Perl script that retrieves a list of all the fully sequenced species that are in the TreeFam database, and stores them in a Perl pickle

store_treefam_species.pl

#!/usr/local/bin/perl

# 
# Perl script store_treefam_species.pl
# Written by Avril Coghlan ([email protected])
# 3-Apr-09. 
#
# For the TreeFam project.
#
# This perl script connects to the TreeFam database and stores
# a list of fully sequenced species in a pickle.
#
# The command-line format is:
# % perl <store_treefam_species.pl> version
# where version is the version of the TreeFam database to use.
#
#------------------------------------------------------------------#

# CHECK IF THERE ARE THE CORRECT NUMBER OF COMMAND-LINE ARGUMENTS:

$num_args                  = $#ARGV + 1;
if ($num_args != 1)
{
   print "Usage of store_treefam_species.pl\n\n";
   print "perl store_treefam_species.pl <version>\n";
   print "where <version> is the version of the TreeFam database to use.\n";
   print "For example, >perl -w store_treefam_species.pl 7\n";
   exit;
}

# FIND THE RELEASE OF TREEFAM TO USE:

$version                   = $ARGV[0];

# READ IN MY PERL MODULES:
use Avril_modules;
use Treefam::DBConnection;
use DBI;
use Storable;

$VERBOSE                   = 0;

#------------------------------------------------------------------#

# GET A LIST OF ALL FULLY SEQUENCED TREEFAM SPECIES:

$database                  = 'treefam_'.$version; 
%SEQUENCED                 = ();
%TAXID                     = ();
$num                       = 0;
$dbh                       = DBI->connect("dbi:mysql:$database:db.treefam.org:3308", 'anonymous', '') || return;
$table_w                   = 'species';
$st                        = "SELECT TAXNAME, SWCODE, TAX_ID from $table_w WHERE FLAG=\"1\"";
$sth                       = $dbh->prepare($st) or die "Cannot prepare $st: $dbh->errstr\n";
$rv                        = $sth->execute or die "Cannot execute the query: $sth->errstr";
if ($rv >= 1)
{
   while ((@array) = $sth->fetchrow_array)
   {
      $num++;
      $TAXNAME             = $array[0];
      $swcode              = $array[1]; # eg. CAEEL
      $taxid               = $array[2];
      if ($VERBOSE == 1) { print "$num: species $TAXNAME ($swcode), taxid $taxid\n";}
      $SEQUENCED{$swcode}  = 1;
      $TAXID{$swcode}      = $taxid;
   }
}
$rc                        = $dbh->disconnect();
$rc                        = "";
print STDERR "Read in fully sequenced species...\n";

# STORE THE HASH TABLES %SEQUENCED AND %TAXID IN A PICKLE:
$output1                   = "treefam.".$version."_sequenced";
store \%SEQUENCED,$output1; 
$output2                   = "treefam.".$version."_taxid";
store \%TAXID,$output2;

#------------------------------------------------------------------#

print STDERR "FINISHED.\n";