From 1e9070660246b20b55997f628e69a2f0f4878e8f Mon Sep 17 00:00:00 2001
From: Don Armstrong <don@donarmstrong.com>
Date: Mon, 16 Dec 2013 18:28:50 -0800
Subject: [PATCH] add start of geo downloader

---
 geo_downloader | 124 +++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 124 insertions(+)
 create mode 100755 geo_downloader

diff --git a/geo_downloader b/geo_downloader
new file mode 100755
index 0000000..541d319
--- /dev/null
+++ b/geo_downloader
@@ -0,0 +1,124 @@
+#!/usr/bin/perl
+# geo_downloader downloads expression files from GEO (NCBI)
+# and is released under the terms of the GNU GPL version 3, or any
+# later version, at your option. See the file README and COPYING for
+# more information.
+# Copyright 2013 by Don Armstrong <don@donarmstrong.com>.
+
+
+use warnings;
+use strict;
+
+use Getopt::Long;
+use Pod::Usage;
+
+=head1 NAME
+
+geo_downloader - downloads expression files from GEO (NCBI)
+
+=head1 SYNOPSIS
+
+geo_downloader [options] [GSE...]
+
+ Options:
+   --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 usage information.
+
+=item B<--man, -m>
+
+Display this manual.
+
+=back
+
+=head1 EXAMPLES
+
+geo_downloader GSE20400
+
+=cut
+
+
+use vars qw($DEBUG);
+use Net::FTP;
+
+my %options = (debug           => 0,
+               help            => 0,
+               man             => 0,
+               host => 'ftp.ncbi.nlm.nih.gov',
+              );
+
+GetOptions(\%options,
+           'debug|d+','help|h|?','man|m');
+
+pod2usage() if $options{help};
+pod2usage({verbose=>2}) if $options{man};
+
+$DEBUG = $options{debug};
+
+my @USAGE_ERRORS;
+if (not @ARGV) {
+    push @USAGE_ERRORS,"You must give at least one GEO accession";
+}
+
+if (@ARGV != grep {/^(gpl|gse|gsm|gds)\d+$/i} @ARGV) {
+    push @USAGE_ERRORS,"Invalid GEO accession(s): ".
+        join(',',grep {$_ !~ /^(gpl|gse|gsm)\d+$/} @ARGV);
+}
+
+
+pod2usage(join("\n",@USAGE_ERRORS)) if @USAGE_ERRORS;
+
+my $ftp = Net::FTP->new($options{host},Debug=>$DEBUG,Passive=>1) or
+    die "Unable to create new Net::FTP object";
+print STDERR "Connected\n" if $DEBUG;
+
+$ftp->login('anonymous') or
+    die "Unable to login";
+print STDERR "logged in\n" if $DEBUG;
+
+$ftp->binary();
+print STDERR "binary\n" if $DEBUG;
+
+
+
+for my $geo_acc (@ARGV) {
+    my $geo_directory = geo_directory($geo_acc);
+    $ftp->cwd($geo_directory);
+    print STDERR "changed to $geo_directory\n" if $DEBUG;
+    print STDERR "(really) changed to ".$ftp->pwd()."\n" if $DEBUG;
+    my @files = $ftp->ls();
+    print map {$_."\n"} @files;
+    print STDERR "transferred listing\n" if $DEBUG;
+}
+
+sub geo_directory {
+    my $geo_acc = shift;
+    $geo_acc = uc($geo_acc);
+    my $geo_acc_dir = $geo_acc;
+    $geo_acc_dir =~ s/\d{3}$/nnn/;
+    my $geo_type_dir = undef;
+    if ($geo_acc =~ /^GSE/) {
+        $geo_type_dir = 'series';
+    } elsif ($geo_acc =~ /^GDS/) {
+        $geo_type_dir = 'datasets';
+    } elsif ($geo_acc =~ /^GPL/) {
+        $geo_type_dir = 'platforms';
+    } elsif ($geo_acc =~ /^GSM/) {
+        $geo_type_dir = 'samples';
+    }
+    return "/geo/".$geo_type_dir.'/'.$geo_acc_dir.'/'.$geo_acc;
+}
+
+__END__
-- 
2.39.2