|
9
|
1 #Declaration de l espace de travail et du chemin de l executable R :
|
|
|
2 use strict;
|
|
|
3 use Carp;
|
|
|
4 use Statistics::R;
|
|
|
5 use Cwd qw(abs_path);
|
|
|
6 use File::Basename;
|
|
|
7 use Getopt::Long;
|
|
|
8
|
|
|
9
|
|
|
10 #Variables Globales ###########################
|
|
|
11 defined $ENV{'MY_GALAXY_DIR'} || die "MY_GALAXY_DIR environment variable not defined";
|
|
|
12 use lib "$ENV{'MY_GALAXY_DIR'}";
|
|
|
13 use GalaxyPath;
|
|
|
14
|
|
|
15 my ($r_bin,$log_dir,$R,$cmd,$line, $out, $input, $reference);
|
|
|
16
|
|
|
17
|
|
|
18 #test:
|
|
|
19 #perl sm_Tax4Fun.pl --input DATA/testtax4fun---ssu---fingerprint----Total---sim_93---tax_silva---td_20.csv --out DATA/Tax4FunProfile.txt
|
|
|
20
|
|
|
21
|
|
|
22 #Parametres d'entree
|
|
|
23 Getopt::Long::Configure( 'no_ignorecase', 'bundling' );
|
|
|
24 GetOptions (
|
|
|
25 'out=s' => \$out,
|
|
|
26 'input=s' => \$input,
|
|
|
27 'reference=s' => \$reference
|
|
|
28 ) or die "Usage: Error in command line arguments\n";
|
|
|
29
|
|
|
30 # SETTING GLOBAL VARIABLES ####################################################
|
|
|
31 my $cfg = GalaxyPath->new( -file => $ENV{"GALAXY_CONFIG_FILE"} );
|
|
|
32 my $file_path = $cfg->my_path( 'workPath', 'MYWORKSPACE' );
|
|
|
33 $r_bin = $cfg->my_path( 'toolsPath', 'R_BIN_PATH' );
|
|
|
34 my $SCRIPTS_R_path = $cfg->my_path( 'toolsPath', 'SCRIPTS_R_path' );
|
|
|
35 ###############################################################################
|
|
|
36
|
|
|
37 #workspace
|
|
|
38 my $debug = 0; #Mode debug (0)
|
|
|
39 if ($debug == 0)
|
|
|
40 {
|
|
|
41 print STDOUT "Debug mode OK \n";
|
|
|
42 }
|
|
|
43 else
|
|
|
44 {
|
|
|
45 $file_path = dirname($out);
|
|
|
46 print STDOUT "No debug \n";
|
|
|
47 }
|
|
|
48
|
|
|
49
|
|
|
50
|
|
|
51 my ($nb) = ($out=~/dataset_(\d+)\.\S+$/);
|
|
|
52 #For debugging in command line on vm-galaxy-preprod
|
|
|
53 #my ($nb) = (1);
|
|
|
54 #creation du repertoire de sortie
|
|
|
55 `cd $file_path; mkdir $nb; chmod -R 777 $nb `;
|
|
|
56 my $dirresults= $file_path."/".$nb;
|
|
|
57
|
|
|
58
|
|
|
59 print STDOUT "Job working directory : $dirresults \n";
|
|
|
60
|
|
|
61
|
|
|
62
|
|
|
63 #remplacer les [a-z]__ du fichier entrant $input par rien
|
|
|
64 #$str =~ s/$find/$replace/g;
|
|
|
65 `cp -a $input $dirresults/INPUT.csv`;
|
|
|
66 open(FILE, "<$dirresults/INPUT.csv") || die "File not found";
|
|
|
67 my @lines = <FILE>;
|
|
|
68 close(FILE);
|
|
|
69
|
|
|
70 my @newlines;
|
|
|
71 my %Taxo=();
|
|
|
72 foreach(@lines) {
|
|
|
73 $_ =~ s/[a-z]\_\_//g; #homogeneiser la taxonomie issues de frogs avec celle utilisée par tax4fun
|
|
|
74 $_ =~ s/\n/;\n/g; #pour qu'il y ait un point virgule a la fin de chaque ligne
|
|
|
75 $_ =~ s/;;/;/g; #remplacer les ;; par un seul ;
|
|
|
76 $_ =~ s/;;/;/g; #remplacer les ;; par un seul ; une seconde fois pour les triples points virgules
|
|
|
77 #push(@newlines,$_);
|
|
|
78 /^(\d+)\t(.+)/||die "bad line: $_";
|
|
|
79 $Taxo{$2}+=$1;
|
|
|
80 }
|
|
|
81
|
|
|
82 open(FILE, ">$dirresults/INPUT.csv") || die "File not found";
|
|
|
83 #print FILE "abondance_sum\tTaxonomy\n";
|
|
|
84 print FILE "abondance_sum;\n";
|
|
|
85 #print FILE @newlines;
|
|
|
86 foreach my $taxo (sort {$a cmp $b} keys %Taxo) {
|
|
|
87 printf FILE "%d\t%s\n",$Taxo{$taxo},$taxo;
|
|
|
88 }
|
|
|
89 close(FILE);
|
|
|
90
|
|
|
91
|
|
|
92 #Log dir
|
|
|
93 $log_dir = $dirresults;
|
|
|
94
|
|
|
95 #déaration de l'objet $R et ouverture du pont :
|
|
|
96 $R = Statistics::R->new(
|
|
|
97 "r_bin" => $r_bin,
|
|
|
98 "log_dir" => $log_dir,
|
|
|
99 ) or die "Problem with R : $!\n";
|
|
|
100 # Ouverture du pont
|
|
|
101 $R->startR;
|
|
|
102
|
|
|
103
|
|
|
104 print STDOUT "Ouverture du pont R \n";
|
|
|
105
|
|
|
106 #Read R script
|
|
|
107 $cmd="";
|
|
|
108 #[galaxy-preprod@vm-galaxy-preprod sm_Tax4Fun]$ ln -s /galaxydata/galaxy-preprod/my_tools/sm_Tax4Fun/sm_Tax4Fun.R /galaxydata/galaxy-preprod/my_bin/scripts_R/.
|
|
|
109 open IN,"< $SCRIPTS_R_path/sm_Tax4Fun.R"
|
|
|
110 or die "Unable to read R script : $!\n";
|
|
|
111 while ($line=<IN>) {
|
|
|
112 $cmd.=$line;
|
|
|
113 }
|
|
|
114 close(IN);
|
|
|
115
|
|
|
116 #Declare R function
|
|
|
117 print STDOUT "Declaration du script R \n";
|
|
|
118
|
|
|
119 #$dirresults/INPUT.csv #"input.file = '$input', ".
|
|
|
120 $R->send(
|
|
|
121 "setwd('$dirresults')\n".
|
|
|
122 "$cmd\n".
|
|
|
123 "GalaxyFrogsTax( ".
|
|
|
124 "input.file = '$dirresults/INPUT.csv', ".
|
|
|
125 "reference = '$reference', ".
|
|
|
126 "output.file = '$out')\n");
|
|
|
127
|
|
|
128 print STDOUT "Envoi du script R ";
|
|
|
129
|
|
|
130
|
|
|
131
|
|
|
132
|
|
|
133 #Fermeture du pont
|
|
|
134 $R->stopR();
|
|
|
135
|
|
|
136 print STDOUT "Fermeture du pont R \n";
|
|
|
137
|
|
|
138 #Recuperation des outputs dans Galaxy
|
|
|
139 my $cmdOUT ='';
|
|
|
140 my $outTax= "$dirresults/Tax4FunProfile.txt";
|
|
|
141 if (! -e $out){print STDERR "Pas de fichier Tax4FunProfile.txt produit \n";}
|
|
|
142 else {$cmdOUT = "(mv $outTax $out) >& ./cp_taxout.log 2>&1";
|
|
|
143 system $cmdOUT;}
|
