cgi-bin/search/family.pl

   1 #!/usr/bin/perl -w
   2 use strict;
   3 use warnings;
   4 use CXGN::Page;
   5 use CXGN::Page::FormattingHelpers qw/  page_title_html
   6                                                                            info_section_html
   7                                        blue_section_html  /;
   8 use CXGN::DB::Connection;
   9 use CXGN::Phylo::Alignment;
  10 use CXGN::Tools::Identifiers;
  11 use CXGN::Tools::WebImageCache;
  12 use CatalystX::GlobalContext '$c';
  13 use File::Temp;
  14
  15 our $page = CXGN::Page->new( "SGN Gene Family", "Chenwei Lin");
  16 our ($sum_q, $at_family_member_q, $sgn_family_member_q, $organism_group_q, $other_family_q, $sgn_align_seq_q, $at_align_seq_q);
  17
  18
  19 my $dbh = CXGN::DB::Connection->new("sgn");
  20
  21 $sum_q = $dbh->prepare("
  22         SELECT
  23                 family_build.family_build_id,
  24                 family_build.status,
  25                 count(family_member_id),
  26                 build_date,
  27                 build_nr,
  28                 i_value,
  29                 family_annotation,
  30                 family_build.group_id,
  31                 comment,
  32                 family.family_nr
  33         FROM sgn.family_member
  34         LEFT JOIN sgn.family USING (family_id)
  35         LEFT JOIN sgn.family_build USING (family_build_id)
  36         LEFT JOIN sgn.groups USING (group_id)
  37         WHERE family.family_id = ?
  38         GROUP BY family.family_id, family.family_nr, family_build.family_build_id, family_build.status, build_date, build_nr, i_value, family_annotation, group_id, comment");
  39
  40 $at_family_member_q = $dbh->prepare("
  41         SELECT sequence_name
  42         FROM family_member
  43         WHERE family_id = ?
  44         AND database_name LIKE 'Arabidopsis'");
  45
  46 $sgn_family_member_q = $dbh->prepare("
  47         SELECT unigene_id, family_member.cds_id, organism_group_id
  48         FROM sgn.family_member
  49         LEFT JOIN sgn.cds USING (cds_id)
  50         WHERE family_id = ?
  51         AND (   database_name LIKE 'SGN'
  52                         OR database_name LIKE 'CGN'
  53                 )
  54         ");
  55
  56 $organism_group_q = $dbh->prepare("
  57         SELECT comment
  58         FROM sgn.groups
  59         WHERE group_id = ?");
  60
  61 $other_family_q = $dbh->prepare("
  62         SELECT family_member.family_id, build_nr, i_value, member_count, family_nr
  63         FROM sgn.family_member
  64         LEFT JOIN sgn.family USING (family_id)
  65         LEFT JOIN sgn.family_build USING (family_build_id)
  66         WHERE cds_id = ?
  67         AND family_member.family_id != ?
  68         AND group_id = ?
  69         AND status='C'
  70         ");
  71
  72 my $family_size_align_limit = 200;
  73
  74 my ($family_id, $family_nr, $i_value) = $page->get_arguments(qw/ family_id family_nr i_value/ );
  75 if ($family_id eq ""){
  76         empty_search($page) unless ($family_nr && $i_value);
  77         my $family_id_q = $dbh->prepare("
  78                 SELECT family_id
  79                 FROM sgn.family
  80                 LEFT JOIN sgn.family_build USING(family_build_id)
  81                 WHERE
  82                 family_nr=?
  83                 AND family_build.i_value=?");
  84         $family_id_q->execute($family_nr, $i_value);
  85         ($family_id) = $family_id_q->fetchrow_array();
  86         empty_search($page) unless $family_id;
  87 }
  88
  89 my ($sum_content, $member_content, $link_content, $align_content);
  90
  91
  92 ##################################
  93 #Get information for the summary section
  94
  95 my ($annotation_content, $family_nr_content, $date_content, $i_value_content, $bn_content, $group_content, $total_gene_content);
  96 my ($family_build_id, $family_build_status, $total_gene, $date, $build_nr, $annotation, $group_id, $group_comment);
  97 $sum_q->execute($family_id);
  98 if (($family_build_id, $family_build_status, $total_gene, $date, $build_nr, $i_value, $annotation, $group_id, $group_comment, $family_nr) = $sum_q->fetchrow_array()){
  99   my $date_display = substr ($date, 0,10);
 100   $total_gene_content = "<tr><th>Total Genes</th><td>" . $total_gene . "</td></tr>";
 101   $date_content = "<tr><th>Build Date</th><td>" . $date_display . "</td></tr>";
 102   $family_nr_content = "<tr><th>Family&nbsp;Number</th><td>" . $family_nr . "</td></tr>";
 103   if ($i_value < 2){
 104     $i_value_content =  $i_value . "  (Low stringency in grouping genes together)";
 105   }
 106   elsif ($i_value >=2 && $i_value < 5){
 107     $i_value_content = $i_value . "  (Normal stringency in grouping genes together)";
 108   }
 109   elsif ($i_value >= 5){
 110     $i_value_content =  $i_value . "  (High stringency in grouping genes together)";
 111   }
 112
 113   $i_value_content = "<tr><th>i Value</th><td> ". $i_value_content . "</td></tr>";
 114   $bn_content = "<tr><th>Build</th><td>" . $build_nr . " <a href=\"family_build.pl?family_build_id=$family_build_id\">[Details of the Overall Family Build]</a>" . "</td></tr>";
 115   $annotation_content = "<tr><th valign=\"top\">Annotation</th><td>" . $annotation . "</td></tr>";
 116   $group_content = "<tr><th>Data Set</th><td>" . $group_comment . "</td></tr>";
 117 }
 118 my $deprecated_content = '';
 119 if($family_build_status ne 'C'){
 120         $deprecated_content .= "<center><div class=\"deprecated\" style=\"width:550px\">This family is from an out-of-date build.  <br />";
 121         $deprecated_content .= "<span style=\"color:black;font-size:0.9em\">See below for relations to current builds</span></div></center>";
 122 }
 123 $sum_content = $family_nr_content . $total_gene_content . $date_content .$i_value_content . $bn_content . $annotation_content . $group_content;
 124 $sum_content .= "<tr><td colspan=\"2\" align=\"center\"><a href=\"/about/family_analysis.pl\" target=\"blank\">
 125 Explanation of family analysis and terms used on this page
 126 </a></td></tr>";
 127 ####################################################
 128 #Get family member information.
 129 my $at_member_nr;
 130 my @at_member_url = ();
 131 my @at_member = ();
 132 my %sgn_member = ();
 133 my @all_sgn_member = ();
 134 my %group_comment = ();
 135 my %group_member_nr = ();
 136 my $family_member_content;
 137
 138 $at_family_member_q->execute($family_id);
 139 while( my ($member) = $at_family_member_q->fetchrow_array()){
 140   my $locus = substr ($member, 0,9);
 141   my $member_url = CXGN::Tools::Identifiers::link_identifier($locus);
 142 #  my $member_url  = "<a href=\"http://www.arabidopsis.org/servlets/TairObject?type=locus&amp;name=$locus\" target=\"blank\">" . $member . "</a>";
 143   push @at_member_url, $member_url;
 144   push @at_member, $member;
 145 }
 146 $at_member_nr = int (@at_member);
 147
 148 $sgn_family_member_q->execute($family_id);
 149 while( my ($member, $cds_id, $organism_group_id) = $sgn_family_member_q->fetchrow_array()){
 150   push @{$sgn_member{$organism_group_id}}, $member;
 151   push @all_sgn_member, $cds_id;
 152 }
 153
 154 foreach (keys %sgn_member){
 155   $organism_group_q->execute($_);
 156   if (my ($comment) = $organism_group_q->fetchrow_array()){
 157     $group_comment{$_} = $comment;
 158     $group_member_nr{$_} = int @{$sgn_member{$_}};
 159   }
 160 }
 161
 162 my %sgn_member_content = ();
 163 foreach (sort keys %sgn_member){
 164   my $id = $_;
 165   foreach (@{$sgn_member{$id}}){
 166         $sgn_member_content{$id} .= CXGN::Tools::Identifiers::link_identifier("SGN-U" . $_) . " ";
 167         #"<a href=\"unigene.pl?unigene_id=$_\">SGN-U" . $_ . "</a>  ";
 168   }
 169 }
 170
 171 $family_member_content = "<tr><th>Organism</th><th># of Members</th><th>Member Id</th></tr>";
 172 $family_member_content .= "<tr><td>Arabidopsis</td>" . "<td>$at_member_nr</td>" . "<td>@at_member_url</td></tr>";
 173 foreach (sort keys %sgn_member){
 174   $family_member_content .=  "<tr><td>$group_comment{$_}</td>" . "<td>$group_member_nr{$_}</td>" . "<td>$sgn_member_content{$_}</td></tr>";
 175 }
 176
 177 ######################################################
 178 #Link to other families
 179 my %other_family = ();
 180 my %other_i = ();
 181 my %other_size = ();
 182 my %other_num = ();
 183 my $other_family_content;
 184 my $number;
 185
 186 #use the @all_sgn_member and $group from previous sections
 187 my %family_regist = ();
 188 foreach (@all_sgn_member){
 189   $other_family_q->execute($_, $family_id, $group_id);
 190   while (my ($other_family_id, $other_build, $other_i_value, $other_size, $other_family_num) = $other_family_q->fetchrow_array()){
 191     $other_i{$other_build} = $other_i_value;
 192         $other_size{$other_family_id} = $other_size;
 193         $other_num{$other_family_id} = $other_family_num;
 194     if (!defined $family_regist{$other_family_id}){
 195       push @{$other_family{$other_build}}, $other_family_id;
 196       $family_regist{$other_family_id} = 1;
 197     }
 198   }
 199 }
 200 my %other_family_content_family = ();
 201 foreach (sort keys %other_family) {
 202   my $id = $_;
 203   foreach (@{$other_family{$id}}){
 204         my $content = $other_num{$_};
 205         $content .= "(" . $other_size{$_} . ")";
 206     $other_family_content_family{$id} .= qq|
 207                 <a href="family.pl?family_id=$_">$content</a>&nbsp;
 208                 |;
 209   }
 210 }
 211
 212
 213 $other_family_content .= "<tr><th>Build id</th><th>i Value</th><th>Family Number (Size)</th></tr>";
 214
 215 foreach (sort keys %other_family){
 216   $other_family_content .= "<tr><td>$_</td>" . "<td>$other_i{$_}</td>" . "<td>$other_family_content_family{$_}</td></tr>";
 217 }
 218
 219
 220 ######################################################
 221 #Retrieve alignment sequences and draw an alignment image
 222
 223 ##First check family size
 224 #
 225 my $pep_align_file = "/data/prod/public/family/$i_value/pep_align/$family_nr.pep.aligned.fasta";
 226 my $newick_file = "/data/prod/public/family/$i_value/newick/$family_nr.newick";
 227
 228 my $total_member_nr = int (@all_sgn_member) + $at_member_nr;
 229 if ($total_member_nr == 1 || $family_build_status ne 'C') {
 230   $align_content = "<tr><td>Not applicable.</td></tr>";
 231 }
 232 elsif( ! -f $pep_align_file ){
 233         my $extra = " due to its large size";
 234         $extra = "" unless ($total_member_nr > 100); #actual limit is 200, usually
 235         $align_content = "<tr><td>Alignment not available for this family$extra.</td></tr>";
 236 }
 237 elsif ($total_member_nr > $family_size_align_limit) {
 238   $align_content = "<tr><td>Family size too large for alignment on this page.";
 239   $align_content .= qq|&nbsp;&nbsp;<a href="/tools/align_viewer/show_align.pl?family_id=$family_id">See Alignment in Viewer</a> |;
 240   $align_content .= "</td></tr>";
 241 }
 242 else {
 243
 244         my $img_height = $total_member_nr * 20;
 245         my $img_width = 700;
 246
 247         my $cache = CXGN::Tools::WebImageCache->new();
 248         $cache->set_key($i_value . '_' . $family_nr);
 249         $cache->set_expiration_time(1);
 250         $cache->set_map_name("family_alignment_tree");
 251         $cache->set_basedir($c->config->{"basepath"});
 252         $cache->set_temp_dir("/documents/tempfiles/family");
 253         if(!$cache->is_valid()){
 254                 my $treealign = undef;
 255                 my $alignment = undef;
 256
 257                 my $alignment_only = 1;
 258                 $alignment_only = 0 if (-f $newick_file);
 259
 260                 unless($alignment_only){
 261                         use CXGN::Phylo::Tree;
 262                         $treealign = CXGN::Phylo::Tree->new({
 263                                 from_files => {
 264                                         newick => $newick_file,
 265                                         alignment => $pep_align_file
 266                                 }
 267                         });
 268                         $treealign->get_layout()->set_image_width($img_width);
 269                         $treealign->get_layout()->set_image_height($img_height);
 270                         $alignment = $treealign->get_alignment();
 271                 }
 272                 else {
 273                         $alignment = CXGN::Phylo::Alignment->new(
 274                                  width=>$img_width,
 275                                  height=>$img_height,
 276                                  type=>'pep',
 277                                  from_file=>$pep_align_file
 278                         );
 279                 }
 280
 281
 282                   ##Draw family_alignment image and write map file
 283                 my $show_num = $alignment->get_nonhidden_member_nr();
 284                 if ($show_num > 1) {
 285                                 my $image_mode = "s";
 286                                 $image_mode = "c" if ($show_num < 60 && !-f $newick_file);
 287                                 $alignment->set_display_type($image_mode);
 288
 289                                 my $tool_link = undef;
 290                                 unless($alignment_only){
 291                                         $cache->set_image_data($treealign->render_png());
 292                                         $tool_link = 'tree_browser/index.pl?&align_type=pep';
 293                                 }
 294                                 else {
 295                                         $alignment->render();
 296                                         $cache->set_image_data($alignment->{image}->png());
 297                                         $tool_link = 'align_viewer/show_align.pl?';
 298                                 }
 299
 300                             $align_content = "<tr><td><center><a target=\"new_tab\" href=\"../tools/${tool_link}&family_nr=$family_nr&i_value=$i_value\">";
 301                                 $align_content .= $cache->get_image_html();
 302                                 $align_content .= "</a></center></td></tr>";
 303                             $align_content .= "<tr><td align=\"center\">Click on the image above to view the detailed alignment</td></tr>";
 304                 }
 305                 else {
 306                         $align_content = "No alignment image available.";
 307                 }
 308         }
 309         else {
 310                         $align_content = "<tr><td><center><a target=\"new_tab\" href=\"../tools/align_viewer/show_align.pl?family_nr=$family_nr&i_value=$i_value\">";
 311                         $align_content .= $cache->get_image_html();
 312                         $align_content .= "</a></center></td></tr>";
 313                         #<img src=\"$tmp_image\" alt=\"\" border=\"0\"/></a></center></td></tr>";
 314                     $align_content .= "<tr><td align=\"center\">Click on the image above to view the detailed alignment</td></tr>";
 315
 316         }
 317 }
 318
 319 ######################################################
 320 #Page output
 321 $page->header();
 322 print page_title_html("SGN Gene Family $family_id");
 323 print $deprecated_content if $deprecated_content;
 324 print info_section_html(title => 'Summary', contents => '<table width="100%" cellpadding="5" cellspacing="0" border="0">'.$sum_content.'</table>');
 325 print info_section_html(title => 'Relations to Other Builds', contents => '<table width="100%" cellpadding="5" cellspacing="0" border="0">'.$other_family_content.'</table>');
 326 print info_section_html(title => 'Family Sequence Alignment', contents => '<table width="100%" cellpadding="5" cellspacing="0" border="0">'.$align_content.'</table>');
 327 print info_section_html(title => 'Family Members', contents => '<table width="100%" cellpadding="5" cellspacing="0" border="1">'.$family_member_content.'</table>', collapsible => 1);
 328
 329 $page->footer();
 330
 331 sub empty_search {
 332   my ($page, $family_id) = @_;
 333
 334   $page->header();
 335
 336   print <<EOF;
 337
 338   <b>No family id specified</b>
 339
 340 EOF
 341
 342   $page->footer();
 343   exit 0;
 344 }
 345
 346 sub invalid_search {
 347   my ($page, $family_id) = @_;
 348
 349   $page->header();
 350
 351   print <<EOF;
 352
 353   <b>The specified family identifer ($family_id) does not result in a valid search.</b>
 354
 355 EOF
 356
 357   $page->footer();
 358   exit 0;
 359 }
 360
 361