lib/CXGN/Tools/RestrictionEnzyme.pm

   1 package CXGN::Tools::RestrictionEnzyme;
   2 use strict;
   3 use CXGN::DB::Connection;
   4
   5 =head1 NAME
   6
   7  CXGN::Tools::RestrictionEnzyme
   8
   9 =head1 SYNOPSIS
  10
  11  Abstract representation for pulling information about restriction enzymes
  12  from SGN database.  Will work in concert with CXGN::Tools::Sequence
  13  to find restriction sites.
  14
  15 =head1 AUTHOR
  16
  17  C. Carpita <ccarpita@gmail.com>
  18
  19 =cut
  20
  21 #GLOBAL CLASS VARIABLES
  22 our $DBH;
  23
  24 #Prepared statement references:
  25 our ($ALL_ENZ_NAME_Q);
  26 our ($VALID_ENZ_NAME_Q, $ENZ_SEQ_Q);
  27
  28 our %VALID_NAMES;
  29
  30 =head1 CLASS METHODS
  31
  32 =cut
  33
  34 sub setDBH {
  35         my $class = shift;
  36         my $dbh = shift;
  37         die "Must call this function on the class and send a DBH argument\n" unless ref($dbh);
  38         $DBH = $dbh;
  39         $class->prepare();
  40 }
  41
  42 sub prepare {
  43         my $class = shift;
  44         if($DBH->can("add_search_path")){
  45                 $DBH->add_search_path("sgn");
  46         }
  47         else {
  48                 $DBH->do("set search_path=sgn");
  49         }
  50
  51         $ALL_ENZ_NAME_Q = $DBH->prepare("
  52                         SELECT DISTINCT enzyme_name
  53                         FROM enzymes
  54                         WHERE enzyme_id
  55                         IN (
  56                                 SELECT DISTINCT enzyme_id
  57                                 FROM enzyme_restriction_sites
  58                                 WHERE restriction_site IS NOT NULL
  59                         )
  60                         ORDER BY enzyme_name
  61                         ");
  62
  63         $VALID_ENZ_NAME_Q = $DBH->prepare("
  64                         SELECT enzyme_name
  65                         FROM enzymes
  66                         WHERE enzyme_id
  67                                 IN (
  68                                         SELECT DISTINCT enzyme_id
  69                                         FROM enzyme_restriction_sites
  70                                         WHERE restriction_site IS NOT NULL
  71                                 )
  72                         AND enzyme_name = ?
  73                         ORDER BY enzyme_name
  74                         ");
  75
  76         $ENZ_SEQ_Q = $DBH->prepare("
  77                 SELECT restriction_site
  78                 FROM enzyme_restriction_sites
  79                 WHERE enzyme_id =
  80                         (       SELECT enzyme_id
  81                                 FROM enzymes
  82                                 WHERE enzyme_name = ?
  83                         )
  84                 AND restriction_site IS NOT NULL
  85         ");
  86 }
  87
  88 sub preload_valid_names {
  89         my $class = shift;
  90         $ALL_ENZ_NAME_Q->execute();
  91         while(my $row = $ALL_ENZ_NAME_Q->fetchrow_hashref){
  92                 my $enzyme_name = $row->{enzyme_name};
  93                 $VALID_NAMES{$enzyme_name} = 1;
  94         }
  95 }
  96
  97 sub createDBH {
  98         my $class = shift;
  99         print STDERR "$class is creating SGN database handle...\n";
 100         my $dbh = CXGN::DB::Connection->new("sgn");
 101         $class->setDBH($dbh);
 102 }
 103
 104 =head2 all_enzymes
 105
 106  Ex: my @enzymes = CXGN::Tools::Enzyme::all_enzymes();
 107
 108  Returns an array of all enzyme objects, created from
 109  the enzyme tables in the database.
 110
 111 =cut
 112
 113 sub all_enzymes {
 114         my $class = shift;
 115         $class->createDBH() unless ref($DBH);
 116         $class->preload_valid_names() unless keys(%VALID_NAMES);
 117         my @enzymes = ();
 118         foreach my $name (keys %VALID_NAMES){
 119                 my $enzyme = $class->new($name);
 120                 push(@enzymes, $enzyme);
 121         }
 122         return @enzymes;
 123 }
 124
 125
 126 =head1 CONSTRUCTOR
 127
 128  Usage: my $enzyme = CXGN::Tools::Enzyme->new("hindIII");
 129  Auto-fetches the matching sequences from the database,
 130  dies if the name is not valid.  Use eval{} tags to avoid
 131  death.
 132
 133 =cut
 134
 135 sub new {
 136         my $class = shift;
 137         my $self = bless {}, $class;
 138
 139         my $name = shift;
 140
 141         #You can provide a different DBH to use instead of the
 142         #class global, which we use to prepare queries efficiently
 143         my $dbh = shift;
 144
 145         die "No enzyme name provided\n" unless $name;
 146         $self->name($name);
 147
 148         if(ref($dbh)){
 149                 $self->setDBH($dbh);
 150         }
 151         else {
 152                 $class->createDBH() unless ref($DBH);
 153         }
 154
 155         $self->{match_seqs} = [];
 156         $self->check_name_valid();
 157         $self->fetch_match_seqs();
 158
 159         return $self;
 160 }
 161
 162 =head1 INSTANCE METHODS
 163
 164 =cut
 165
 166 sub name {
 167         my $self = shift;
 168         my $name = shift;
 169         return $self->{name} unless $name;
 170         $self->{name} = $name;
 171 }
 172
 173 sub check_name_valid {
 174         my $self = shift;
 175         my $name = $self->name();
 176         __PACKAGE__->preload_valid_names() unless defined %VALID_NAMES;
 177         die "Name of enzyme ($name) not valid, or no restriction site identified\n" unless $VALID_NAMES{$name};
 178 }
 179
 180 sub fetch_match_seqs {
 181         my $self = shift;
 182         $ENZ_SEQ_Q->execute($self->name());
 183         while(my ($seq) = $ENZ_SEQ_Q->fetchrow_array){
 184                 push(@{$self->{match_seqs}}, $seq);
 185         }
 186 }
 187
 188 1;