/* false positives */
/* crops_fp.c

   Compile with:   gcc -g crops_fp.c -lm -o crops_fp
   Example run:    crops_fp beta_DEVIATION.X.l6 60 93 14

   Example output: from: 61 to: 80 fp: 40 fn: 13 total: 53 (93)   60  l=6

   Given an agree-DEVIATION file, it outputs the interval of p-values
   for which the best "fp" count is obtained, providing that only the
   runs with at least the specified percentage of true positives are
   included.  In the example above, the best p-interval for phylogen
   runs on the HBB promoter region with l=6 (ct) which exceed 60% true
   positives (ie, < 40% fn) is [61,80], when gaps are not allowed in
   the regions (X).  The fp, fn and total counts are: 40, 13 and 53
   respectively. 93 is the total length of the known functional regions
   (maxfn). The chars_to_len value is 14=strlen("agree.beta.X.l") (see
   the denomination for a run in the DEVIATION file).


   Produces an interval file as an intermediate (tmp) file.

   NOTE: For use in 'cut' experiments on individual regions.
         Fp-based evaluation criteria were not used in the "Comparison
         of five methods..." paper. Refer to crops_total.c, instead.
*/
#include <stdio.h>
#include <stdlib.h>
#include <strings.h>
#include <ctype.h>
#include <sys/types.h>
#include <sys/param.h>


#define Usage "%s <input file> <pct> <length> <chars_to_len>"

void fatal(char *message);
void fatalf(char *msg, char *val);
FILE *ckopen(char *name, char *mode);
int get_len(char *s);



int main(int argc, char *argv[])
{
    char   buffer[200], cmd[100];
    int    i, k, length, pct, best_k, best, thresh, chars_to_len;
    int    from[1000], to[1000];
    int    fn[1000], fp[1000];
    char  *s;
    FILE  *file;
    
    if (argc!=5) fatalf(Usage, argv[0]);
    pct = atoi(argv[2]);
    length = atoi(argv[3]);
    chars_to_len = atoi(argv[4]);
    thresh = ((100-pct)*length)/100;
    best = 100000; i = best_k = -1; 
    
    sprintf(cmd,"interval %s %d > tmp",argv[1], chars_to_len);
    system(cmd);

    file = ckopen("tmp","r");
    k = 0;
    while (fgets(buffer,200,file)!=NULL) {
       if (buffer[0]=='#') continue;
 
       sscanf(buffer,"%d %d %d %d", 
                     &from[k], &to[k], &fp[k], &fn[k]);
       if (fn[k]>thresh) continue;
       if (fp[k]+fn[k]<best) { best = fp[k]+fn[k]; best_k = k; }
       k++;
    }
    fclose(file);
    system("rm -f tmp");
    
    i = k-1;
    while ((i>=1) && (fp[i]==fp[i-1])) i--;

    if (i>=0) 
    printf("from: %d to: %d fp: %d fn: %d total: %d (%d)   %d l=%d\n",
            from[i], to[i], fp[i], fn[i], fp[i]+fn[i], length, pct, get_len(argv[1]));

    
    return 0;
}

int get_len(char *s)
{
    int len;
    char *p = s;
    
    while (*p && !isspace(*p) && (*p!='l')) p++; 
    sscanf(p+1,"%d",&len);

    return len;
}

void fatal(char *msg)
{                
        (void)fprintf(stderr, "%s\n", msg);
        exit(1); 
}    
     
void fatalf(char *msg, char *val)
{    
        (void)fprintf(stderr, msg, val);  
        (void)putc('\n', stderr);
        exit(1);
}            
          
FILE *ckopen(char *name, char *mode)
{    
        FILE *fp;
     
        if ((fp = fopen(name, mode)) == NULL)
                fatalf("Cannot open %s.", name);
        return fp;
}