From b19c162e20dfc02467db28b7449d1e14417e0c99 Mon Sep 17 00:00:00 2001 From: tjc <tjc@ee4ac58c-ac51-4696-9907-e4b3aa274f04> Date: Mon, 21 Feb 2005 11:54:10 +0000 Subject: [PATCH] generic version git-svn-id: svn+ssh://svn.internal.sanger.ac.uk/repos/svn/pathsoft/artemis/trunk@2323 ee4ac58c-ac51-4696-9907-e4b3aa274f04 --- etc/run_blastn | 29 ++--------- etc/run_blastp | 28 ++-------- etc/run_blastx | 29 ++--------- etc/run_clustalx | 24 ++------- etc/run_fasta | 132 ++++------------------------------------------- etc/run_fastx | 24 ++------- 6 files changed, 32 insertions(+), 234 deletions(-) diff --git a/etc/run_blastn b/etc/run_blastn index fdcaa65a6..513ff047a 100755 --- a/etc/run_blastn +++ b/etc/run_blastn @@ -6,7 +6,7 @@ # to customise this script see the function called run_one_prog below -RCS_HEADER="$Header: //tmp/pathsoft/artemis/etc/run_blastn,v 1.2 2004-10-05 13:59:39 tjc Exp $" +RCS_HEADER="$Header: //tmp/pathsoft/artemis/etc/run_blastn,v 1.3 2005-02-21 11:42:46 tjc Exp $" PROG=`echo $RCS_HEADER | sed 's/.*run_\(.*\),v.*/\1/'` @@ -50,30 +50,9 @@ run_one_prog () { 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 + nice -19 $EXEC -d $DATABASE -i $INPUT_FILE -p blastn \ + $EXTRA_ARGS 2>&1 > $OUTPUT_FILE | + tee ${PROG}_errors.new 1>&2 #### end of changes diff --git a/etc/run_blastp b/etc/run_blastp index 95a98a318..853a249d3 100755 --- a/etc/run_blastp +++ b/etc/run_blastp @@ -6,7 +6,7 @@ # to customise this script see the function called run_one_prog below -RCS_HEADER="$Header: //tmp/pathsoft/artemis/etc/run_blastp,v 1.2 2004-10-05 12:40:05 tjc Exp $" +RCS_HEADER="$Header: //tmp/pathsoft/artemis/etc/run_blastp,v 1.3 2005-02-21 11:45:33 tjc Exp $" PROG=`echo $RCS_HEADER | sed 's/.*run_\(.*\),v.*/\1/'` @@ -49,28 +49,10 @@ run_one_prog () { 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 + nice -19 $EXEC -d $DATABASE -i $INPUT_FILE -p blastp \ + $EXTRA_ARGS 2>&1 > $OUTPUT_FILE | + tee ${PROG}_errors.new 1>&2 #### end of changes @@ -95,7 +77,7 @@ if [ x$ONEFILE = x ] then for i in `cat $1` do - run_one_prog $i $i.out $EXPANDED_DATABASE + run_one_prog $i $i.out $EXPANDED_DATABASE done else diff --git a/etc/run_blastx b/etc/run_blastx index 700f09f3b..4172bfbad 100755 --- a/etc/run_blastx +++ b/etc/run_blastx @@ -6,7 +6,7 @@ # to customise this script see the function called run_one_prog below -RCS_HEADER="$Header: //tmp/pathsoft/artemis/etc/run_blastx,v 1.2 2004-10-05 14:53:49 tjc Exp $" +RCS_HEADER="$Header: //tmp/pathsoft/artemis/etc/run_blastx,v 1.3 2005-02-21 11:46:53 tjc Exp $" PROG=`echo $RCS_HEADER | sed 's/.*run_\(.*\),v.*/\1/'` @@ -50,30 +50,9 @@ run_one_prog () { 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 + nice -19 $EXEC -d $DATABASE -i $INPUT_FILE -p blastx \ + $EXTRA_ARGS 2>&1 > $OUTPUT_FILE | + tee ${PROG}_errors.new 1>&2 #### end of changes diff --git a/etc/run_clustalx b/etc/run_clustalx index f7b5a603a..0b9dd6a12 100755 --- a/etc/run_clustalx +++ b/etc/run_clustalx @@ -4,7 +4,7 @@ # contents of files listed in a file of filenames -RCS_HEADER="$Header: //tmp/pathsoft/artemis/etc/run_clustalx,v 1.2 2004-10-05 15:21:47 tjc Exp $" +RCS_HEADER="$Header: //tmp/pathsoft/artemis/etc/run_clustalx,v 1.3 2005-02-21 11:50:01 tjc Exp $" PROG=`echo $RCS_HEADER | sed 's/.*run_\(.*\),v.*/\1/'` @@ -21,7 +21,7 @@ chmod a+x $PROG.kill file_of_filenames=$1 date=`date +"%y_%m_%d"` hostname=`hostname` -temp_file_name=clustalx/artemis_temp.$$.$hostname.$date.clustalx_input.fasta +temp_file_name=/tmp/artemis_temp.$$.$hostname.$date.clustalx_input.fasta cat $file_of_filenames | xargs cat > $temp_file_name @@ -36,24 +36,6 @@ perl -pne 'if (/^>(\S+)/) { } }' $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 +clustalx $temp_file_name.processed exit 0 diff --git a/etc/run_fasta b/etc/run_fasta index 6e4f9b95c..31147c37b 100755 --- a/etc/run_fasta +++ b/etc/run_fasta @@ -6,7 +6,7 @@ # to customise this script see the function called run_one_prog below -RCS_HEADER="$Header: //tmp/pathsoft/artemis/etc/run_fasta,v 1.11 2005-01-14 14:24:40 tjc Exp $" +RCS_HEADER="$Header: //tmp/pathsoft/artemis/etc/run_fasta,v 1.12 2005-02-21 11:51:32 tjc Exp $" PROG=`echo $RCS_HEADER | sed 's/.*run_\(.*\),v.*/\1/'` @@ -28,30 +28,10 @@ else 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 () { @@ -59,41 +39,21 @@ run_one_prog () { 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 + EXEC=fasta34 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 $DATABASE ktup 2" + COMMAND="$EXEC -B -S -q -b 40 -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 + nice -19 $COMMAND 2>&1 > $OUTPUT_FILE | + tee ${PROG}_errors.new 1>&2 #### end of changes @@ -110,87 +70,17 @@ run_one_prog () { 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 + for i in `cat $1` + do + run_one_prog $i $i.out $EXPANDED_DATABASE + done + else run_one_prog $1 $2 $EXPANDED_DATABASE fi diff --git a/etc/run_fastx b/etc/run_fastx index 833c89690..e37d4b4cd 100644 --- a/etc/run_fastx +++ b/etc/run_fastx @@ -6,7 +6,7 @@ # to customise this script see the function called run_one_prog below -RCS_HEADER="$Header: //tmp/pathsoft/artemis/etc/run_fastx,v 1.2 2004-10-05 15:43:08 tjc Exp $" +RCS_HEADER="$Header: //tmp/pathsoft/artemis/etc/run_fastx,v 1.3 2005-02-21 11:54:10 tjc Exp $" PROG=`echo $RCS_HEADER | sed 's/.*run_\(.*\),v.*/\1/'` @@ -49,26 +49,12 @@ run_one_prog () { 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 + echo "command line: $COMMAND" 1>&2 + + $COMMAND 2>&1 > $OUTPUT_FILE | + tee ${PROG}_errors.new 1>&2 #### end of changes -- GitLab