|
18
|
1 #!/usr/bin/perl -w
|
|
|
2 ###
|
|
|
3 # But : protocol permettant la detection d'RNA non codant potentiel
|
|
|
4 #
|
|
|
5 # Entrees : fichier de mapping Smart gff3
|
|
|
6 # fichier gff des gènes
|
|
|
7 # fichier gff des clusters Cis regulateur potentiel
|
|
|
8 #
|
|
|
9 # Sortie : fichier gff des clusters ARN nc
|
|
|
10 #
|
|
|
11 ###------------------------------------------------------
|
|
|
12
|
|
|
13 use vars qw($USAGE);
|
|
|
14 use strict;
|
|
|
15
|
|
|
16 =head1 NAME
|
|
|
17
|
|
|
18 protocol_NC_V2_CTN3.pl
|
|
|
19
|
|
|
20 =head1 SYNOPSIS
|
|
|
21
|
|
|
22 % strictlyIncludeGff.pl -i toSelect.gff3 -t template.gff3 > result.gff3
|
|
|
23
|
|
|
24 =head1 DESCRIPTION
|
|
|
25
|
|
|
26 strictlyIncludeGff.pl - print elements strictly include in template (gff files)
|
|
|
27
|
|
|
28 -i|--input fileName gff input file name
|
|
|
29 -t|--template fileName gff template file name
|
|
|
30 [-h|--help] help mode then die
|
|
|
31
|
|
|
32 =head1 AUTHOR - CTN - mar.11
|
|
|
33 (from RNA-Vibrio/protocol_NC_V2_CTN3.pl - Claire KUCHLY)
|
|
|
34
|
|
|
35 =cut
|
|
|
36
|
|
|
37 #----------------------------------------------------------------------------
|
|
|
38 # check command line :
|
|
|
39 my $outFileName = "outSIG.gff3";
|
|
|
40 if ($#ARGV==0) {
|
|
|
41 die (exec("pod2text $0\n"));
|
|
|
42 } else {
|
|
|
43 foreach my $num (0 .. $#ARGV) {
|
|
|
44 SWITCH: for ($ARGV[$num]) {
|
|
|
45 /--input|-i/ && do { open(ARN,"<$ARGV[$num+1]")
|
|
|
46 or die "Error: Can't open \"$ARGV[$num+1]\", $!";
|
|
|
47 last };
|
|
|
48 /--template|-t/ && do { open(SEED,"<$ARGV[$num+1]")
|
|
|
49 or die "Error : Can't open file \"$ARGV[$num+1]\", $!";
|
|
|
50 last };
|
|
|
51 /--help|-h/ && do { exec("pod2text $0\n") ; die };
|
|
|
52 }
|
|
|
53 }
|
|
|
54 ##NC_011753.2 RefSeq gene 367 834 . - . locus_tag=VS_0001;db_xref=GeneID:7162789
|
|
|
55 # open(OUT,">$outFileName") or die "Error can't $outFileName open for output. $!\n";
|
|
|
56 my @seed ;
|
|
|
57 my $s=0;
|
|
|
58 while (my $seedLine = <SEED> ) {
|
|
|
59 my @list = split(/\t/,$seedLine);
|
|
|
60 $seed[$s][0]= $list[3] ; # position begin seed
|
|
|
61 $seed[$s][1]= $list[4] ; # position end seed
|
|
|
62 $seed[$s][2]= $list[6] ; # seed sens
|
|
|
63 $seed[$s][3]= $list[0] ; # chromesome name
|
|
|
64 $s++;
|
|
|
65 }
|
|
|
66 close SEED ;
|
|
|
67 while(my $ligne = <ARN>){
|
|
|
68 $s=0;
|
|
|
69 my @list = split(/\t/,$ligne);
|
|
|
70 while (($s <= $#seed)) {
|
|
|
71 if (($seed[$s][3] eq $list[0]) and ($seed[$s][0] <= $list[3]) and ($seed[$s][1] >= $list[4]) and ($seed[$s][2] eq $list[6])) { # if list include in seed + same direction
|
|
|
72 print "$ligne";
|
|
|
73 }
|
|
|
74 $s++;
|
|
|
75 }
|
|
|
76 }
|
|
|
77 close ARN ;
|
|
|
78 exit(0);
|
|
|
79 }
|
