res_stats.c

序列对齐 Compare a protein sequence to a protein sequence database or a DNA sequence to a DNA sequenc

    return;
  }

  max_dev = 0.0;
  mh1 = hist.maxh-1;
  mht = (3*hist.maxh-3)/4 - 1;

  if (!m_msg.nohist && mh1 > 0) {
    for (i=0,maxval=0,maxvalt=0; i<hist.maxh; i++) {
      if (hist.hist_a[i] > maxval) maxval = hist.hist_a[i];
      if (i >= mht &&  hist.hist_a[i]>maxvalt) maxvalt = hist.hist_a[i];
    }
    n_chi_sq = 0;
    cum_hl = -hist.hist_a[0];
    dotsiz = (maxval-1)/60+1;
    ddotsiz = (maxvalt-1)/50+1;
    doinset = (ddotsiz < dotsiz && dotsiz > 2);

    if (pst.zsflag>=0)
      fprintf(fd,"       opt      E()\n");
    else 
      fprintf(fd,"     opt\n");

    prev_e =  zs_to_Ec((double)(hist.min_hist-hist.histint/2),hist.entries);
    for (i=0; i<=mh1; i++) {
      pch = (i==mh1) ? '>' : ' ';
      pch = (i==0) ? '<' : pch;
      hll = hl = hist.hist_a[i];
      if (pst.zsflag>=0) {
	cum_hl += hl;
	f_int = (double)(i*hist.histint+hist.min_hist)+(double)hist.histint/2.0;
	cur_e = (double)zs_to_Ec(f_int,hist.entries);
	ev = el = ell = (int)(cur_e - prev_e + 0.5);
	if (hl > 0  && i > 5 && i < (90-hist.min_hist)/hist.histint) {
	  x_tmp  = fabs(cum_hl - cur_e);
	  if ( x_tmp > max_dev) {
	    max_dev = x_tmp;
	    max_i = i;
	  }
	  n_chi_sq++;
	}
	if ((el=(el+dotsiz-1)/dotsiz) > 60) el = 60;
	if ((ell=(ell+ddotsiz-1)/ddotsiz) > 40) ell = 40;
	fprintf(fd,"%c%3d %5d %5d:",
		pch,(i<mh1)?(i)*hist.histint+hist.min_hist :
		mh1*hist.histint+hist.min_hist,hl,ev);
      }
      else fprintf(fd,"%c%3d %5d :",
		   pch,(i<mh1)?(i)*hist.histint+hist.min_hist :
		   mh1*hist.histint+hist.min_hist,hl);

      if ((hl=(hl+dotsiz-1)/dotsiz) > 60) hl = 60;
      if ((hll=(hll+ddotsiz-1)/ddotsiz) > 40) hll = 40;
      for (j=0; j<hl; j++) hline[j]='='; 
      if (pst.zsflag>=0) {
	if (el <= hl ) {
	  if (el > 0) hline[el-1]='*';
	  hline[hl]='\0';
	}
	else {
	  for (j = hl; j < el; j++) hline[j]=' ';
	  hline[el-1]='*';
	  hline[hl=el]='\0';
	}
      }
      else hline[hl] = 0;
      if (i==1) {
	for (j=hl; j<10; j++) hline[j]=' ';
	sprintf(&hline[10]," one = represents %d library sequences",dotsiz);
      }
      if (doinset && i == mht-2) {
	for (j = hl; j < 10; j++) hline[j]=' ';
	sprintf(&hline[10]," inset = represents %d library sequences",ddotsiz);
      }
      if (i >= mht&& doinset ) {
	for (j = hl; j < 10; j++) hline[j]=' ';
	hline[10]=':';
	for (j = 11; j<11+hll; j++) hline[j]='=';
	hline[11+hll]='\0';
	if (pst.zsflag>=0) {
	  if (ell <= hll) hline[10+ell]='*';
	  else {
	    for (j = 11+hll; j < 10+ell; j++) hline[j]=' ';
	    hline[10+ell] = '*';
	    hline[11+ell] = '\0';
	  }
	}
      }

      fprintf(fd,"%s\n",hline);
      prev_e = cur_e;
    }
  }

  if (ntt.carry==0) {
    fprintf(fd, "%7ld residues in %5ld sequences\n", ntt.length, ntt.entries);
  }
  else {
    db_tt = (double)ntt.carry*(double)LONG_MAX + (double)ntt.length;
    fprintf(fd, "%.0f residues in %5ld library sequences\n", db_tt, ntt.entries);
  }

  if (pst.zsflag>=0) {
    if (MAX_STATS < hist.entries)
      fprintf(fd," statistics extrapolated from %d to %ld sequences\n",
	      MAX_STATS,hist.entries);
    /*    summ_stats(stat_info); */
    fprintf(fd," %s\n",hist.stat_info);
    if (!m_msg.nohist && cum_hl > 0)
      fprintf(fd," Kolmogorov-Smirnov  statistic: %6.4f (N=%d) at %3d\n",
	      max_dev/(double)cum_hl, n_chi_sq,max_i*hist.histint+hist.min_hist);
    if (m_msg.markx & MX_M10FORM) {
      while ((bp=strchr(hist.stat_info,'\n'))!=NULL) *bp=' ';
      if (cum_hl <= 0) cum_hl = -1;
      sprintf(info_hstring1,"; mp_extrap: %d %ld\n; mp_stats: %s\n; mp_KS: %6.4f (N=%d) at %3d\n",
	      MAX_STATS,hist.entries,hist.stat_info,max_dev/(double)cum_hl, n_chi_sq,max_i*hist.histint+hist.min_hist);
    }
  }
  fprintf(fd,"\n%s\n",info_gstring2);
  fflush(fd);
}

showbest(struct db_str ntt)
  {
    int ib, istart, istop;
    char bline[200], fmt[40], pad[200];
    char rline[20];
    int ntmp;
    int lcont, ccont, loff;
    int hcutoff;

    sprintf(fmt,"%%-%ds (%%3d)",llen-10);

    nshow = min(20,nbest);
    mshow = min(20,nbest);

    if (outtty) {
      printf(" How many scores would you like to see? [%d] ",nshow);
      fflush(stdout);
      if (fgets(rline,sizeof(rline),stdin)==NULL) exit(0);
      if (rline[0]!='\n' && rline[0]!=0) sscanf(rline,"%d",&nshow);
      if (nshow<=0) nshow = min(20,nbest);
    }
    else nshow=mshow;

    memset(pad,' ',llen-10);
    pad[llen-31]='\0';
    if (zsflag)
      fprintf(outfd,"The best scores are:%s s-w Z-score E(%ld)\n",pad,ntt.entries);
    else
      fprintf(outfd,"The best scores are:%s s-w\n",pad);

    if (outfd != stdout)
      if (zsflag)
	fprintf(stdout,"The best scores are:%s s-w Z-score E(%ld)\n",pad,ntt.entries);
      else
	fprintf(stdout,"The best scores are:%s s-w\n",pad);

    istart = 0;
  l1:	istop = min(nbest,nshow);
  for (ib=istart; ib<istop; ib++) {
    bbp = bptr[ib];

    if (!outtty && zsflag && bbp->escore > e_cut) {
      nshow = ib;
      goto done;
    }

    sprintf(bline,"%-12s %d",bbp->libstr,bbp->lib);
    bline[13]='\0';

    fprintf(outfd,fmt,bline,bbp->n1);

    if (zsflag)
      fprintf(outfd,"%4d %4.1f %6.2g\n",
	      bbp->score,bbp->zscore,
	      bbp->escore);
    else 
      fprintf(outfd,"%4d\n",bbp->score);

    if (outfd!=stdout) {
      fprintf(stdout,fmt,bline,bbp->n1);
      if (zsflag)
	printf("%4d %4.1f %6.2g\n",
	       bbp->score,bbp->zscore,
	       bbp->escore);
      else 
	printf("%4d\n",bbp->score);
    }
  }

  fflush(outfd); if (outfd!=stdout) fflush(stdout);

  if (outtty) {
    printf(" More scores? [0] ");
    fflush(stdout);
    if (fgets(rline,sizeof(rline),stdin)==NULL) exit(0);
    ntmp = 0;
    if (rline[0]!='\n' && rline[0]!=0) sscanf(rline,"%d",&ntmp);
    if (ntmp<=0) ntmp = 0;
    if (ntmp>0) {
      istart = istop;
      nshow += ntmp;
      mshow += ntmp;
      goto l1;
    }
  }
  else if (zsflag && bbp->escore < e_cut) {
    istart=istop;
    nshow += 10;
    goto l1;
  }

  done:
  if (outfd!=stdout) fprintf(outfd,"\n");
}

selectz(k,n)	/* k is rank in array */
     int k,n;
{
  int t, i, j, l, r;
  double v;
  struct beststr *tmptr;

  l=0; r=n-1;

  while ( r > l ) {
    i = l-1;
    j = r;
    v = bptr[r]->zscore;
    do {
      while (bptr[++i]->zscore > v ) ;
      while (bptr[--j]->zscore < v ) ;
      tmptr = bptr[i]; bptr[i]=bptr[j]; bptr[j]=tmptr;
    } while (j > i);
    bptr[j]=bptr[i]; bptr[i]=bptr[r]; bptr[r]=tmptr;
    if (i>=k) r = i-1;
    if (i<=k) l = i+1;
  }
}

sortbest()
{
  int cmps(), cmp1(), cmpa(), cmpz();
  ksort(bptr,nbest,cmps);
}

sortbeste()
{
  int cmpe();
  ksort(bptr,nbest,cmpe);
}

sortbestz()
{
  int cmpz();
  ksort(bptr,nbest,cmpz);
}

cmps(ptr1,ptr2)
     struct beststr *ptr1, *ptr2;
{
  if (ptr1->score < ptr2->score) return (1);
  else if (ptr1->score > ptr2->score) return (-1);
  else return (0);
}

cmpe(ptr1,ptr2)
     struct beststr *ptr1, *ptr2;
{
  if (ptr1->escore < ptr2->escore) return (-1);
  else if (ptr1->escore > ptr2->escore) return (1);
  else return (0);
}

cmpz(ptr1,ptr2)
     struct beststr *ptr1, *ptr2;
{
  if (ptr1->zscore < ptr2->zscore) return (1);
  else if (ptr1->zscore > ptr2->zscore) return (-1);
  else return (0);
}

ksort(v,n,comp)
     char *v[]; int n, (*comp)();
{
  int gap, i, j;
  char *tmp;
	
  for (gap=n/2; gap>0; gap/=2)
    for (i=gap; i<n; i++)
      for (j=i-gap; j>=0; j -= gap) {
	if ((*comp)(v[j],v[j+gap]) <=0)
	  break;
	tmp = v[j]; v[j]=v[j+gap]; v[j+gap]=tmp;
      }
}

/*
do_bout(FILE *bout,struct stat_str **bptr, int nbest)
{
  int i, min_hist, max_hist;
  double mu, var;

  if (bout==NULL) return;

  inithist();
  for (i = 0; i<nbest; i++)
    addhist(bptr[i]->score,bptr[i]->n1);

  for (i=0; i<MAX_LLEN; i++)
    if (llen_hist[i]>0) {
      min_hist=i;
      break;
    }

  for (i=MAX_LLEN-1; i>=0; i--)
    if (llen_hist[i]>0) {
      max_hist=i;
      break;
    }

  for (i=min_hist; i<=max_hist; i++) {
    mu=(double)score_sums[i]/(double)llen_hist[i];
    if (llen_hist[i]>1) {
      var = ((double)score2_sums[i]-(double)llen_hist[i]*mu*mu)/
	(double)(llen_hist[i]-1);

      fprintf(bout,"%d\t%d\t%.1f\t%.1f\t%.1f\t%.4f\t%.4f\n",
	      i,llen_hist[i],exp(((double)(i))/LN_FACT),
	      score_sums[i],score2_sums[i],mu,var);
    }
  }
  free_hist();
  fclose(bout);
}
*/

s_abort()
{
  exit(1);
}