highlight.cc

unix或linux下的DNA分析软件源码 其功能如下 1. Edit up to 256 peptide or DNA sequences simultaneously. 2. Transla

//-------------------------------------------------------------------------//
// highlight.cc                                                            //
// latest revision: 07-11-2000                                             //
// Copyright (C) 2000 by Thomas J. Nelson                                  //
//-------------------------------------------------------------------------//

#include <stdio.h>
#include <stdlib.h>
#include <math.h>
#include <string.h>
#include <ctype.h>
#define min(a,b) (((a)<(b)) ? (a) : (b))
#define max(a,b) (((a)>(b)) ? (a) : (b))
enum{ TITLE, BOX, COMMENT, LABEL, TEXT };

char *getstring(char* heading, char* string, char *defstring);
int getint(char *heading, int &answer, int defint);
double getdouble(char *heading, double &answer, double defdouble);
void remove_terminal_cr(char *s);
void remove_terminal_spaces(char *s);
void calculate_matches(char **s1, char **match, int count, int pos, 
     int matchmode, int identitymode);
int ismatch(char a, char b, int identitymode);
void fill_sequence_strings(char *norm, char *hilight, char **s1, char **match,
     int seqno, int offset, int width);
void read_data(FILE *fp1, char **s1, char **match, int &count, char **label, 
   int &labelcount, int &longest_length);
void print_title(FILE *fp, char **label, int labelcount, int xstart,
     int &ystart, int size, char *lcolor, char *ps_color_string, char *titlefont);
void print_box(FILE *fp, char **label, int count, int labelcount, int startpos,
     int width, int size, int chunkspacesize, int chunk, int xseqstart, int ystart);
void print_label(FILE *fp, char **label, int count, int labelcount, int startpos,
     int width, int size, int chunkspacesize, int chunk, int xseqstart, int ystart,
     char *font);
void print_text(FILE *fp, char **label, int count, int labelcount, int startpos,
     int separatorpos, int width, int size, int chunkspacesize, int chunk, 
     int xseqstart, int ystart, char *font);
void get_sequence_text(char *result, char *label, int offset, int width, 
     int &text_xpos);
int labeltype(char *s, char *&ptr);
int nextarg(char *&ptr);
int between(int a,int b,int c);
int between(double a, double b, double c);

int main(int argc, char **argv)
{
   int k,count,labelcount,page,offset;
   FILE *fp1, *fp2;
   char file1[128], file2[128];
   char **s1;
   char **label;
   char **match;
   char titlefont[64];
   char namefont[64];
   char font[64];
   char ps_color_string[64];
   char fcolor[64],bcolor[64],gcolor[64],hcolor[64],lcolor[64];
   char tempstring[1024];
   char normalstring[1024];
   char matchstring[1024];
   char titlestring[1024];
   char formatstring[1024];
   int matchmode      = 1;
   int size           = 10;
   int wantcolor      = 1;
   int width          = 40;
   int longest_length = 0;
   int chunkspacesize = size;
   int chunksperpage;
   int identitymode   = 1;
   double leftmargin  = 0.5;
   double topmargin   = 1.0;
   int xtranslate;
   int ytranslate;
   int chunk, separatorpos, startpos, xstart, x, y, seqno, xseqstart, ystart;
   s1    = new char*[1000];
   match = new char*[1000];
   label = new char*[1000];

   printf("Highlight - converts alignment file to PostScript, highlighting\n");
   printf("positions that match the controlling sequence(s).              \n");
   printf("See manual for format of alignment files.\n");

   getstring("Alignment file ", file1, "1.txt");
   sprintf(tempstring, "%s.ps", file1);
   getstring("PostScript file", file2, tempstring);
   getint("Controlling sequence(1=1st, 2=1st+2nd, 3=any 4=best)", matchmode, matchmode);
   getint("Enter 1=identity 2=homology", identitymode, identitymode);
   getstring("Font for title        ", titlefont, "Times-Roman");
   getstring("Font for sequence name", namefont, "Times-Roman");
   getstring("Font for sequence     ", font, "Helvetica");
   getint("Char size", size, size);
   getint("Sequence characters per line", width, width);
   getint("Enter 0=B/W 1=color", wantcolor, wantcolor);
   if(wantcolor)
   {
      strcpy(ps_color_string, "setrgbcolor");
      getstring("Normal character RGBcolor     ", fcolor, "0.000 0.000 0.000");
      getstring("Normal background RGBcolor    ", bcolor, "0.950 0.950 0.950");
      getstring("Highlighted character RGBcolor", gcolor, "0.000 0.000 0.000");
      getstring("Highlight background RGBcolor ", hcolor, "1.000 1.000 0.000");
      getstring("Label character RGBcolor      ", lcolor, "0.000 0.000 0.000");
   }else
   {  strcpy(ps_color_string, "setgray");
      getstring("Normal character gray value     ", fcolor, "0.000");
      getstring("Normal background gray value    ", bcolor, "0.950");
      getstring("Highlighted character gray value", gcolor, "1.000");
      getstring("Highlight background gray value ", hcolor, "0.050");
      getstring("Label character gray value      ", lcolor, "0.000");
   }

   if((fp1=fopen(file1,"rb")) == NULL) 
   {  printf("\nUnable to open file xxx%sxxx\n",file1);
      exit(1);
   }
   if((fp2=fopen(file2,"wb")) == NULL) 
   {  printf("\nUnable to open file xxx%sxxx\n",file2);
      exit(1);
   }

   //// Read the sequences
   count = 0;
   labelcount = 0;
   page = 1;
   read_data(fp1, s1, match, count, label, labelcount, longest_length);

   printf("%d sequences counted\n", count);
   getint("Spacing between chunks", chunkspacesize, chunkspacesize);
   ////  684 = 9.5 inches * 72 points/inch
   chunksperpage = 684 / ((size*count) + chunkspacesize);
   getint("Chunks per page       ", chunksperpage, chunksperpage);
   getdouble("Top margin (inches)   ", topmargin, topmargin);
   getdouble("Left margin (inches)  ", leftmargin, leftmargin);
   xtranslate = int(leftmargin * 72);
   ytranslate = int(-topmargin * 72);

 
   //// Create postscript
   fprintf(fp2, "%s\n", "%!PS-Adobe-2.0 EPSF-2.0");
   fprintf(fp2, "%s\n", "%%Title: None");
   fprintf(fp2, "%s %s\n", "%%Creator:", argv[0]);
   fprintf(fp2, "%s\n", "%%BoundingBox: 0 0 596 842");
   fprintf(fp2, "%s\n", "%%Pages: 1");
   fprintf(fp2, "%s\n", "%%DocumentFonts:");
   fprintf(fp2, "%s\n", "%%EndComments");
   fprintf(fp2, "%s\n", "%%EndProlog");

   fprintf(fp2, "%s %d %d\n", "%%Page:", page, page);
   fprintf(fp2, "%s\n", "/origstate save def");
   fprintf(fp2, "%d %d translate\n", xtranslate, ytranslate);

   formatstring[0] = 0;
   sprintf(tempstring, "%d ", size);
   for(k=0;k<width;k++) strcat(formatstring, tempstring);
   if(!strchr(s1[0], '|'))
   {  printf("Error: Sequence name must be separated by sequence by a vertical bar (|)\n");
      exit(EXIT_FAILURE);
   }
   separatorpos = (int)(strchr(s1[0], '|') - s1[0]);       
   calculate_matches(s1, match, count, separatorpos, matchmode, identitymode);

   xstart = 20;
   xseqstart = 0;
   startpos = 0;
   chunk = 0;
   ystart = 800;
   seqno = 0;

   print_title(fp2, label, labelcount, xstart, ystart, size, lcolor,
       ps_color_string, titlefont);

   for(y = ystart; y>0; y-=size)
   {
       ////  Print sequence name
       fprintf(fp2, "%s %s\n", lcolor, ps_color_string);
       fprintf(fp2, "/%s %s\n", namefont, "findfont");
       fprintf(fp2, "%d %s\n",  size,"scalefont setfont");
       strncpy(titlestring, s1[seqno], separatorpos);
       titlestring[separatorpos] = 0;
       fprintf(fp2, "%d %d moveto\n", xstart, y);
       fprintf(fp2, "(%s) show\n", titlestring);
       if(xseqstart==0) xseqstart = xstart + int(0.6 * size * strlen(titlestring));

       //// Print match boxes
       fprintf(fp2, "%s %s\n", hcolor, ps_color_string);
       for(k=startpos+1; k<=startpos+width; k++)
       {  x = xseqstart + (k-startpos-1) * size;
          if(match[seqno][k+separatorpos])
              fprintf(fp2, "%d %d %d %d rectfill\n",x-1,y-1,size,size);
       }

       fprintf(fp2, "%s %s\n", bcolor, ps_color_string);
       for(k=startpos+1; k<=startpos+width; k++)
       {  x = xseqstart + (k-startpos-1) * size;
          if(!match[seqno][k+separatorpos])
              fprintf(fp2, "%d %d %d %d rectfill\n",x-1,y-1,size,size);
       }

       //// Print sequence
       fprintf(fp2, "/%s %s\n", font, "findfont");
       fprintf(fp2, "%d %s\n",  size,"scalefont setfont");
       offset = 1 + separatorpos + startpos;
       fill_sequence_strings(normalstring, matchstring, s1, match, seqno, 
           offset, width); 
       fprintf(fp2, "%d %d moveto\n", xseqstart, y);
       fprintf(fp2, "%s %s\n", fcolor, ps_color_string);
       fprintf(fp2, "(%s) [%s] xshow\n", normalstring, formatstring);

       fprintf(fp2, "%d %d moveto\n", xseqstart, y);
       fprintf(fp2, "%s %s\n", gcolor, ps_color_string);
       fprintf(fp2, "(%s) [%s] xshow\n", matchstring, formatstring);

       seqno++;
       if(seqno>=count) 
       {   seqno=0;
           fprintf(fp2, "%s %s\n", lcolor, ps_color_string);
           ////  Print boxes after each chunk
           print_box(fp2, label, count, labelcount, startpos, width, size, 
                chunkspacesize, chunk, xseqstart, ystart);
           ////  Print labels
           print_label(fp2, label, count, labelcount, startpos, width, size, 
                chunkspacesize, chunk, xseqstart, ystart, namefont);
           ////  Print text
           print_text(fp2, label, count, labelcount, startpos, separatorpos, 
                width, size, chunkspacesize, chunk, xseqstart, ystart, namefont);
           chunk++;
           startpos += width; 
           if(startpos >= (int)strlen(s1[seqno])) break;   
           y -=chunkspacesize;
           if(chunk >= chunksperpage)
           {    chunk=0;
                page++;
                y = ystart + size;
                fprintf(fp2, "%s\n", "showpage");
                fprintf(fp2, "%s\n", "origstate restore");
                fprintf(fp2, "%s %d %d\n", "%%Page:",page,page);
                fprintf(fp2, "%s\n\n", "/origstate save def");
                fprintf(fp2, "%d %d translate\n", xtranslate, ytranslate);
           }
       }
   }

   fprintf(fp2, "%s\n", "showpage");
   fprintf(fp2, "%s\n", "origstate restore");
   fprintf(fp2, "%s\n", "%%Trailer");
 
   fclose(fp1);
   fclose(fp2);

   for(k=0;k<labelcount;k++) delete[] label[k];
   for(k=0;k<count;k++){ delete[] s1[k]; delete[] match[k]; }
   delete[] label;
   delete[] s1;
   delete[] match;
   exit(EXIT_SUCCESS);
}


//-------------------------------------------------------------------------//
// getstring                                                               //
//-------------------------------------------------------------------------//
char *getstring(char *heading, char *string, char *defstring)
{
   printf("%s [%s]:", heading, defstring); fflush(stdout);
   fgets(string, 64, stdin);
   if(string == NULL || string[0]=='\n') strcpy(string, defstring);
   remove_terminal_cr(string);
   return(string);
}

//-------------------------------------------------------------------------//
// getint                                                                  //
//-------------------------------------------------------------------------//
int getint(char *heading, int &answer, int defint)
{
   char string[1024];
   printf("%s [%d]:", heading, defint); fflush(stdout);
   fgets(string, 64, stdin);
   if(string == NULL || string[0]=='\n') sprintf(string, "%d", defint);
   remove_terminal_cr(string);
   answer = atoi(string);
   return atoi(string);
}

//-------------------------------------------------------------------------//
// getdouble                                                               //
//-------------------------------------------------------------------------//
double getdouble(char *heading, double &answer, double defdouble)
{
   char string[1024];
   printf("%s [%g]:", heading, defdouble); fflush(stdout);
   fgets(string, 64, stdin);
   if(string == NULL || string[0]=='\n') sprintf(string, "%g", defdouble);
   remove_terminal_cr(string);
   answer = atof(string);
   return atof(string);
}


//-------------------------------------------------------------------------//
// remove_terminal_cr                                                      //
//-------------------------------------------------------------------------//
void remove_terminal_cr(char *s)
{
    int pos = max(0, (int)(strlen(s))-1);
    if(s[pos]=='\n') s[pos]=0;
}

//-------------------------------------------------------------------------//
// remove_terminal_spaces                                                  //
//-------------------------------------------------------------------------//
void remove_terminal_spaces(char *s)
{
    int k;
    int pos = max(0, (int)(strlen(s))-1);
    for(k=pos; k>=0; k--) if(s[k]==' ') s[k]=0; else break;
}

//-------------------------------------------------------------------------//
// calculate_matches                                                       //
//-------------------------------------------------------------------------//
void calculate_matches(char **s1, char **match, int count, int pos, 
     int matchmode, int im)
{
   int i,j,j2, hit[count], most, highest;
   int len=0;
   for(j=0; j<count; j++) len = max(len, (int)strlen(s1[j]));
   for(j=0; j<count; j++)  
   for(i=0; i<(int)strlen(s1[j]); i++)  
       match[j][i]=0;

   for(i=pos; i<len; i++)       
   {  switch(matchmode)
      {  case 1:  // match 1st sequence only
            for(j=1; j<count; j++)  
               if(ismatch(s1[j][i], s1[0][i], im)){ match[0][i]=1; match[j][i]=1; }
            break;