Perl script that checks for cases where a TreeFam family seems to have disappeared from a particular version of TreeFam, even though it was present in the previous version of TreeFam and has not been curated since.

treefam_QC12.pl

#!/usr/local/bin/perl

# 
# Perl script treefam_QC12.pl
# Written by Avril Coghlan ([email protected])
# 6-Feb-09.
#
# This perl script checks for cases where a TreeFam family seems to have
# disappeared from a particular version of TreeFam, even though it was
# present in the previous version of TreeFam and has not been curated since.
#
# The command-line format is:
# % perl <treefam_QC12.pl> <release> 
#   where <release> is the release 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 treefam_QC12.pl\n\n";
   print "perl treefam_QC12.pl <release>\n";
   print "where <release> is the release of the TreeFam database to use.\n";
   print "For example, >perl -w treefam_QC12.pl 7\n";
   exit;
}

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

# DECLARE MYSQL USERNAME AND HOST:
use DBI;

# FIND WHICH RELEASE OF THE TREEFAM DATABASE TO USE:

$release                   = $ARGV[0];

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

# FIND ALL FAMILIES IN THE FAMILYA/FAMILYB/FAMILYC TABLES, IN
# THE CURRENT TREEFAM RELEASE:

$database                  = "dbi:mysql:treefam_".$release.":db.treefam.org:3308";
$dbh                       = DBI->connect("$database", 'anonymous', '') || return;
$RELEASE                   = &read_all_families1($dbh);
# READ IN THE FAMILYB FAMILIES WHICH WERE TURNED INTO FAMILYA FAMILIES
# SINCE THE PREVIOUS RELEASE:
$CURATED                   = &read_curated_families($dbh);
# NOW DISCONNECT FROM THE DATABASE:
$rc                        = $dbh->disconnect();
$rc                        = "";

# FIND ALL FAMILIES IN THE FAMILYA/FMAILYB/FAMILYC TABLES, IN
# THE PREVIOUS TREEFAM RELEASE:

$prev_release              = $release - 1;
$database                  = "dbi:mysql:treefam_".$prev_release.":db.treefam.org:3308";
$dbh                       = DBI->connect("$database", 'anonymous', '') || return;
&read_all_families2($dbh,$RELEASE,$CURATED);
# NOW DISCONNECT FROM THE DATABASE:
$rc                        = $dbh->disconnect();
$rc                        = "";

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

print STDERR "FINISHED.\n";
print "FINISHED\n";


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

# READ IN THE FAMILYB FAMILIES WHICH WERE TURNED INTO FAMILYA FAMILIES
# SINCE THE PREVIOUS RELEASE:

sub read_curated_families
{
   my $dbh                 = $_[0];
   my $table_w;
   my $st;
   my $sth;
   my $rv;
   my @array;
   my $AC1;
   my $AC2;
   my %CURATED             = ();

   $table_w                = "famB2A";
   $st                     = "SELECT famB, famA from $table_w";
   $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)
      {
         $AC1              = $array[0]; # eg. TF300003
         $AC2              = $array[1]; # eg. TF101534
         $CURATED{$AC1}    = $AC2;
      }
   }
   return(\%CURATED);
}

#------------------------------------------------------------------#
# READ ALL FAMILIES IN THE CURRENT RELEASE OF TREEFAM:

sub read_all_families1
{
   my $dbh                 = $_[0];
   my $table_w;
   my $st;
   my $sth;
   my $rv;
   my @array;
   my $i;
   my $AC;
   my %SEEN                = ();

   for ($i = 1; $i <= 3; $i++)
   {
      if    ($i == 1) { $table_w = 'familyA';}
      elsif ($i == 2) { $table_w = 'familyB';}
      elsif ($i == 3) { $table_w = 'familyC';}

      $st                  = "SELECT AC from $table_w"; 
      $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)    
         {
            $AC            = $array[0];  
            $SEEN{$AC}     = 1;
         }
      }
   }

   print "Read all families in the TreeFam version $release\n";
   print STDERR "Read all families in the TreeFam version $release\n";
   return(\%SEEN);
}

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

# READ ALL FAMILIES IN THE CURRENT RELEASE OF TREEFAM:

sub read_all_families2
{
   my $dbh                 = $_[0];
   my $RELEASE             = $_[1];
   my $CURATED             = $_[2];
   my $table_w;
   my $st;
   my $sth;
   my $rv;
   my @array;
   my $i;
   my $AC;
   my $curated;

   for ($i = 1; $i <= 3; $i++)
   {
      if    ($i == 1) { $table_w = 'familyA';}
      elsif ($i == 2) { $table_w = 'familyB';}
      elsif ($i == 3) { $table_w = 'familyC';}

      $st                  = "SELECT AC from $table_w";
      $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)
         {
            $AC            = $array[0];
            # CHECK IF THIS FAMILY IS IN THE CURRENT RELEASE: 
            if (!($RELEASE->{$AC}))
            {
               # CHECK IF THE FAMILY $AC WAS CURATED:
               if ($CURATED->{$AC})
               {
                  $curated = $CURATED->{$AC};
                  if (!($RELEASE->{$curated}))
                  {
                     print "WARNING: family $AC was curated to $curated which is not in TreeFam release $release, even though $AC was in version $prev_release\n";
                  }
               }
               else
               {
                  print "WARNING: family $AC is not in TreeFam version $release, but was in version $prev_release (and was not curated since)\n";
               }
            }
         }
      }
   }
   print STDERR "Read all families in TreeFam version $prev_release\n";
   print "Read all families in TreeFam version $prev_release\n";
}

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