| 1 |
#!/usr/bin/perl |
| 2 |
use strict; |
| 3 |
use Getopt::Std; |
| 4 |
use File::Basename; |
| 5 |
use Cwd qw(abs_path cwd); |
| 6 |
umask 0002; |
| 7 |
|
| 8 |
my $output; |
| 9 |
my $myfasta; |
| 10 |
my $evidence_file; |
| 11 |
my $treedir; |
| 12 |
my $idir_list; |
| 13 |
my $addPrefix; |
| 14 |
my $useLinComb; |
| 15 |
my $noOverWrite; |
| 16 |
my $penStr; |
| 17 |
my $inLen; |
| 18 |
|
| 19 |
my $usage = q~ |
| 20 |
Wrapper script for running jigsaw (not the training part). |
| 21 |
|
| 22 |
run_jigsaw.pl [options] |
| 23 |
|
| 24 |
Options: |
| 25 |
-e <evidence_fie> - evidence descriptor file (required) |
| 26 |
-l <dirlist_file> - file with full paths to the genomic sequence and data |
| 27 |
directories, one per line (required) |
| 28 |
-o <output_file> - name of output file for gene predictions (required) |
| 29 |
-f <fasta_suffix> - the filename suffix of the genomic sequence file |
| 30 |
(default 'fa') |
| 31 |
-d <training_dir> - directory with training data (required, unless -L) |
| 32 |
-n <intron_penalty> - specify maximum intron length and penalty for |
| 33 |
exceeding the length |
| 34 |
-g <grid_engine> - grid engine to use: 'smp', 'sge' or 'condor'; |
| 35 |
(default is: run locally, do not use the grid) |
| 36 |
-c <numCPUs> - max grid nodes/CPUs to use (default 20; requires -g) |
| 37 |
-m <e-mail> - e-mail to notify when all jobs are finished |
| 38 |
-i <min_intron_len> - specify minimum intron length |
| 39 |
-C - do not clobber existing output file (default is to |
| 40 |
overwrite) |
| 41 |
-L - run the linear combiner instead |
| 42 |
|
| 43 |
Miscellaneous Options: |
| 44 |
-h, -help - print this help message |
| 45 |
-V - obtain program version |
| 46 |
~; |
| 47 |
|
| 48 |
my $cmdline="$0 ".join(' ',@ARGV); |
| 49 |
getopts('LCm:c:g:f:e:d:l:p:n:o:') || die($usage."\n"); |
| 50 |
|
| 51 |
($evidence_file, $output, $treedir, $idir_list, $myfasta) = |
| 52 |
($Getopt::Std::opt_e, $Getopt::Std::opt_o, $Getopt::Std::opt_d, $Getopt::Std::opt_l, |
| 53 |
$Getopt::Std::opt_f); |
| 54 |
$myfasta='fa' unless $myfasta; |
| 55 |
$useLinComb=$Getopt::Std::opt_L; |
| 56 |
$penStr=$Getopt::Std::opt_n; |
| 57 |
$inLen=$Getopt::Std::opt_i; |
| 58 |
$noOverWrite=$Getopt::Std::opt_C; |
| 59 |
my $mailnotify=$Getopt::Std::opt_m; |
| 60 |
my $gridengine=$Getopt::Std::opt_g; |
| 61 |
my $maxCPUs=$Getopt::Std::opt_c || 20; |
| 62 |
|
| 63 |
|
| 64 |
$addPrefix = 1; |
| 65 |
|
| 66 |
# if ( defined $addPrefix ) { |
| 67 |
# $addPrefix = 0; |
| 68 |
# } else { |
| 69 |
# $addPrefix = 1; |
| 70 |
# } |
| 71 |
|
| 72 |
die("$usage\n Error: not all the required parameters where given!\n") |
| 73 |
unless $output && -f $idir_list && -f $evidence_file; |
| 74 |
|
| 75 |
die("$usage\nError: training directory not given!\n") if (!-d $treedir && !$useLinComb); |
| 76 |
|
| 77 |
print STDERR "#Command line:\n$cmdline\n"; |
| 78 |
|
| 79 |
my $penArg=$penStr ? "-n \"$penStr\"" : ''; |
| 80 |
|
| 81 |
my $linArg = $useLinComb ? '-l' : ''; |
| 82 |
|
| 83 |
my $ilArg=""; |
| 84 |
|
| 85 |
#if ( defined $penStr ) { |
| 86 |
#$ilArg="-i $inLen"; |
| 87 |
#} |
| 88 |
|
| 89 |
my $evidence; |
| 90 |
open(FILE,$evidence_file) || die("Cannot open $evidence_file: $!"); |
| 91 |
while(my $line = <FILE>) { |
| 92 |
chomp($line); |
| 93 |
## ignore the curation line, not used when actually running combiner |
| 94 |
$evidence.="$line:" unless ($line =~ /\scuration\s*/ ); |
| 95 |
} |
| 96 |
close(FILE); |
| 97 |
|
| 98 |
my @testOn; |
| 99 |
open(FILE,$idir_list) || die "unable to open [$idir_list]\n"; |
| 100 |
while(my $line = <FILE>) { |
| 101 |
chomp($line); |
| 102 |
push(@testOn,$line) if $line; |
| 103 |
} |
| 104 |
close(FILE); |
| 105 |
my $combid = basename($evidence_file); |
| 106 |
|
| 107 |
my $CMDQFile='.runJIGSAW.'.$combid.'.'.$$.'.gridcmds'; |
| 108 |
|
| 109 |
my @cmdqueue; |
| 110 |
my $treearg = $treedir ? '-d '.abs_path($treedir) : ''; |
| 111 |
for(my $cnt = 0; $cnt < @testOn; $cnt++) { |
| 112 |
my $dir = $testOn[$cnt]; |
| 113 |
my $dname = basename($dir); |
| 114 |
my $prefix; |
| 115 |
if( $addPrefix ) { |
| 116 |
$prefix = "$dir/$dname."; |
| 117 |
} else { |
| 118 |
$prefix = "$dir/"; |
| 119 |
} |
| 120 |
my $combd = "${prefix}${combid}.run"; |
| 121 |
writeEvidenceFile($combd, $evidence, $prefix); |
| 122 |
my $combout = "${prefix}$output"; |
| 123 |
#my $cmd=; |
| 124 |
push(@cmdqueue, "jigsaw -f ${prefix}$myfasta $treearg $linArg -e $combd -m $combout $penArg"); |
| 125 |
|
| 126 |
#print "[$cmd]\n"; |
| 127 |
#if (!defined($noOverWrite) || ! -e $combout || -z $combout ) { |
| 128 |
# runCmd($cmd); |
| 129 |
# } |
| 130 |
} |
| 131 |
|
| 132 |
close(RUNQ); |
| 133 |
$mailnotify = $mailnotify ? "-m '$mailnotify'": ''; |
| 134 |
if ($gridengine) { |
| 135 |
open(RUNQ, '>'.$CMDQFile) || die("Error creating $CMDQFile!\n"); |
| 136 |
foreach my $l (@cmdqueue) { |
| 137 |
print RUNQ $l."\n"; |
| 138 |
} |
| 139 |
close(RUNQ); |
| 140 |
&runCmd("gridx -g $gridengine -p $maxCPUs -S -f $CMDQFile $mailnotify -O logs_jigs_run -q"); |
| 141 |
} |
| 142 |
else { # run locally: |
| 143 |
foreach my $cmd (@cmdqueue) { |
| 144 |
&runCmd($cmd); |
| 145 |
} |
| 146 |
} |
| 147 |
|
| 148 |
exit(0); |
| 149 |
|
| 150 |
sub writeEvidenceFile { |
| 151 |
my ($tfname,$evlst,$prefix) = @_; |
| 152 |
my @evlst = split(/:/,$evlst); |
| 153 |
my $evdata = $tfname.'.run'; |
| 154 |
my $status = 0; |
| 155 |
open(EVDATA, '>'.$tfname) || die ("Cannot create evidence file: $tfname\n"); |
| 156 |
for(my $iter = 0; $iter <= $#evlst; $iter++) { |
| 157 |
print EVDATA "$prefix$evlst[$iter]\n"; |
| 158 |
} |
| 159 |
close(EVDATA); |
| 160 |
} |
| 161 |
|
| 162 |
sub runCmd { |
| 163 |
my $cmd = $_[0]; |
| 164 |
print STDERR "#>running: $cmd\n"; |
| 165 |
system($cmd); |
| 166 |
} |
