cgi-bin/search/family_build.pl

   1
   2 #!/usr/bin/perl -w
   3 use strict;
   4 use warnings;
   5 use CXGN::Page;
   6 use CXGN::Page::FormattingHelpers qw/  page_title_html
   7                                        blue_section_html  /;
   8 use GD::Graph::bars;
   9 use GD::Text;
  10 use GD;
  11 use File::Temp;
  12 use CXGN::DB::Connection;
  13 use CatalystX::GlobalContext '$c';
  14
  15 my $page = CXGN::Page->new( "SGN Gene Family Build", "Chenwei Lin");
  16 my ($family_build_sum_q, $total_family_q, $total_gene_q, $other_build_q, $organism_member_q, $family_organism_q, $family_size_q);
  17 my ($family_build_id, $family_build_nr) = $page->get_arguments("family_build_id", "family_build_nr");
  18
  19
  20 my $dbh = CXGN::DB::Connection->new("public");
  21
  22 if($family_build_nr && !$family_build_id) {
  23         my $fam_id_q = $dbh->prepare("SELECT family_build_id FROM family_build WHERE build_nr=?");
  24         $fam_id_q->execute($family_build_nr);
  25         ($family_build_id) = $fam_id_q->fetchrow_array();
  26 }
  27
  28 empty_search($page) unless $family_build_id;
  29
  30 $family_build_sum_q = $dbh->prepare("select family_build.group_id, i_value, build_date, comment, build_nr from family_build left join sgn.groups using (group_id) where family_build_id = ?");
  31
  32 $total_family_q = $dbh->prepare("select count(family_id) from family where family_build_id = ?");
  33 $total_gene_q = $dbh->prepare("select count(family_member_id) from family left join family_member using (family_id) where family_build_id = ?");
  34
  35 $other_build_q = $dbh->prepare("
  36         SELECT family_build_id, build_nr, i_value
  37         FROM family_build
  38         WHERE group_id = ?
  39         AND build_nr != ?
  40         AND status='C'
  41         ");
  42
  43 $family_organism_q = $dbh->prepare("select family_member.family_id,organism_group_id, comment from sgn.groups left join family_member on (family_member.organism_group_id = sgn.groups.group_id) left join family using (family_id) left join family_build using (family_build_id) where family_build.family_build_id = ?");
  44
  45 $family_size_q = $dbh->prepare("select count(family_member_id) from family_member left join family using (family_id) where family_build_id = ? group by family_member.family_id");
  46
  47
  48 my ($sum_content, $member_content, $other_build_content, $dist_content);
  49 ###############################################
  50 #Family build summary
  51 $family_build_sum_q->execute($family_build_id);
  52 my ($group_id, $i_value, $build_date, $group_comment, $build_nr, $total_family, $total_gene);
  53 my ($i_value_content, $date_content, $group_comment_content, $build_nr_content, $total_family_content, $total_gene_content);
  54
  55 if (($group_id, $i_value, $build_date, $group_comment, $build_nr, $total_family) = $family_build_sum_q->fetchrow_array()){
  56   my $date_content = substr($build_date, 0, 10);
  57   if ($i_value < 2){
  58     $i_value_content =  $i_value . "  (Low stringency in grouping genes together)";
  59   }
  60   elsif ($i_value >= 2 && $i_value < 5){
  61     $i_value_content = $i_value . "  (Normal stringency in grouping genes together)";
  62   }
  63   elsif ($i_value >= 5){
  64     $i_value_content =  $i_value . "  (High stringency in grouping genes together)";
  65   }
  66   else { $i_value_content = $i_value }
  67
  68   $total_family_q->execute($family_build_id);
  69   ($total_family) = $total_family_q->fetchrow_array();
  70   $total_gene_q->execute($family_build_id);
  71   ($total_gene) = $total_gene_q->fetchrow_array();
  72   $date_content = "<tr><th>Build Date</th><td>" . $date_content . "</td></tr>";
  73   $i_value_content = "<tr><th>i Value</th><td> ". $i_value_content . "</td></tr>";
  74   $group_comment_content = "<tr><th>Data Set</th><td>" . $group_comment . "</td></tr>";
  75   $build_nr_content = "<tr><th>Build number</th><td>" . $build_nr . "</td></tr>";
  76   $total_family_content = "<tr><th>Total Families</th><td>" . $total_family . "</td></tr>";
  77   $total_gene_content = "<tr><th>Total Genes/Unigenes</th><td>" . $total_gene . "</td></tr>";
  78   $sum_content = $date_content . $i_value_content . $group_comment_content . $build_nr_content . $total_family_content . $total_gene_content;
  79   $sum_content .= "<tr><td colspan=\"2\" align=\"center\" bgcolor=\"gray\"><a href=\"/about/family_analysis.pl\" target=\"blank\">For help with gene family analysis, please click here.</a></td></tr>";
  80 }
  81 else {
  82   &invalid_search;
  83 }
  84
  85 ###############################################
  86 #Member datasets
  87 my %organism_gene_count = ();
  88 my %organism_family = ();
  89 my %family_organism = ();
  90 my %organism_uniq_family_count = ();
  91 my %organism_comment = ();
  92 my %organism_family_count = ();
  93
  94 $family_organism_q->execute($family_build_id);
  95 while (my ($family_id, $organism_group_id, $organism_group_comment) = $family_organism_q->fetchrow_array()){
  96   $organism_comment{$organism_group_id} = $organism_group_comment;
  97   if (!defined $organism_gene_count{$organism_group_id}){
  98     $organism_gene_count{$organism_group_id} = 1;
  99   }
 100   else {
 101     $organism_gene_count{$organism_group_id}++;
 102   }
 103   $organism_family{$organism_group_id}{$family_id} = 1;
 104   $family_organism{$family_id}{$organism_group_id} = 1;
 105 }
 106
 107 foreach (keys %organism_family){
 108   $organism_family_count{$_} = 0;
 109   my $count = int (keys %{$organism_family{$_}});
 110   $organism_family_count{$_} += $count;
 111 }
 112
 113 foreach (keys %family_organism){
 114   my $count = int (keys %{$family_organism{$_}});
 115   my $family_id = $_;
 116   if ($count == 1){
 117     foreach (keys %{$family_organism{$family_id}}){
 118       if (!defined $organism_uniq_family_count{$_}){
 119         $organism_uniq_family_count{$_} = 1;
 120       }
 121       else {
 122         $organism_uniq_family_count{$_}++;
 123       }
 124     }
 125   }
 126 }
 127
 128 my $member_data_content = "<tr><th>Species</th><th># Genes or Unigenes</th><th># Families</th><th># Unique Families</th></tr>";
 129 foreach (sort {$organism_gene_count{$b} <=> $organism_gene_count{$a}} keys %organism_gene_count){
 130   $member_data_content .= "<tr><td>$organism_comment{$_}</td><td>$organism_gene_count{$_}</td><td>$organism_family_count{$_}</td><td>$organism_uniq_family_count{$_}</td></tr>";
 131 }
 132
 133
 134 ###############################################
 135 #Other builds of the same group
 136 my %other_build = ();
 137 my %other_build_id = ();
 138
 139 $other_build_q->execute($group_id, $build_nr);
 140 while (my ($other_build_id, $other_build_nr, $other_i_value) = $other_build_q->fetchrow_array())
 141 {
 142
 143   if ($other_i_value < 2){
 144     $other_i_value .= " (Low stringency in grouping genes together)";
 145   }
 146   elsif($other_i_value >= 2 && $other_i_value < 5){
 147     $other_i_value .= " (Normal stringency in grouping genes together)";
 148   }
 149   else {
 150     $other_i_value .= " (High stringency in grouping genes together)";
 151   }
 152   $other_build{$other_build_nr} = $other_i_value;
 153   $other_build_id{$other_build_nr} = $other_build_id;
 154
 155 }
 156
 157 $other_build_content = "<tr><th>Build number</th><th>i Value</th></tr>";
 158 foreach (sort keys %other_build){
 159   $other_build_content .= "<tr><td><a href=\"family_build.pl?family_build_id=$other_build_id{$_}\">$_</a></td><td>$other_build{$_}</td></tr>";
 160 }
 161
 162 ###############################################
 163 #Size distribution
 164 my %size_gene_count = ();
 165 my %size_family_count  = ();
 166 $family_size_q->execute($family_build_id);
 167 while (my ($count) = $family_size_q->fetchrow_array()){
 168   if ($count == 1) {
 169     $size_gene_count{"1"}++;
 170   }
 171   elsif ($count == 2){
 172     $size_gene_count{"2"} += $count;
 173   }
 174   elsif ($count == 3){
 175     $size_gene_count{"3"} += $count;
 176   }
 177   elsif ($count >3 && $count <=20){
 178     $size_gene_count{"4-20"} += $count;
 179   }
 180   elsif ($count >20 && $count <=40){
 181     $size_gene_count{"21-40"} += $count;
 182   }
 183   elsif ($count >40 && $count <=100){
 184     $size_gene_count{"41-100"} += $count;
 185   }
 186   else {
 187     $size_gene_count{"100 up"} += $count;
 188   }
 189 }
 190
 191 ###############################################
 192 #Draw bar chart
 193 #First generte a random file.
 194 my $html_root_path = $c->config->{'basepath'};
 195 my $doc_path = $c->tempfiles_subdir('family_images');
 196 my $path = $html_root_path . $doc_path;
 197 my $tmp = new File::Temp(
 198                         DIR => $path,
 199                         SUFFIX => '.png',
 200                         UNLINK => 0,
 201                     );
 202
 203 #Draw the bar chart
 204 my $graph = new GD::Graph::bars(700, 400);
 205
 206 $graph->set(
 207                    x_label => 'Family Size',
 208                    y_label => '%',
 209                    dclrs => [ qw(lblue) ],
 210                    bar_spacing => 10,
 211                    x_labels_vertical => 1,
 212                    x_label_position => 0.5,
 213                    text_space => 16,
 214                    two_axes => 1,
 215                    bar_width => 5,
 216
 217 );
 218
 219 $graph->set_x_label_font(gdGiantFont);
 220 $graph->set_y_label_font(gdGiantFont);
 221 $graph->set_x_axis_font(gdMediumBoldFont);
 222 $graph->set_y_axis_font(gdMediumBoldFont);
 223 $graph->set_legend_font(gdGiantFont);
 224 my @x_values = ();
 225 my @y_values = ();
 226
 227 foreach (sort {$a<=>$b} keys %size_gene_count){
 228   push @x_values, $_;
 229   push @y_values, $size_gene_count{$_} / $total_gene * 100;
 230 }
 231
 232 my $gd = $graph->plot([\@x_values, \@y_values]);
 233 print $tmp $gd->png;
 234 $tmp =~ s/$html_root_path//;
 235 close $tmp;
 236
 237 my $size_content = "<tr><td><center><img src=\"$tmp\" alt=\"\" /></center></td></tr>";
 238
 239 ###############################################
 240 #Page printout
 241 $page->header();
 242 print page_title_html("SGN Gene Family Build $family_build_id");
 243 print blue_section_html('Summary','<table width="100%" cellpadding="5" cellspacing="0" border="0">' . $sum_content . '</table>');
 244 print blue_section_html('Member Data Sets','<table width="100%" cellpadding="5" cellspacing="0" border="1">' . $member_data_content . '</table>');
 245 print blue_section_html('Other Builds with Different Inflation Factor','<table width="100%" cellpadding="5" cellspacing="0" border="0">' . $other_build_content . '</table>');
 246 print blue_section_html('Unigene Family Size Distribution','<table width="100%" cellpadding="5" cellspacing="0" border="0" align="center">' . $size_content . '</table>');
 247
 248
 249   $page->footer();
 250 sub empty_search {
 251   my ($page, $family_id) = @_;
 252
 253   $page->header();
 254
 255   print <<EOF;
 256
 257   <b>No family id specified</b>
 258
 259 EOF
 260
 261   $page->footer();
 262   exit 0;
 263 }
 264
 265 sub invalid_search {
 266   my ($page, $family_build_id) = @_;
 267
 268   $page->header();
 269
 270   print <<EOF;
 271
 272   <b>The specified family identifer ($family_build_id) does not result in a valid search.</b>
 273
 274 EOF
 275
 276   $page->footer();
 277   exit 0;
 278 }
 279