| 1 |
gpertea |
23 |
#!/usr/bin/perl |
| 2 |
|
|
use strict; |
| 3 |
|
|
use FindBin; |
| 4 |
|
|
|
| 5 |
|
|
umask 0002; |
| 6 |
|
|
#the line below is needed if pvmsx is used |
| 7 |
|
|
# also, the error condition is set only by the presence of $file |
| 8 |
|
|
$ENV{'PATH'}=$FindBin::Bin.':'.$ENV{'PATH'}; |
| 9 |
|
|
|
| 10 |
|
|
#so for pvmsx to consider the task was successful, $file must be deleted! |
| 11 |
|
|
#============== |
| 12 |
|
|
# 1 is the name of the fasta sequence input file |
| 13 |
|
|
# 2 is the # of sequences in ${1} should = 1 for this script |
| 14 |
|
|
# 3 is the slice no. being processed by sx |
| 15 |
|
|
# 4 is 0 if not the last file, 1 if the last file |
| 16 |
|
|
# 5 is the # of sequences skipped initially |
| 17 |
|
|
# 6 is the # of sequences to be processed (-1 = ALL) |
| 18 |
|
|
# 7 user parameter |
| 19 |
|
|
# 1 2 3 4 5 6 |
| 20 |
|
|
my ($file, $numpass, $slice_num, $last, $skipped, $total, $serverport, $dbdir)=@ARGV; |
| 21 |
|
|
#print STDERR "running: $0 ".join(' ',@ARGV)."\n"; |
| 22 |
|
|
my $usage=q{ |
| 23 |
|
|
Slice processing script for blat protein-to-dnax mode; |
| 24 |
|
|
Never use without a parent controllers (e.g. gridx). |
| 25 |
|
|
Must start the server prior to running this gridx command. |
| 26 |
|
|
Usage example: |
| 27 |
|
|
gridx -q -N -O logs -g condor -p 20 -n 2000 -i proteins.fa blat_client.psx server:port /fullpath/to/dbseqdir |
| 28 |
|
|
}; |
| 29 |
|
|
|
| 30 |
|
|
die $usage."\n" unless -s $file && $total && $dbdir; |
| 31 |
|
|
my $fout=$file.'.gff3'; |
| 32 |
|
|
my ($server, $port)=split(/:/, $serverport); |
| 33 |
|
|
die $usage."\n" unless $port>255; |
| 34 |
|
|
die $usage."\nInvalid data directory $dbdir" unless -d $dbdir; |
| 35 |
|
|
my $log_file='log_std'; |
| 36 |
|
|
my $err_file='err_log'; |
| 37 |
|
|
open(STDERR, '>>'.$err_file); |
| 38 |
|
|
open(STDOUT, '>>'.$log_file); |
| 39 |
|
|
|
| 40 |
|
|
#my $toskip=($file =~ m/_\@(\d+)_v\d+\.\d+/) ? $1 : $skipped+$numpass*($slice_num-1); |
| 41 |
|
|
#my ($srvhost, $srvport)=split(/\:/,$srv); |
| 42 |
|
|
#$srvport=8080 unless $srvport; |
| 43 |
|
|
my $cmd="gfClient -minScore=50 -minIdentity=70 -t=dnax -q=prot -out=gff -nohead $server $port $dbdir $file $fout"; |
| 44 |
|
|
|
| 45 |
|
|
#my $cmd="gmap -D $dbdir -d $gmapdb -B 2 -f 1 $file > $gmap_res"; |
| 46 |
|
|
my $slno=sprintf("slice:%09d",$slice_num); |
| 47 |
|
|
print STDERR ">>$slno: $cmd\n"; |
| 48 |
|
|
&runCmd($cmd, $fout); |
| 49 |
|
|
|
| 50 |
|
|
print STDERR "<<$slno: done.\n"; |
| 51 |
|
|
unlink($file); |
| 52 |
|
|
exit 0; |
| 53 |
|
|
|
| 54 |
|
|
sub runCmd { |
| 55 |
|
|
my ($docmd, @todel) = @_; |
| 56 |
|
|
my $errmsg = `($docmd) 2>&1`; |
| 57 |
|
|
if ($? || ($errmsg =~ m/Error|Segmentation|Failed|Invalid/is)) { |
| 58 |
|
|
print STDERR "!Error at:\n$docmd\n"; |
| 59 |
|
|
print STDERR "$errmsg\n"; |
| 60 |
|
|
system("cp $file err.at.$file"); |
| 61 |
|
|
foreach (@todel) { |
| 62 |
|
|
unlink($_); |
| 63 |
|
|
} |
| 64 |
|
|
# exit(1); |
| 65 |
|
|
return 0; |
| 66 |
|
|
} |
| 67 |
|
|
} |
