add bin files for search routines

[function2gene.git] / bin / get_location_from_uniprot

diff --git a/bin/get_location_from_uniprot b/bin/get_location_from_uniprot
new file mode 100755 (executable)
index 0000000..f4582e0
--- /dev/null
+++ b/bin/get_location_from_uniprot
@@ -0,0 +1,143 @@
+#! /usr/bin/perl
+
+# get_location_from_uniprot retreives files of search results from ncbi,
+# and is released under the terms of the GPL version 2, or any later
+# version, at your option. See the file README and COPYING for more
+# information.
+
+# Copyright 2004 by Don Armstrong <don@donarmstrong.com>.
+
+# $Id: ss,v 1.1 2004/06/29 05:26:35 don Exp $
+
+
+use warnings;
+use strict;
+
+
+use Getopt::Long;
+use Pod::Usage;
+
+=head1 NAME
+
+  get_location_from_uniprot [options]
+
+=head1 SYNOPSIS
+
+
+ Options:
+  --terms, -t file of search terms [default -]
+  --debug, -d debugging level [default 0]
+  --help, -h display this help
+  --man, -m display manual
+
+=head1 OPTIONS
+
+=over
+
+=item B<--debug, -d>
+
+Debug verbosity. (Default 0)
+
+=item B<--help, -h>
+
+Display brief useage information.
+
+=item B<--man, -m>
+
+Display this manual.
+
+=back
+
+=head1 EXAMPLES
+
+  get_location_from_uniprot -t terms.txt > output.txt
+
+Will pretty much do what you want
+
+=cut
+
+# http://www.ebi.uniprot.org/uniprot-srv/extendedView.do?proteinId=1A01_HUMAN
+
+use vars qw($DEBUG $REVISION);
+
+BEGIN{
+     ($REVISION) = q$LastChangedRevision: 1$ =~ /LastChangedRevision:\s+([^\s]+)/;
+     $DEBUG = 0 unless defined $DEBUG;
+}
+
+use IO::File;
+use URI::ParamMunge;
+use LWP::UserAgent;
+
+# XXX parse config file
+
+my %options = (debug    => 0,
+              help     => 0,
+              man      => 0,
+              format   => 'xml',
+              database => 'gene',
+              dir      => '.',
+              name     => '${search}_results_harvester',
+              terms    => '-',
+              uniprot_site => 'http://www.ebi.uniprot.org',
+              uniprot_search_url  => '/uniprot-srv/extendedView.do?proteinId=1A01_HUMAN',
+             );
+
+GetOptions(\%options,'terms|t=s','dir|D=s','debug|d+','help|h|?','man|m');
+
+pod2usage() if $options{help};
+pod2usage({verbose=>2}) if $options{man};
+
+$DEBUG = $options{debug};
+
+use constant {NAME     => 0,
+             LOCATION => 1,
+             FULLNAME => 2,
+            };
+
+#open search terms file
+my $terms;
+if ($options{terms} eq '-') {
+     $terms = \*STDIN;
+}
+else {
+     $terms = new IO::File $options{terms}, 'r' or die "Unable to open file $options{terms}: $!";
+}
+
+my $ua = new LWP::UserAgent(agent=>"DA_get_location_from_uniprot/$REVISION");
+
+#For every term
+print STDOUT qq("NAME","LOCATION","FULL NAME"\n);
+while (<$terms>) {
+     my @gene;
+     # Get uids to retrieve
+     chomp;
+     my $search = $_;
+     my $url = uri_param_munge($options{uniprot_site}.$options{uniprot_search_url},
+                              {proteinId => $search,
+                              },
+                             );
+     my $request = HTTP::Request->new('GET', $url);
+     my $response = $ua->request($request);
+     $response = $response->content;
+     $gene[NAME] = $search;
+     ($gene[LOCATION]) = $response =~ m{<!--Chromosome\s+locus-->\s*<tr>\s*
+                                     <td\s+class="import_title"\s+valign="top">&nbsp;</td>\s*
+                                     <td\s+class="value"\s+colspan="5">\s*
+                                     <table\s+width="100%"><tr\s+class="value"><td>Gene\s+name:[^\&]+&nbsp;&nbsp;Location:([^\<]+)</td></tr>\s*
+                                     </table></td></tr>\s*
+                                     <!--\s*end\s+chromosome\s+locus\s+-->}xis;
+     ($gene[FULLNAME]) = $response =~ m{>Protein\s+name</a>\s*
+                                     </td>\s*<td\s+class="value"\s+colspan="5">\s*
+                                     <b>([^\<]+)</b>\s*
+                                     </td>\s*</tr>\s*<!--end\s+title-->}xis;
+     print STDOUT join(',', map {if (defined $_) {qq("$_");} else {qq("NO DATA");}} @gene[0..2]),qq(\n);
+     sleep 2;
+}
+
+
+
+
+
+
+__END__