bin/get_location_from_uniprot

   1 #! /usr/bin/perl
   2
   3 # get_location_from_uniprot retreives files of search results from ncbi,
   4 # and is released under the terms of the GPL version 2, or any later
   5 # version, at your option. See the file README and COPYING for more
   6 # information.
   7
   8 # Copyright 2004 by Don Armstrong <don@donarmstrong.com>.
   9
  10 # $Id: ss,v 1.1 2004/06/29 05:26:35 don Exp $
  11
  12
  13 use warnings;
  14 use strict;
  15
  16
  17 use Getopt::Long;
  18 use Pod::Usage;
  19
  20 =head1 NAME
  21
  22   get_location_from_uniprot [options]
  23
  24 =head1 SYNOPSIS
  25
  26
  27  Options:
  28   --terms, -t file of search terms [default -]
  29   --debug, -d debugging level [default 0]
  30   --help, -h display this help
  31   --man, -m display manual
  32
  33 =head1 OPTIONS
  34
  35 =over
  36
  37 =item B<--debug, -d>
  38
  39 Debug verbosity. (Default 0)
  40
  41 =item B<--help, -h>
  42
  43 Display brief useage information.
  44
  45 =item B<--man, -m>
  46
  47 Display this manual.
  48
  49 =back
  50
  51 =head1 EXAMPLES
  52
  53   get_location_from_uniprot -t terms.txt > output.txt
  54
  55 Will pretty much do what you want
  56
  57 =cut
  58
  59 # http://www.ebi.uniprot.org/uniprot-srv/extendedView.do?proteinId=1A01_HUMAN
  60
  61 use vars qw($DEBUG $REVISION);
  62
  63 BEGIN{
  64      ($REVISION) = q$LastChangedRevision: 1$ =~ /LastChangedRevision:\s+([^\s]+)/;
  65      $DEBUG = 0 unless defined $DEBUG;
  66 }
  67
  68 use IO::File;
  69 use URI::ParamMunge;
  70 use LWP::UserAgent;
  71
  72 # XXX parse config file
  73
  74 my %options = (debug    => 0,
  75                help     => 0,
  76                man      => 0,
  77                format   => 'xml',
  78                database => 'gene',
  79                dir      => '.',
  80                name     => '${search}_results_harvester',
  81                terms    => '-',
  82                uniprot_site => 'http://www.ebi.uniprot.org',
  83                uniprot_search_url  => '/uniprot-srv/extendedView.do?proteinId=1A01_HUMAN',
  84               );
  85
  86 GetOptions(\%options,'terms|t=s','dir|D=s','debug|d+','help|h|?','man|m');
  87
  88 pod2usage() if $options{help};
  89 pod2usage({verbose=>2}) if $options{man};
  90
  91 $DEBUG = $options{debug};
  92
  93 use constant {NAME     => 0,
  94               LOCATION => 1,
  95               FULLNAME => 2,
  96              };
  97
  98 #open search terms file
  99 my $terms;
 100 if ($options{terms} eq '-') {
 101      $terms = \*STDIN;
 102 }
 103 else {
 104      $terms = new IO::File $options{terms}, 'r' or die "Unable to open file $options{terms}: $!";
 105 }
 106
 107 my $ua = new LWP::UserAgent(agent=>"DA_get_location_from_uniprot/$REVISION");
 108
 109 #For every term
 110 print STDOUT qq("NAME","LOCATION","FULL NAME"\n);
 111 while (<$terms>) {
 112      my @gene;
 113      # Get uids to retrieve
 114      chomp;
 115      my $search = $_;
 116      my $url = uri_param_munge($options{uniprot_site}.$options{uniprot_search_url},
 117                                {proteinId => $search,
 118                                },
 119                               );
 120      my $request = HTTP::Request->new('GET', $url);
 121      my $response = $ua->request($request);
 122      $response = $response->content;
 123      $gene[NAME] = $search;
 124      ($gene[LOCATION]) = $response =~ m{<!--Chromosome\s+locus-->\s*<tr>\s*
 125                                       <td\s+class="import_title"\s+valign="top">&nbsp;</td>\s*
 126                                       <td\s+class="value"\s+colspan="5">\s*
 127                                       <table\s+width="100%"><tr\s+class="value"><td>Gene\s+name:[^\&]+&nbsp;&nbsp;Location:([^\<]+)</td></tr>\s*
 128                                       </table></td></tr>\s*
 129                                       <!--\s*end\s+chromosome\s+locus\s+-->}xis;
 130      ($gene[FULLNAME]) = $response =~ m{>Protein\s+name</a>\s*
 131                                       </td>\s*<td\s+class="value"\s+colspan="5">\s*
 132                                       <b>([^\<]+)</b>\s*
 133                                       </td>\s*</tr>\s*<!--end\s+title-->}xis;
 134      print STDOUT join(',', map {if (defined $_) {qq("$_");} else {qq("NO DATA");}} @gene[0..2]),qq(\n);
 135      sleep 2;
 136 }
 137
 138
 139
 140
 141
 142
 143 __END__