my @awords = split /_/, $a;
  my @bwords = split /_/, $b;

  my $anum = $awords[0];
  my $bnum = $bwords[0];


  #  if ($anum !~ /^chr/ || $bnum !~ /^chr/) {
  #    die "Chr name doesn't begin with chr for $a or $b";
  #  }

  $anum =~ s/chr//;
  $bnum =~ s/chr//;

  if ($anum !~ /^[0-9]*$/) {
    if ($bnum !~ /^[0-9]*$/) {
      return $anum cmp $bnum;
    } else {
      return 1;
    }
  }
  if ($bnum !~ /^[0-9]*$/) {
    return -1;
  }

  if ($anum <=> $bnum) {
    return $anum <=> $bnum;
  } else {
    if ($#awords == 0) {
      return -1;
    } elsif ($#bwords == 0) {
      return 1;
    } else {
      return $awords[1] cmp $bwords[1];
    }
  }

  my ($chromosomes, $chrhash, $dbname) = @_;

  #filter to specified chromosome names only
  if (scalar(@$chromosomes)) {
    foreach my $chr (@$chromosomes) {
      my $found = 0;
      foreach my $chr_from_hash (keys %$chrhash) {
        if ($chr_from_hash =~ /^${chr}$/) {
          $found = 1;
          last;
        }
      }
      if (!$found) {
        print "Didn't find chromosome named $chr in database " . $dbname . "\n";
      }
    }
    HASH: foreach my $chr_from_hash (keys %$chrhash) {
      foreach my $chr (@$chromosomes) {
        if ($chr_from_hash =~ /^${chr}$/) { 
          #print "Keeping $chr_from_hash\n"; 
          next HASH;
        }
      }
      #print "Removing $chr_from_hash\n";
      delete($chrhash->{$chr_from_hash});
    }
  }
  return 1;
}

return 1;

  my $db = shift;
  my $type = shift;

  if (!$db->isa('Bio::EnsEMBL::DBSQL::DBAdaptor')) {
    die "get_chrlengths should be passed a Bio::EnsEMBL::DBSQL::DBAdaptor\n";
  }

  my %chrhash;

  my $q = qq( SELECT chrom.name,max(chr_end) FROM assembly as ass, chromosome as chrom
              WHERE ass.type = '$type' and ass.chromosome_id = chrom.chromosome_id
              GROUP BY chrom.name
            );

  my $sth = $db->dbc->prepare($q) || $db->throw("can't prepare: $q");
  my $res = $sth->execute || $db->throw("can't execute: $q");

  while( my ($chr, $length) = $sth->fetchrow_array) {
    $chrhash{$chr} = $length;
  }
  return\% chrhash;

  my $db   = shift;
  my $type = shift;
  my $coordsystem = shift;

  my %chrhash;

  if ($coordsystem ne 'toplevel') {
    if (!$db->isa('Bio::EnsEMBL::DBSQL::DBAdaptor')) {
      die "get_chrlengths should be passed a Bio::EnsEMBL::DBSQL::DBAdaptor\n";
    }
  
    my $query = "select seq_region.name, seq_region.length as mce from seq_region,coord_system where" .
                " seq_region.coord_system_id=coord_system.coord_system_id and" .
                " coord_system.version = '" . $type . "' and coord_system.name='$coordsystem'";
  
    my $sth = $db->dbc->prepare($query);
  
    $sth->execute;
  
  
    my $hashref;
    while (($hashref = $sth->fetchrow_hashref) && defined($hashref)) {
      $chrhash{$hashref->{'name'}} = $hashref->{mce};
      # print $hashref->{'name'} . " " . $hashref->{'mce'} . "\n";
    }
  } else {
    my $sa = $db->get_SliceAdaptor;

    my @slices = @{$sa->fetch_all('toplevel')};

    foreach my $slice (@slices) {
      #print $slice->seq_region_name . " " . $slice->length . " " . $slice->coord_system->version . "\n";
      if ($slice->coord_system->version eq $type) {
        $chrhash{$slice->seq_region_name} = $slice->length;
      } else {
        print STDERR "WARNING: " . $slice->seq_region_name . " " . $slice->length . " " . $slice->coord_system->version . " NOT on requested assembly version but is toplevel\n";
      }
    }
  }
  return\% chrhash;

  my ($chrhash) = @_;

  my @sorted_names = reverse sort bychrnum keys %$chrhash;
  return\@ sorted_names;