//process input into three profiles Hs,Hu,Hn
//s = selected profiles, u = unselected in sample,n=in native only
//connects them, puts Hs in cluster with freq = 0 if not already there
int process_one_input(FILE *fp) {
HASHTBL *halfbrac;
FILE *file;
char temp[100],tmp[ARRAYSIZE],*profile,*fullprofile,*diff,*native,*diffn,*dup;
int i,j,k,*id,last=0,insample;
int numhelix = 0,fullnum = 0,natnum = 0;
int size = INIT_SIZE,size2 = INIT_SIZE,size3 = INIT_SIZE,size4 = INIT_SIZE,size5 = INIT_SIZE;
if (!(halfbrac = hashtbl_create(HASHSIZE,NULL))) {
fprintf(stderr, "ERROR: hashtbl_create() for halfbrac failed");
exit(EXIT_FAILURE);
}
if (!(input = hashtbl_create(HASHSIZE,NULL))) {
fprintf(stderr, "ERROR: hashtbl_create() for input failed");
exit(EXIT_FAILURE);
}
//longest = hashtbl_get(max,"longest");
profile = malloc(sizeof(char)*ARRAYSIZE*size);
fullprofile = malloc(sizeof(char)*ARRAYSIZE*size2);
diff = malloc(sizeof(char)*ARRAYSIZE*size3);
native = malloc(sizeof(char)*ARRAYSIZE*size4);
diffn = malloc(sizeof(char)*ARRAYSIZE*size5);
profile[0] = '\0';
fullprofile[0] = '\0';
diff[0] = '\0';
native[0] = '\0';
diffn[0] = '\0';
if (!(file = fopen(INPUT,"r")))
fprintf(stderr,"Cannot open %s\n",INPUT);
while (fgets(temp,100,file)) {
// if (sscanf(temp,"Structure %d (%d)",&i,&prob) == 2)
if (sscanf(temp,"%d %d %d",&i,&j,&k) == 3) {
insample = 1;
sprintf(tmp,"%d %d",i,j);
id = hashtbl_get(bp,tmp);
if (!id) {
id = process_native(i,j,k);
//printf("number in marginals %d\n",hashtbl_numkeys(marginals));
if (*id > hashtbl_numkeys(marginals))
insample = 0;
}
if (*id != last) {
sprintf(tmp,"%d",*id);
if (strlen(native)+strlen(tmp) > (ARRAYSIZE*size4-2))
native = realloc(native,ARRAYSIZE*(++size4));
sprintf(native,"%s%s ",native,tmp);
natnum++;
if (insample) {
fullnum++;
if (strlen(fullprofile)+strlen(tmp) > (ARRAYSIZE*size2-2))
fullprofile = realloc(fullprofile,ARRAYSIZE*(++size2));
sprintf(fullprofile,"%s%s ",fullprofile,tmp);
if (hashtbl_get(freq,tmp)) { //is a freq helix, save to profile
numhelix++;
if (strlen(profile)+strlen(tmp) > (ARRAYSIZE*size-2))
profile = realloc(profile,ARRAYSIZE*(++size));
//printf("adding %d to profile\n",*id);
sprintf(profile,"%s%s ",profile,tmp);
}
else { //if not freq record diff
if (strlen(diff)+strlen(tmp)+2 > ARRAYSIZE*size3)
diff = realloc(diff,ARRAYSIZE*(++size3));
sprintf(diff,"%s%s ",diff,tmp);
//printf("printing diff %s\n",diff);
}
}
else {
if (strlen(diffn)+strlen(tmp)+2 > ARRAYSIZE*size5)
diffn = realloc(diffn,ARRAYSIZE*(++size5));
sprintf(diffn,"%s%s ",diffn,tmp);
}
last = *id;
make_brackets(halfbrac,i,j,*id);
}
else if (VERBOSE)
printf("helix %d %d %d is duplicate with id %d: process_input()\n",i,j,k,*id);
}
}
native = sort_input(native,natnum);
//printf("native is now %s\n",native);
make_bracket_rep(halfbrac,native);
hashtbl_destroy(halfbrac);
fullprofile = sort_input(fullprofile,fullnum);
profile = sort_input(profile,numhelix);
//printf("native %s, fullprofile %s, profile %s, diff %s, diffn %s\n",native,fullprofile,profile,diff,diffn);
if (fullnum != natnum)
make_edge_and_node(fp,fullprofile,native,diffn,natnum);
if (numhelix != fullnum)
make_edge_and_node(fp,profile,fullprofile,diff,fullnum);
fprintf(fp,"\"%s\" [style = filled, fillcolor = gray60];\n",profile);
sprintf(tmp,"%d ",numhelix);
if (strlen(tmp)+strlen(profile) > ARRAYSIZE*size+1)
profile = realloc(profile,ARRAYSIZE*(++size));
dup = mystrdup(profile);
sprintf(profile,"%s%s",tmp,dup);
id = hashtbl_get(cluster,profile);
free(dup);
if (!id) {
id = malloc(sizeof(int));
*id = 0;
hashtbl_insert(cluster,profile,id);
//printf("inserting input %s into cluster\n",profile);
}
free(profile);
return numhelix;
}
//takes input profile and checks if in graph
//if so, changes node shape to hexagon
//if not, insert new vertex
HASHTBL* process_input_profile(FILE *fp,HASHTBL *brac,char *fullprofile, int fullnum,char *profile,int numhelix, char *diff, int prob) {
HASHTBL *hash, *temp=NULL;
char *diff1,*bracket,*difftrip;
int *val,k1=0,k2=0;
KEY *parent,*next;
if (!(temp = hashtbl_create(HASHSIZE,NULL))) {
fprintf(stderr, "ERROR: hashtbl_create() failed");
exit(EXIT_FAILURE);
}
make_bracket_rep(brac,fullprofile);
hashtbl_destroy(brac);
profile = sort_input(profile,numhelix);
//printf("(sorted) profile is %s with fullprofile %s and diff %s\n",profile,fullprofile,diff);
if ((hash = graph[numhelix-1]) && (hashtbl_get(hash,profile))) {
if (numhelix == fullnum) {
//printf("case 1: full profile found in graph\n");
fullprofile = sort_input(fullprofile,numhelix);
} else {
//cannot use find_diff because fullprofile has helices not in table[]
//puts("case 2: profile found in graph");
diff = insert_diff(temp,diff);
bracket = edge_label(temp,profile,fullprofile,fullnum);
fprintf(fp,"\"%s\"-> \"%s\" [label =\" %s\\n%s \",fontsize=8,style = dotted];\n",profile,fullprofile,diff,bracket);
}
}
else {
/*
if (numhelix == fullnum)
puts("case 3: full profile not found");
else
puts("case 4: profile not found");
*/
for (parent = find_parents(profile); parent; parent = next) {
diff1 = find_diff(temp,parent->data,profile,&k1,&k2);
if (numhelix != fullnum) {
difftrip = insert_diff(temp,diff);
diff1 = realloc(diff1,strlen(diff1)+strlen(difftrip)+4);
//printf("for parent %s, diff1 is now %s, diff is %s and difftrip is %s\n",parent->data,diff1,diff,difftrip);
sprintf(diff1,"%s\\n%s",diff1,difftrip);
//printf("Diff is %s for parent %s of profile %s; diff %s for %s\n",diff1,parent->data,profile,difftrip,fullprofile);
}
bracket = edge_label(temp,parent->data,fullprofile,fullnum);
fprintf(fp,"\"%s\"-> \"%s\" [label =\" %s\\n%s \",fontsize=8,style = dotted];\n",parent->data,fullprofile,bracket,diff1);
next = parent->next;
free(parent);
}
}
//printf("%s has size %d and prob %d\n",fullprofile,fullnum,prob);
fprintf(fp,"\"%s\" [shape = hexagon];\n",fullprofile);
val = malloc(sizeof(int)*2);
val[0] = fullnum;
val[1] = prob;
hashtbl_insert(input,fullprofile,val);
hashtbl_destroy(temp);
if (!(brac = hashtbl_create(HASHSIZE,NULL))) {
fprintf(stderr, "ERROR: hashtbl_create() failed");
exit(EXIT_FAILURE);
}
return brac;
}
extern input_list* unflatten_area(input_list *entry, int entry_number)
{
input_list *pred = NULL, *input, *new_input;
chain_list* abl;
cell_list* cell;
note_list* note, *all_note=NULL, *best_note;
int cell_source_positiv, neg_cell_source_positiv; /*flag*/
if (!entry) {
fprintf(stderr,"unflatten_area: no input\n");
exit(1);
}
/*test all the simple logic library*/
for (cell=getcell_oper_lib(); cell; cell=cell->NEXT) {
all_note=eval_note(all_note, cell, entry, entry_number);
}
/*no result pattern, impossible to map, signal it to caller function*/
if (!all_note) return entry;
/*take the best for a positiv and negativ results*/
best_note=all_note;
best_note->AVERAGE_COST=(best_note->COST+best_note->NEG_COST)/2;
for (note=all_note->NEXT; note; note=note->NEXT) {
note->AVERAGE_COST=(note->COST+note->NEG_COST)/2;
if (note->AVERAGE_COST<best_note->AVERAGE_COST) best_note=note;
else if (note->AVERAGE_COST==best_note->AVERAGE_COST
&& note->ARITY>best_note->ARITY) best_note=note;
else if (note->AVERAGE_COST==best_note->AVERAGE_COST
&& note->ARITY==best_note->ARITY
&& note->DELAY+note->NEG_DELAY<best_note->DELAY+best_note->NEG_DELAY)
best_note=note;
}
/*one of those cells are missing, impossible to unflat*/
if (!best_note->CELL || !best_note->NEG_CELL) {
fprintf(stderr,
"unflatten_area: opposite cell of %s is missing (%d inputs)\n",
best_note->CELL?best_note->CELL->NAME:best_note->NEG_CELL->NAME,
entry_number);
exit(1);
}
/*build new input*/
new_input=newinput();
new_input->DELAY=best_note->DELAY;
new_input->NEG_DELAY=best_note->NEG_DELAY;
new_input->R=best_note->R;
new_input->NEG_R=best_note->NEG_R;
new_input->ABL=createabloper(ABL_OPER(best_note->CELL->ABL));
new_input->NEG_ABL=createabloper(ABL_OPER(best_note->NEG_CELL->ABL));
cell_source_positiv=is_source_positiv(ABL_OPER(best_note->CELL->ABL));
neg_cell_source_positiv=is_source_positiv(
ABL_OPER(best_note->NEG_CELL->ABL)); /*put operands*/
for (input=entry; input; input=input->NEXT) {
if (best_note->ARITY==0) break;
best_note->ARITY--;
/*positiv*/
if (cell_source_positiv) {
abl=input->ABL;
input->ABL=NULL; /* protect from freeinput() */
}
else {
abl=input->NEG_ABL;
input->NEG_ABL=NULL; /* protect from freeinput() */
}
ABL_CDR(new_input->ABL)=addchain(ABL_CDR(new_input->ABL),abl);
/*opposite*/
if (neg_cell_source_positiv) {
if (input->ABL) abl=input->ABL;
else abl=dupablexpr(abl); /*already used above*/
input->ABL=NULL; /* protect from freeinput() */
}
else {
if (input->NEG_ABL) abl=input->NEG_ABL;
else abl=dupablexpr(abl); /*already used above*/
input->NEG_ABL=NULL; /* protect from freeinput() */
}
ABL_CDR(new_input->NEG_ABL)=addchain(ABL_CDR(new_input->NEG_ABL),abl);
pred=input;
}
pred->NEXT=NULL;
freeinput(entry);
freenote(all_note);
return sort_input(input,new_input); /*return a sort list*/
}
