Added GPML hex id hash, nodes_by_gene_id hash, and new nodes array. Revised gene...

algorithm and successfully tested node-swapping DOM approach all the way to file.

svn path=/; revision=143

diff --git a/preecej/perl_singletons/pathway_gene_swapper.pl b/preecej/perl_singletons/pathway_gene_swapper.pl
index 8a490c05a974b3f6795cb9908eec02dd1647448a..6a9e03c4c442c0d0829ae734e29937967cba22c8 100644 (file)
--- a/preecej/perl_singletons/pathway_gene_swapper.pl
+++ b/preecej/perl_singletons/pathway_gene_swapper.pl
@@ -310,49 +310,134 @@ sub swap_genes
     my $biopax_node = ($pathway_node->getElementsByTagName("Biopax"))[0];
     $pathway_node->removeChild($biopax_node);
 
-    # for (@$biopax_nodes)
-    # {
-    #     # print $_->getTagName . "\n";
-    #     $pathway_node->removeChild($_);
-    # }
-
     my $data_nodes = $pathway_node->getElementsByTagName("DataNode");
+
     #print $data_nodes->getLength . "\n";
+    # if ($debug) { print "...<DEBUG: \@$data_nodes>\n"
+    #     . Dumper(@$data_nodes) . "\n\n"; }
+
     for (@$data_nodes)
     {
         my $curr_datanode = $_;
         my $biopaxref_nodes = $curr_datanode->getElementsByTagName("BiopaxRef");
         for (@$biopaxref_nodes)
         {
+            # print $_->getTagName . "\n";
             $curr_datanode->removeChild($_);
         }
     }
     
-    # my $biopaxref_nodes = $gpml_doc->getElementsByTagName("BiopaxRef");
-    # print $biopaxref_nodes->getLength . "\n";
-    # for (@$biopaxref_nodes)
-    # {
-    #     @$pathway_nodes[0]->removeChild($_);
-    # }
+    # will hold a convenient list of data node references in the gpml doc,
+    # indexed by the id of the gene located in the <Xref> element for each
+    # node
+    my %data_nodes_by_gene_id;
     
+    # create a hash of all 5-digit hex ids in the gpml doc (this is the black list)
+    # one list of DataNode.GraphId, Group.GroupId, and Group.GraphId
+    my %existing_hex_ids;
+    for (@$data_nodes)
+    {
+        #print $_ . "\n";
+        if ($_->getAttributeNode("GraphId"))
+        {
+            $existing_hex_ids{$_->getAttributeNode("GraphId")->getValue}
+                = $_->getTagName . ".GraphId";
+        }
+        # also build a data node hash to make lookup easier in the next section
+        my $curr_xref_id = ($_->getElementsByTagName("Xref"))[0]
+            ->getAttributeNode("ID")->getValue;
+        $curr_xref_id =~ s/\s+$//; # rtrim whitespace 
+        $data_nodes_by_gene_id{$curr_xref_id} = $_;
+    }
+
+    # if ($debug) { print "...<DEBUG: \%data_nodes_by_gene_id>\n"
+    #     . Dumper(\%data_nodes_by_gene_id) . "\n\n"; }
+
+    my $group_nodes = $pathway_node->getElementsByTagName("Group");
+    for (@$group_nodes)
+    {
+        if ($_->getAttributeNode("GroupId"))
+        {
+            $existing_hex_ids{$_->getAttributeNode("GroupId")->getValue}
+                = $_->getTagName . ".GroupId";
+        }
+        if ($_->getAttributeNode("GraphId"))
+        {
+            $existing_hex_ids{$_->getAttributeNode("GraphId")->getValue}
+                = $_->getTagName . ".GraphId";
+        }
+    }
+
+    # if ($debug) { print "...<DEBUG: \%existing_hex_ids>\n"
+    #     . Dumper(\%existing_hex_ids) . "\n\n"; }
+    
+    # iterate through gene mappings from csv file
+    foreach my $old_gene (keys %swap_genes)
+    {
+        #print $old_gene . "\n";
+
+        # find curr old gene node in doc
+        if (exists $data_nodes_by_gene_id{$old_gene})
+        {
+            my $curr_old_gene_node = $data_nodes_by_gene_id{$old_gene};
+            # print $curr_old_gene_node . "\n";
+
+            # holds list of newly-created nodes, used to replace old node
+            my @new_nodes;
+
+            # copy the curr old gene node
+            my $new_node = $curr_old_gene_node->cloneNode("deep");
+
+            # testing...
+            $new_node->setAttribute("GraphId","zzzzz");
+            #print $new_node->getAttributeNode("GraphId")->getValue . "\n";
+            print $new_node->toString;
+
+            # iterate through new gene replacements
+            # for (@{$swap_genes{$old_gene}})
+            # {
+            #     # add to new nodes ary
+                 push @new_nodes, $new_node;
+            # }
+            # print "@new_nodes\n";
+
+            # if more than one new gene maps, you'll need a group for multiple gene boxes
+            
+                # if curr old gene does not belong to a group
+                    # create a new group node
+                    # SUB: generate a new groupRef hex id not already in use
+                    # add to end of group nodes
+                # for each new gene box
+                    # use new or existing group id for new boxes
+                    # add to new nodes ary
+    
+            # for new nodes ary
+                # SUB: generate a new DataNode hex id not already in use
+                # update new nodes w/ attributes:
+                    # rename TextLabel (prefix: Eu-, suffix: -alpha or numeric, based on regex)
+                    # add Comment back-referencing TAIR locus id (use "source" attribute)
+                    # edit <Xref Database="JGI" ID="Egrandis..." />
+                    # decrement the Z-order
+                    # decrement CenterX and CenterY by 10px each
+                    # change box Width if needed
+                    # color the box (<Graphics...Color="4488ff" />)
+
+            # replace old node w/ new node(s); do this in place
+            for (@new_nodes) {
+                # replace one
+                $pathway_node->replaceChild($_,$curr_old_gene_node);
+                # ...and add the others immediately preceding
+                #insertBefore();
+            }
+        }
+        else
+        {
+            print "ALERT: Gene identifier $old_gene is not present in this "
+                . "PathVisio GPML document.\n";
+        }
+    }
     
-    # .identify mapped genes
-    #   .if mapped gene did not belong to a group, create new group node
-    #   .duplicate <DataNode> for each mapping
-    #     .create new 5-digit hex code "GraphId"
-    #   .if needed, add new group node
-    #     .create new, random 5-digit hex code
-    #       .make sure it doesn't already exist
-    #     .add "GroupRef" on each added node, including the orig.
-    #   .rename TextLabel (prefix: Eu-, suffix: -alpha or numeric, based on regex)
-    #   .add Comment back-referencing TAIR locus id (use "source" attribute)
-    #   .edit <Xref Database="JGI" ID="Egrandis..." />
-    #   .decrement the Z-order
-    #   .decrement CenterX and CenterY by 10px each
-    #   .change box Width if needed
-    #   .color the box (<Graphics...Color="4488ff" />)
-    # 
-    # [hold]
+    # [later]
     # .After PathVisio sanity check, before communication:
     #   .remove back-ref TAIR comments