ontology distance script now accumulates annotations correctly.

scripts/geneontology/go-distance/godist.c

@@ -9,6 +9,7 @@
 void print_terms();
 void add_link(char*, char*);
 struct node* get_bp();
+struct node* get_term(char *);
 
 /* initialisation */
 int godist_init() {
@@ -48,8 +49,26 @@ int godist_init() {
     fclose(tree_fd);
     printf(" %d edges\n", link_count);
 
-    accumulate_evidence(get_bp());
+    for (i=0; i<term_array_size; i++) {
+        clear_flags(get_bp());
+        accumulate_evidence(term_array[i]);
+	printf(".");
+    }
     print_term(get_bp());
+    
+    print_term(get_term("GO:0040007"));
+    print_term(get_term("GO:0007275"));
+    print_term(get_term("GO:0007582"));
+    print_term(get_term("GO:0043473"));
+    print_term(get_term("GO:0000004"));
+    print_term(get_term("GO:0051704"));
+    print_term(get_term("GO:0000003"));
+    print_term(get_term("GO:0016032"));
+    print_term(get_term("GO:0009987"));
+    print_term(get_term("GO:0050896"));
+    print_term(get_term("GO:0050789"));
+
+/*    find_multi_parented();*/
 }
 
 void godist_exit() {
@@ -95,11 +114,7 @@ int godist_read_term(FILE *fd) {
 	e.key = n->term;
 	e.data = (void*)n;
 	res = hsearch(e, ENTER);
-/*	if (res);
+
-	    printf("hash-insert: %s", ((struct node*)res->data)->term);
-        else
-            printf("hash-insert: error");
-*/
         term_array[term_array_size++] = n;
     } 
 
@@ -111,11 +126,6 @@ float go_distance(char *term1, char *term2) {
     return 0.0;
 }
 
-void add_node(char *term, int parentc, struct node **parents) {
-    
-
-}
-
 void clear_flags(struct node *n) {
     int i;
     for (i=0; i<n->childrenc; i++) 
@@ -152,20 +162,20 @@ void add_link(char *parent_id, char *child_id) {
     }
     parent->children[parent->childrenc] = child;
     parent->childrenc++;
-/*    printf("adding %x as child of %x\n", child, parent);*/
     child->parents[child->parentc] = parent;
     child->parentc++;
 }
 
 struct node *get_bp() {
+    return get_term("GO:0008150");
+}
+
+struct node *get_term(char *term) {
     ENTRY e, *ep;
-    e.key = "GO:0008150";
+    e.key = term;
     ep = hsearch(e, FIND);
  
     if (ep) {
-/*	printf("BP is %x and %s\n", (int) ep->data, ep->key);
-	printf("%d\n", ((struct node*)ep->data)->childrenc );
-	printf("%s\n", ((struct node*)ep->data)->term);*/
 	return ep->data;
     }
     return NULL;
@@ -175,6 +185,7 @@ void accumulate_evidence(struct node *n) {
     int i, j;
     if (n->visited)
         return;
+    n->visited = 1;
 
     for (i=0; i<12; i++)
         n->acc_evidence[i] = n->evidence[i];
@@ -186,7 +197,6 @@ void accumulate_evidence(struct node *n) {
                 n->acc_evidence[j] += n->children[i]->acc_evidence[j];
         }
     }
-/*    printf(".");*/
 }
 
 void print_terms() {
@@ -211,3 +221,15 @@ void print_term(struct node *n) {
     printf("\n");
 }
 
+void find_multi_parented() {
+    int i;
+    for (i=0; i<term_array_size; i++) {
+        if (term_array[i]->parentc > 1)
+            printf("%s  -- %d\n", term_array[i]->term, term_array[i]->parentc);
+    }
+}
+
+float calc_ic(unsigned int evidence) {
+    
+}
+

scripts/geneontology/go-distance/godist.h

@@ -7,6 +7,19 @@
 #define MAX_PARENTS 100
 #define MAX_CHILDREN 100
 
+enum EVIDENCE { MP = 1, 
+                IGI = 1 << 1, 
+		IPI = 1 << 2, 
+		ISS = 1 << 3, 
+		IDA = 1 << 4, 
+		IEP = 1 << 5,
+		IEA = 1 << 6, 
+		TAS = 1 << 7, 
+		NAS = 1 << 8, 
+		ND =  1 << 9, 
+		RCA = 1 << 10, 
+		IC =  1 << 11 };
+
 struct node;
 struct node {
     /* GO term id. E.g: "GO:0005180" */