Query any NCBI database (nucleotide, protein, nucgss, etc.) and retrieve resulting records.

query_ncbi.pl

#!/usr/bin/perl
# Author      : Kamil Slowikowski <[email protected]>
# Date        : March 23, 2011
# Version     : 0.1
# Description : Modified code from
#               http://www.bioperl.org/wiki/HOWTO:EUtilities_Cookbook

use strict;
use warnings;

use Bio::DB::EUtilities;
use Getopt::Long::Descriptive;

my ($opt, $usage) = describe_options(
    '%c [options...]',
    [ 'output|o=s',
      "write returned data to this file, default is stdout",
      { default => 'stdout' }
    ],
    [ 'query|q=s',
      "query sent to NCBI",
      { default => '' }
    ],
    [ 'db|d=s',
      "NCBI database to query, default is nucgss",
      { default => 'nucgss' } # nucleotide
    ],
    [ 'listdatabases|l',
      "list all available databases and exit",
    ],
    [ 'email|e=s',
      "your email, so NCBI can can track your project "
      . "and contact you if there is a problem",
      { default => '[email protected]' }
    ],
    [ 'retstart|s=i',
      "number of results to skip, default is 0",
      { default => 0 }
    ],
    [ 'retmax|m=i',
      "number of results to download at a time, default is 500",
      { default => 500 }
    ],
    [ 'rettype|t=s',
      "format of returned data, default is fasta",
      { default => 'fasta' }
    ],
    [ 'retries|r=i',
      "number of retries before giving up if the server doesn't respond",
      { default => 5 }
    ],
    [],
    [ 'help|h', "print usage message and exit" ],
);

# print a list of all available databases and exit
if ($opt->listdatabases) {
    my $factory = Bio::DB::EUtilities->new(
        -eutil => 'einfo',
        -email => $opt->email,
    );
    print join(' ', sort $factory->get_available_databases), "\n";
    exit;
}

# print help if query is missing or help is requested
if ($opt->help || length $opt->query < 1) {
    print($usage->text);
    exit;
}

# create an ESearch object
my $factory = Bio::DB::EUtilities->new(
    -eutil      => 'esearch',
    -db         => $opt->db,
    -term       => $opt->query,
    -email      => $opt->email,
    -usehistory => 'y',
);

# die if the query fails to be translated
die unless my $translation = $factory->get_query_translation;

# check how many results we got back
my $count = $factory->get_count;

warn "Query translates to '$translation'\n";
warn "Found $count results\n";
warn "Download and print all data? [no] \n";
<STDIN> =~ /^ye?s?/i || exit;

# get history from queue
my $hist = $factory->next_History || die 'No history data returned';

# note db carries over from above
$factory->set_parameters(
    -eutil   => 'efetch',
    -rettype => $opt->rettype,
    -history => $hist,
);

# count how many times we retry 
my $retry = 0;
# how many results to download at a time, which result index to start with
my ($retmax, $retstart) = ($opt->retmax, $opt->retstart);

# either print to stdout or to the file specified
my $out;
if ($opt->output eq 'stdout') {
    $out = \*STDOUT;
} else {
    open($out, '>', $opt->output) || die "Can't open file: $!";
}

RETRIEVE_RESULTS:
while ($retstart < $count) {
    $factory->set_parameters(
        -retmax   => $retmax,
        -retstart => $retstart
    );
    eval{
        $factory->get_Response(
            -cb => sub {
                       my ($data) = @_;
                       print $out $data;
                   }
        );
    };
    if ($@) {
        if ($retry == $opt->retries) {
            # quit if we get to 5 retries
            die "Server error: $@. Try again later";
        } else {
            # otherwise, retry
            warn "Server error, retry #", ++$retry, "\n";
            redo RETRIEVE_RESULTS;
        }
    }
    warn "Retrieved $retstart-", ($retstart + $retmax), "\n";
    $retstart += $retmax;
}

close $out;