From 989fe550e8720e4bd2b05aaf8e632809c725f4f4 Mon Sep 17 00:00:00 2001
From: preecej <preecej@localhost>
Date: Fri, 8 Oct 2010 19:30:50 +0000
Subject: [PATCH] hash loads of CAS, LIGAND, RiceCyc

svn path=/; revision=60
---
 .../reactome_chebi_mapping.pl                 | 73 ++++++++++++-------
 1 file changed, 46 insertions(+), 27 deletions(-)

diff --git a/preecej/perl_singletons/reactome_chebi_mapping/reactome_chebi_mapping.pl b/preecej/perl_singletons/reactome_chebi_mapping/reactome_chebi_mapping.pl
index cc12279..4ff66d9 100755
--- a/preecej/perl_singletons/reactome_chebi_mapping/reactome_chebi_mapping.pl
+++ b/preecej/perl_singletons/reactome_chebi_mapping/reactome_chebi_mapping.pl
@@ -77,36 +77,41 @@ sub init
         chomp $line;
         my @reactome_entry = split(/\t/, $line); # break up our tab-del line
 
+        # load up this reactome entry's ID, CAS, LIGAND, and RiceCyc values
+        my $reactome_id = $reactome_entry[0];
+        my $CAS_id = $reactome_entry[2];
+        my $LIGAND_id = $reactome_entry[3];
+        my $RiceCyc_term = $reactome_entry[4];
+
         # There is a possibility that a single CAS, LIGAND, or RiceCyc
         # identifier may appear in more than one reactome entry. This
-        # temp array allows each hash match to hold more than 
-        # ReactomeID, if necessary.
-        my @temp_ary = ();
+        # temp array allows each matched hash value to hold more than 
+        # one ReactomeID, if necessary.
         
         # --CAS Hash Load--
-        # if this reactome entry is already in the hash, just 
-        # append it to the hash's array
-        if(defined($reactome_CAS{$reactome_entry[2]}))
+        if ($CAS_id ne "-") # keep those "-" placeholders out
+        {
+            # build the CAS hash; each value may hold 1...n reactome
+            # ids (as an array)
+            push @{$reactome_CAS{$CAS_id}}, $reactome_id;
+        }
+
+        # similarly...
+        
+        # --LIGAND Hash Load--
+        if ($LIGAND_id ne "-")
         {
-            $reactome_CAS{push(@_,$reactome_entry[2])} = $reactome_entry[0];
-        # otherwise, just add the reactome entry to the hash
-        } else {
-            @temp_ary = (); # clear it for re-use
-            if ($reactome_entry[2] != "-") # keep those "-" placeholders out
-            {
-                $reactome_CAS{$temp_ary[$reactome_entry[2]]}
-                    = $reactome_entry[0];
-            }
+            push @{$reactome_LIGAND{$LIGAND_id}}, $reactome_id;
+        }
+
+        # --RiceCyc Hash Load--
+        if ($RiceCyc_term ne "-")
+        {
+            push @{$reactome_RiceCyc{$RiceCyc_term}}, $reactome_id;
         }
     }
 
     close REACTOME_FILE;
-    
-    # read rice reactome file into array
-    #open(REACTOME_FILE,$data_path . $reactome_file);
-    #@reactome_ref_molecules = <REACTOME_FILE>;
-    #chomp(@reactome_ref_molecules);
-    #close REACTOME_FILE;
 }
 
 
@@ -120,12 +125,29 @@ sub test_inputs
         scalar($ont->get_root_terms)," root terms, and ",
         scalar($ont->get_all_terms)," total terms, and ",
         scalar($ont->get_leaf_terms)," leaf terms\n";
-    
+
+    # all chebi terms in the ontology
+    #print "\n[CHEBI Term List from \$ont]\n";
+    #foreach my $term ($ont->get_all_terms()) {
+        #print $term->identifier() . "\n";
+    #}
+
     # show reactome hashes
+    my $k; my @v;
     print "\n[Reactome Hashes]\n";
     print "\n--CAS Hash--\n";
-    print "$_\n" foreach %reactome_CAS;
-    print "\n";
+    for $k (keys %reactome_CAS) {
+        print "$k: @{$reactome_CAS{$k}}\n";
+    }
+    print "\n--LIGAND Hash--\n";
+    for $k (keys %reactome_LIGAND) {
+        print "$k: @{$reactome_LIGAND{$k}}\n";
+    }
+    print "\n--RiceCyc Hash--\n";
+    for $k (keys %reactome_RiceCyc) {
+        print "$k: @{$reactome_RiceCyc{$k}}\n";
+    }
+
 }
 
 
@@ -143,9 +165,6 @@ sub perform_map
     #       if match chebi to reactome hashes on CAS, LIGAND, and/or RiceCyc name;
     #           write tab-del mapping string directly to results array for EACH type of match
         
-    foreach my $term (@chebi_obo_terms) {
-        print $term->identifier() . "\n";
-    }
 }
 
 
-- 
2.34.1