wrote outline for sub perform_map(); successfully added code for mapping

on RiceCyc term name; added header line to output file

svn path=/; revision=61

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 4ff66d98f6249896acb592e8907dc937831314ae..5d01cd72c5a439d74e71f2a1bf0a1a14d287e0be 100755 (executable)
--- a/preecej/perl_singletons/reactome_chebi_mapping/reactome_chebi_mapping.pl
+++ b/preecej/perl_singletons/reactome_chebi_mapping/reactome_chebi_mapping.pl
@@ -12,13 +12,13 @@ use strict;
 #   CHEBI OBO file (preset)
 #   Rice Reactome file (preset, provided by YuanMing Wu)
 #   (Header)    [ReactomeID]    [Compound_Name]            [CAS]           [LIGAND]    [RiceCyc]
-#   (Row)       923893          S-adenosyl-L-methionine    29908-03-0      C00019      S-ADENOSYLMETHIONINE        ** please use a '-' (dash) symbol for any empty columns
+#   (Row)       923893          S-adenosyl-L-methionine    29908-03-0      C00019      S-ADENOSYLMETHIONINE        ** the '-' (dash) symbol will be applied to any empty columns
 #
 # Outputs: tab-del mapping file (reactome_chebi_mapping.txt)
 #   (Header)    [ReactomeID]    [CHEBI]    [XREF_Type]    [XREF_ID]       
 #   (Row)       923893          15414      CAS            29908-03-0
 #   (Row)       923893          15414      LIGAND         C00019
-#   (Row)       923893          15414      RiceCyc        S-ADENOSYLMETHIONINE      ** this would be a rare mapping occurrence; only if CAS and LIGAND mappings are not available
+#   (Row)       923893          15414      RiceCyc        S-ADENOSYLMETHIONINE      ** this would be a rare mapping occurrence
 # --------------------------------------------------------------------
 
 
@@ -28,7 +28,6 @@ use strict;
 
 use Bio::OntologyIO;
 
-
 # --------------------------------------------------------------------
 # declarations
 # --------------------------------------------------------------------
@@ -45,7 +44,7 @@ my %reactome_CAS; # rice reactome CAS hash
 my %reactome_LIGAND; # rice reactome LIGAND hash
 my %reactome_RiceCyc; # rice reactome RiceCyc hash
 
-my @map_results; # successful mappings between chebi and reactome
+my @map_results = (); # successful mappings between chebi and reactome
 
 
 # --------------------------------------------------------------------
@@ -97,7 +96,7 @@ sub init
         }
 
         # similarly...
-        
+
         # --LIGAND Hash Load--
         if ($LIGAND_id ne "-")
         {
@@ -107,10 +106,9 @@ sub init
         # --RiceCyc Hash Load--
         if ($RiceCyc_term ne "-")
         {
-            push @{$reactome_RiceCyc{$RiceCyc_term}}, $reactome_id;
+            push @{$reactome_RiceCyc{"\U$RiceCyc_term"}}, $reactome_id;
         }
     }
-
     close REACTOME_FILE;
 }
 
@@ -127,58 +125,114 @@ sub test_inputs
         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";
-    #}
+    print "\n[CHEBI Term List from \$ont]\n";
+    foreach my $term ($ont->get_all_terms) {
+        my @synonyms = $term->get_synonyms;
+        my @xrefs = $term->get_dbxrefs;
+
+        print $term->identifier;
+        print " \|NAME\| ";
+        if (defined($term->name)) {
+            print $term->name;
+        }
+        print " \|SYNONYMS\| ";
+        print "$_," foreach @synonyms;
+        print " \|XREFS\| ";
+        print "$_" foreach @xrefs;
+        foreach my $xref (@xrefs) {
+            print $xref->primary_id;
+        }
+        print "\n\n";
+    }
 
-    # show reactome hashes
+    # show reactome hashes - this is important, give >1 dupes to Pankaj
+    # for manual reference
     my $k; my @v;
     print "\n[Reactome Hashes]\n";
     print "\n--CAS Hash--\n";
     for $k (keys %reactome_CAS) {
-        print "$k: @{$reactome_CAS{$k}}\n";
+        #if (@{$reactome_CAS{$k}} > 1) {
+            print "$k: @{$reactome_CAS{$k}}\n";
+        #}
     }
     print "\n--LIGAND Hash--\n";
     for $k (keys %reactome_LIGAND) {
-        print "$k: @{$reactome_LIGAND{$k}}\n";
+        #if (@{$reactome_LIGAND{$k}} > 1) {
+            print "$k: @{$reactome_LIGAND{$k}}\n";
+        #}
     }
     print "\n--RiceCyc Hash--\n";
     for $k (keys %reactome_RiceCyc) {
-        print "$k: @{$reactome_RiceCyc{$k}}\n";
+        #if (@{$reactome_RiceCyc{$k}} > 1) {
+            print "$k: @{$reactome_RiceCyc{$k}}\n";
+        #}
     }
-
 }
 
 
-# map the chebi terms to the reactome entries (brute-force algo)
+# map the chebi terms to the reactome entries
 # --------------------------------------------------------------------
 sub perform_map
 {
-    # iterate ontology; use brute-force matching to map
-    my @chebi_obo_terms = $ont->get_all_terms();
+    my @chebi_obo_terms = $ont->get_all_terms;
+    #print $_->identifier . "\n" foreach @chebi_obo_terms;
 
-    #print $_->identifier() . "\n" foreach @chebi_obo_terms;
+    # loop through each chebi term
+    foreach my $term (@chebi_obo_terms)
+    {
+        # set locals for matching each term property
+        my $term_name;
+        if (defined($term->name)) {
+            $term_name = $term->name;
+        } else {
+            $term_name = "";
+        }
+        my @term_synonyms = $term->get_synonyms;
 
-    # --pseudo--
-    #   loop through each chebi term (~33K)
-    #       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
+        # attempt CHEBI match on CAS ID
+
+        # attempt CHEBI match on LIGAND ID
         
+        # attempt CHEBI match on RiceCyc names
+        if (defined($reactome_RiceCyc{"\U$term_name"})) {
+            push (@map_results, "$reactome_RiceCyc{$term_name}\t",
+                "$term->identifier\t",
+                "RiceCyc\t",
+                $term_name);
+        } else { # check the term synonyms, if needed
+            foreach my $synonym (@term_synonyms) {
+                print "";
+            }
+        }
+    }
 }
 
+# sample format - remove later
+#   [ReactomeID]    [CHEBI]    [XREF_Type]    [XREF_ID]       
+#   923893          15414      CAS            29908-03-0
+#   923893          15414      LIGAND         C00019
+#   923893          15414      RiceCyc        S-ADENOSYLMETHIONINE
+
 
 # put the results in the mapped output file
 # --------------------------------------------------------------------
 sub create_mapfile
 {
-    # setup output file
-    open(OUTPUT_FILE,">>" . $data_path . $mapped_output_file);
-
-    #format results for file output
-    print OUTPUT_FILE "$_\n" foreach @map_results;
+    if (@map_results > 0)
+    {
+        # add a header to the results array
+        unshift (@map_results, "ReactomeID\tCHEBI\tXREF_Type\tXREF_ID");
+        
+        # setup output file
+        open(OUTPUT_FILE,">>" . $data_path . $mapped_output_file);
     
-    close OUTPUT_FILE;
+        #format results for file output
+        print OUTPUT_FILE "$_\n" foreach @map_results;
+        
+        close OUTPUT_FILE;
+    } else {
+        print "\n\nSorry, there are no mapped results.\n\n";
+    }
 }
 
 
@@ -187,7 +241,7 @@ sub create_mapfile
 # --------------------------------------------------------------------
 
 init;
-test_inputs;
+#test_inputs;
 perform_map;
 create_mapfile;