From c9993a35c89e7498a980b56595d232df528d6f8e Mon Sep 17 00:00:00 2001
From: preecej <preecej@localhost>
Date: Fri, 5 Aug 2011 23:14:12 +0000
Subject: [PATCH] Added MH's gene counting code, changed gene hash tp accept
 1-to-many mappings, removed Biopax publication nodes in output file using
 XML:DOM

svn path=/; revision=141
---
 .../perl_singletons/pathway_gene_swapper.pl   | 49 +++++++++++++++----
 1 file changed, 39 insertions(+), 10 deletions(-)

diff --git a/preecej/perl_singletons/pathway_gene_swapper.pl b/preecej/perl_singletons/pathway_gene_swapper.pl
index f907f45..49e65c7 100644
--- a/preecej/perl_singletons/pathway_gene_swapper.pl
+++ b/preecej/perl_singletons/pathway_gene_swapper.pl
@@ -227,7 +227,12 @@ sub import_data
 
     open(GENE_FILE, $input_gene_file) or die("Could not open $input_gene_file"); 
     
+    # used to generate total counts of each species' gene list; sanity check
+    my $original_gene_count = 0;
+    my $replacement_paralog_count = 0;
+
     my $line = <GENE_FILE>;
+    
     while (<GENE_FILE>)
     {
        $line = $_;
@@ -235,10 +240,31 @@ sub import_data
        my @line_ary = split(',',$line);
        my $data_field = $line_ary[0];
        my $data_val = $line_ary[1];
-
-       $swap_genes{$data_field} = $data_val; 
+       
+       #Does ath_gene exist?
+       if (!exists $swap_genes{$data_field})
+       {
+           $original_gene_count++;
+       }
+       $replacement_paralog_count++; # count this every time
+       
+       # add new gene to hash value (array) for old gene hash key
+       push @{$swap_genes{$data_field}},$data_val;      
     } 
-
+    
+    if ($verbose) # give add'l stats on gene and paralog counts
+    {
+        print "[Total number of original genes and paralogs]\n"
+            . "Original gene count: " . $original_gene_count . "\n"
+            . "Replacement paralog count: $replacement_paralog_count\n\n"; 
+    
+        foreach my $orig_gene_key (keys %swap_genes)
+        {
+            print "$orig_gene_key: " . scalar(@{$swap_genes{$orig_gene_key}}) . "\n";
+        }
+        print "\n";
+    }
+    
     close(GENE_FILE);
 
     print "Opening GPML pathway file and reading data...\n\n";
@@ -282,11 +308,16 @@ sub swap_genes
     # [PathVisio Perl Pseudo-Script]
     # -------------------------------------------------------------------------
     # .remove all <BiopaxRef> and <bp:PublicationXref> elements and children
-    # my $pub_nodes = $gpml_doc->getElementsByTagName("bp:PublicationXref");
-    # for my $pub_node (\$pub_nodes)
-    # {
-    #     $pub_node->removeChild();
-    # }
+    my $pathway_nodes = $gpml_doc->getElementsByTagName("Pathway");
+    print @$pathway_nodes[0] . "\n";
+    my $biopax_nodes = $gpml_doc->getElementsByTagName("Biopax");
+    print $biopax_nodes . "\n";
+    for (@$biopax_nodes)
+    {
+        # print $_->getTagName . "\n";
+        print $_ . "\n";
+        @$pathway_nodes[0]->removeChild($_);
+    }
     
     # .identify mapped genes
     #   .if mapped gene did not belong to a group, create new group node
@@ -305,8 +336,6 @@ sub swap_genes
     #   .color the box (<Graphics...Color="4488ff" />)
     # .After PathVisio sanity check, before communication:
     #   .remove back-ref TAIR comments
-
-    
     
     # [usage example]
     # my $nodes = $gpml_doc->getElementsByTagName("DataNode");
-- 
2.34.1