From 28f0f53e8aab490fe8318f3cd4d298139474997f Mon Sep 17 00:00:00 2001
From: einarr <einarr@pvv.ntnu.no>
Date: Tue, 3 Apr 2007 13:08:48 +0000
Subject: [PATCH] go-distance script soon calculating resnik distances.

---
 scripts/geneontology/go-distance/Makefile |  2 +-
 scripts/geneontology/go-distance/godist.c | 60 ++++++++++++++++++++---
 scripts/geneontology/go-distance/godist.h |  6 +++
 3 files changed, 60 insertions(+), 8 deletions(-)

diff --git a/scripts/geneontology/go-distance/Makefile b/scripts/geneontology/go-distance/Makefile
index 2d05f5d..b2d3978 100644
--- a/scripts/geneontology/go-distance/Makefile
+++ b/scripts/geneontology/go-distance/Makefile
@@ -5,7 +5,7 @@ godist.o: godist.c godist.h
 	gcc -ggdb -c godist.c
 
 go-distance: godist.o main.o
-	gcc -ggdb -o go-distance godist.o main.o
+	gcc -ggdb -o go-distance godist.o main.o -lm
 
 clean:
 	-rm go-distance godist.o main.o 
diff --git a/scripts/geneontology/go-distance/godist.c b/scripts/geneontology/go-distance/godist.c
index 575fbdc..63ae7d0 100644
--- a/scripts/geneontology/go-distance/godist.c
+++ b/scripts/geneontology/go-distance/godist.c
@@ -1,4 +1,5 @@
 
+#include <math.h>
 #include <string.h>
 #include <errno.h>
 #include <stdio.h>
@@ -10,7 +11,8 @@ void print_terms();
 void add_link(char*, char*);
 struct node* get_bp();
 struct node* get_term(char *);
-float calc_ic(struct node *, unsigned int);
+void calc_ic(struct node *, unsigned int);
+struct node *common_subsumer(struct node *, struct node *);
 
 
 /* initialisation */
@@ -19,6 +21,7 @@ int godist_init() {
     hcreate(MAX_NODES);
     term_array_size = 0;
     link_count = 0;
+    struct node *n;
 
     /* Read ontology terms from file */
     printf("Reading GO terms from go-terms.txt...");
@@ -51,13 +54,24 @@ int godist_init() {
     fclose(tree_fd);
     printf(" %d edges\n", link_count);
 
+    printf("Calculating accumulated evidence...");
+    fflush(stdout);
     for (i=0; i<term_array_size; i++) {
         clear_flags(get_bp());
         accumulate_evidence(term_array[i]);
-	printf(".");
     }
+    printf("\n");
+
+    evidence = 0xff;
+    total_ann = 0;
+    n = get_bp();
+    for (i=0; i<12; i++)
+	if (evidence & 1<<i)
+	    total_ann += n->acc_evidence[i];
+    printf("Using %d annotations.\n", total_ann);
+
     print_term(get_bp());
-    
+   /* 
     print_term(get_term("GO:0040007"));
     print_term(get_term("GO:0007275"));
     print_term(get_term("GO:0007582"));
@@ -69,9 +83,14 @@ int godist_init() {
     print_term(get_term("GO:0009987"));
     print_term(get_term("GO:0050896"));
     print_term(get_term("GO:0050789"));
-
-    calc_ic(get_bp(), 0xffff);
+*/
+    printf("Calculation information content...");
+    fflush(stdout);
+    calculate_ics(0xffff);
+    printf("\n");
+/*    calc_ic(get_bp(), 0xffff);*/
 /*    find_multi_parented();*/
+    common_subsumer(get_term("GO:0000003"), get_term("GO:0000004"));
 }
 
 void godist_exit() {
@@ -232,12 +251,39 @@ void find_multi_parented() {
     }
 }
 
-float calc_ic(struct node *n, unsigned int evidence) {
+void calculate_ics(unsigned int evidence) {
+    int i;
+    for (i=0; i<term_array_size; i++)
+	calc_ic(term_array[i], evidence);
+}
+
+void calc_ic(struct node *n, unsigned int evidence) {
     int i;
     float ann=0.0;
     for (i=0; i<12; i++)
 	if (evidence & 1<<i)
 	    ann += (float) n->acc_evidence[i];
-    printf("%f", ann);
+    n->ic = -log(ann/total_ann);
+    printf("%f\n", n->ic);
+}
+
+struct node *common_subsumer(struct node *n1, struct node *n2) {
+    struct node *anc1[MAX_NODES];
+    struct node *anc2[MAX_NODES];
+    int ancc1=0, ancc2=0;
+
+    add_ancestors(&ancc1, anc1, n1);
+    add_ancestors(&ancc2, anc2, n2);
+    printf("Ancestors: %d %d\n", ancc1, ancc2);
+}
+
+void add_ancestors(int *ancc, struct node *anc[], struct node *n) { 
+    int i=0;
+    anc[(*ancc)++] = n;
+    for (i=0; i<n->parentc; i++)
+        add_ancestors(ancc, anc, n->parents[i]);
+}
+
+float resnik(struct node *n1, struct node *n2) {
 }
 
diff --git a/scripts/geneontology/go-distance/godist.h b/scripts/geneontology/go-distance/godist.h
index a074f6e..4a3f918 100644
--- a/scripts/geneontology/go-distance/godist.h
+++ b/scripts/geneontology/go-distance/godist.h
@@ -48,6 +48,8 @@ struct node {
 struct node* term_array[MAX_NODES];
 long term_array_size;
 int link_count;
+int total_ann;
+int evidence; /* bitvector with one bit per evidence code */
 
 /* Ontology initialisation functions. */
 int godist_init();
@@ -58,6 +60,10 @@ void accumulate_evidence(struct node*);
 /* Distance metric functions */
 float resnik_distance(char *term1, char *term2);
 float fussimeg_distance(char *term1, char *term2);
+void calc_ic(struct node *n, unsigned int evidence);
+
+void clear_flags(struct node *n);
+void print_term(struct node *n);
 
 #endif