\1/'`
+
+
+if [ $# != 1 ]
+then
+    echo usage: $0 file_of_filenames
+fi
+
+(echo "#!/bin/sh -"; echo "kill $$") > $PROG.kill
+
+chmod a+x $PROG.kill
+
+file_of_filenames=$1
+date=`date +"%y_%m_%d"`
+hostname=`hostname`
+temp_file_name=clustalx/artemis_temp.$$.$hostname.$date.clustalx_input.fasta
+
+cat $file_of_filenames | xargs cat > $temp_file_name
+
+# make sure that the identifiers are unique for clustalx
+perl -pne 'if (/^>(\S+)/) {
+  $name = $1;
+  if (exists $h{$name}) {
+    $i = $h{$name}++;
+    s/^>(\S+)/>$name.$i/;
+  } else {
+    $h{$name} = 0;
+  }
+}' $temp_file_name > $temp_file_name.processed
+
+# delete it at some point
+echo "rm -f $temp_file_name* > /dev/null 2>&1" | at now + 8 hours
+
+HOSTNAME=`hostname`
+REMOTE=N
+
+case $HOSTNAME in
+    deskpro*)
+      REMOTE=Y ;;
+    *)
+esac
+
+if [ $REMOTE = "Y" ]; then
+  WDIR=`pwd`
+  export WDIR
+  ssh babel "cd $WDIR; clustalx $temp_file_name.processed"
+else
+  clustalx $temp_file_name.processed
+fi
+
+exit 0

etc/run_fasta.sanger

+#!/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,v 1.1 2005-02-21 11:51:10 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=/nfs/disk222/yeastpub/bio-soft/fasta/fasta33_t
+
+    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='
+
+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";
+
+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;
+
+'
+
+(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`
+      rsh 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

etc/run_fastx.sanger

+#!/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_fastx.sanger,v 1.1 2005-02-21 11:53:49 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
+
+
+# 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
+
+
+    ### change these lines:
+
+    FASTLIBS=/nfs/disk222/yeastpub/bio-soft/fasta/pubseqgbs export FASTLIBS
+    EXEC=/nfs/disk222/yeastpub/bio-soft/fasta/fastx33_t
+
+    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:
+    HOSTNAME=`hostname`
+    REMOTE=N
+
+    case $HOSTNAME in
+        deskpro*)
+          REMOTE=Y ;;
+        *)
+    esac
+
+    COMMAND="$EXEC -B -S -q -b 40 -H $INPUT_FILE $DATABASE ktup 2"
+    echo "command line: $COMMAND" 1>&2    
+
+    if [ $REMOTE = "Y" ]; then
+      WDIR=`pwd`
+      export WDIR
+      rsh babel "cd $WDIR; $COMMAND >! $OUTPUT_FILE"
+    else
+      $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
+}
+
+(echo "#!/bin/sh -"; echo "kill $$") > $PROG.kill
+
+chmod a+x $PROG.kill
+
+if [ x$ONEFILE = x ]
+then
+    for i in `cat $1`
+    do
+        run_one_prog $i $i.out $EXPANDED_DATABASE
+    done
+
+else
+    run_one_prog $1 $2 $EXPANDED_DATABASE
+fi
+
+exit 0