Exemple #1
0
void
sff_read_print(FILE *fp, sff_read_t *r, sff_header_t *gh, sff_read_header_t *rh)
{
  uint32_t i;

  // flowgram
  for(i=0;i<gh->flow_length;i++) {
      if(0 < i) {
          if(-1L == fputc(',', fp)) {
            ion_error(__func__, "fputc", Exit, WriteFileError);
          }
      }
      if(fprintf(fp, "%u", r->flowgram[i]) < 0) {
          ion_error(__func__, "fprintf", Exit, WriteFileError);
      }
  }
  if(-1L == fputc('\n', fp)) {
      ion_error(__func__, "fputc", Exit, WriteFileError);
  }

  // flow index
  for(i=0;i<rh->n_bases;i++) {
      if(0 < i) {
          if(-1L == fputc(',', fp)) { 
            ion_error(__func__, "fputc", Exit, WriteFileError);
          }
      }
      if(fprintf(fp, "%u", r->flow_index[i]) < 0) { 
        ion_error(__func__, "fprintf", Exit, WriteFileError);
      }
  }
  if(-1L == fputc('\n', fp)) {
      ion_error(__func__, "fputc", Exit, WriteFileError);
  }

  // bases
  if(fprintf(fp, "%s\n", r->bases->s) < 0) { 
    ion_error(__func__, "fprintf", Exit, WriteFileError);
  }

  // quality
  for(i=0;i<r->quality->l;i++) {
    if(EOF == fputc(QUAL2CHAR(r->quality->s[i]), fp)) {
        ion_error(__func__, "fputc", Exit, WriteFileError);
    }
  }
  fputc('\n', fp);
}
Exemple #2
0
tmap_sff_read_t *
tmap_sff_read_read(tmap_file_t *fp, tmap_sff_header_t *gh, tmap_sff_read_header_t *rh)
{
  tmap_sff_read_t *r = NULL;
  uint32_t i, n = 0;

  r = tmap_calloc(1, sizeof(tmap_sff_read_t), "r");

  r->flowgram = tmap_malloc(sizeof(uint16_t)*gh->flow_length, "r->flowgram");
  r->flow_index = tmap_malloc(sizeof(uint8_t)*rh->n_bases, "r->flow_index");

  r->bases = tmap_string_init(rh->n_bases+1);
  r->quality = tmap_string_init(rh->n_bases+1);

  if(gh->flow_length != tmap_file_fread(r->flowgram, sizeof(uint16_t), gh->flow_length, fp)
     || rh->n_bases != tmap_file_fread(r->flow_index, sizeof(uint8_t), rh->n_bases, fp)
     || rh->n_bases != tmap_file_fread(r->bases->s, sizeof(char), rh->n_bases, fp)
     || rh->n_bases != tmap_file_fread(r->quality->s, sizeof(char), rh->n_bases, fp)) {
      tmap_error("tmap_file_fread", Exit, ReadFileError);
  }
  n += sizeof(uint16_t)*gh->flow_length + 3*sizeof(uint8_t)*rh->n_bases;

  // set length and null-terminators
  r->bases->l = rh->n_bases;
  r->quality->l = rh->n_bases;
  r->bases->s[r->bases->l]='\0';
  r->quality->s[r->quality->l]='\0';

  // convert qualities from int to char
  for(i=0;i<r->quality->l;i++) {
      r->quality->s[i] = QUAL2CHAR(r->quality->s[i]);
  }

  // convert flowgram to host order
  for(i=0;i<gh->flow_length;i++) {
      r->flowgram[i] = ntohs(r->flowgram[i]);
  }

  n += tmap_sff_read_padding(fp, n);

#ifdef TMAP_SFF_DEBUG
  tmap_sff_read_print(stderr, r, gh, rh);
#endif

  return r;
}
Exemple #3
0
int32_t
tmap_sam_io_read(tmap_sam_io_t *samio, tmap_sam_t *sam)
{
  if(NULL != sam->b) {
      bam_destroy1(sam->b);
  }
  sam->b = bam_init1();

  // check if we're past optional end bam virtual file offset
  if (samio->bam_end_vfo > 0) {
      BGZF* bgzf_fp = samio->fp->x.bam;
      if (bam_tell(bgzf_fp) >= samio->bam_end_vfo) {
         fprintf(stderr, "stopping at bam virtual file offset %lu\n", samio->bam_end_vfo);
         return -1;
      }
  }

  if(0 < samread(samio->fp, sam->b)) {
      char *str;
      int32_t i, len;

      // name
      str = bam1_qname(sam->b);
      len = strlen(str);
      tmap_sam_io_update_string(&sam->name, str, len);
      sam->name->s[len] = '\0';
      // seq and qual
      len = sam->b->core.l_qseq;
      tmap_sam_io_update_string(&sam->seq, NULL, len);
      tmap_sam_io_update_string(&sam->qual, (char*)bam1_qual(sam->b), len);
      for(i=0;i<len;i++) {
          sam->seq->s[i] = bam_nt16_rev_table[bam1_seqi(bam1_seq(sam->b), i)];
          sam->qual->s[i] = QUAL2CHAR(sam->qual->s[i]);
      }
      sam->seq->s[len] = sam->qual->s[len] = '\0';
      // reverse compliment if necessary
      if((sam->b->core.flag & BAM_FREVERSE)) {
          tmap_sam_reverse_compliment(sam);
      }
      return 1;
  }
  
  return -1;
}
Exemple #4
0
int
sff_view_main(int argc, char *argv[])
{
  int i, c;
  sff_file_t *sff_file_in=NULL, *sff_file_out=NULL;
  sff_iter_t *sff_iter = NULL;
  sff_t *sff = NULL;
  char *fn_names = NULL;
  char **names = NULL;
  int32_t names_num = 0, names_mem = 0;
  int32_t out_mode, min_row, max_row, min_col, max_col;

  out_mode = 0;
  min_row = max_row = min_col = max_col = -1;

  while((c = getopt(argc, argv, "r:c:R:bqh")) >= 0) {
      switch(c) {
        case 'r':
          if(ion_parse_range(optarg, &min_row, &max_row) < 0) {
              ion_error(__func__, "-r : format not recognized", Exit, OutOfRange);
          }
          break;
        case 'c':
          if(ion_parse_range(optarg, &min_col, &max_col) < 0) {
              ion_error(__func__, "-c : format not recognized", Exit, OutOfRange);
          }
          break;
        case 'R':
          free(fn_names);
          fn_names = strdup(optarg); break;
        case 'q':
          out_mode |= 1;
          break;
        case 'b':
          out_mode |= 2;
          break;
        case 'h': 
        default: 
          return usage();
      }
  }
  if(argc != 1+optind) {
      return usage();
  }
  else {
      sff_header_t *header = NULL;
      if(3 == out_mode) {
          ion_error(__func__, "options -b and -q cannot be used together", Exit, CommandLineArgument);
      }

      // open the input SFF
      if(-1 != min_row || -1 != max_row || -1 != min_col || -1 != max_col) {
          sff_file_in = sff_fopen(argv[optind], "rbi", NULL, NULL);
      }
      else {
          sff_file_in = sff_fopen(argv[optind], "rb", NULL, NULL);
      }

      header = sff_header_clone(sff_file_in->header); /* copy header, but update n_reads if using index or names */

      // read in the names
      if(NULL != fn_names) {
          FILE *fp = NULL;
          char name[1024]="\0"; // lets hope we don't exceed this length
          names_num = names_mem = 0;
          names = NULL;
          if(!(fp = fopen(fn_names, "rb"))) {
              fprintf(stderr, "** Could not open %s for reading. **\n", fn_names);
              ion_error(__func__, fn_names, Exit, OpenFileError);
          }
          while(EOF != fscanf(fp, "%s", name)) {
              while(names_num == names_mem) {
                  if(0 == names_mem) names_mem = 4;
                  else names_mem *= 2;
                  names = ion_realloc(names, sizeof(char*) * names_mem, __func__, "names");
              }
              names[names_num] = strdup(name);
              if(NULL == names[names_num]) {
                  ion_error(__func__, name, Exit, MallocMemory);
              }
              names_num++;
          }
          names = ion_realloc(names, sizeof(char*) * names_num, __func__, "names");
          fclose(fp);
          header->n_reads = names_num;
      }
      else {
	// if using index, then iterate once through the index to count the entries
	// so we can set the count correctly in the header
	if (-1 != min_row || -1 != max_row || -1 != min_col || -1 != max_col) {
	  int entries = 0;
          sff_iter = sff_iter_query(sff_file_in, min_row, max_row, min_col, max_col);
	  while (NULL != (sff = sff_iter_read(sff_file_in, sff_iter)))
	    entries++;
	  header->n_reads = entries;
	  /* reset sff_iter */
	  sff_iter_destroy(sff_iter);
	  sff_iter = sff_iter_query(sff_file_in, min_row, max_row, min_col, max_col);
	}
      }

      // print the header
      switch(out_mode) {
        case 0:
          sff_header_print(stdout, header);
          break;
        case 1:
          // do nothing: FASTQ
          break;
        case 2:
          sff_file_out = sff_fdopen(fileno(stdout), "wb", header, NULL);
          break;
      }


      while(1) {
          int32_t to_print = 1;
          if(-1 != min_row || -1 != max_row || -1 != min_col || -1 != max_col) {
              if(NULL == (sff = sff_iter_read(sff_file_in, sff_iter))) {
                  break;
              }
          }
          else {
              if(NULL == (sff = sff_read(sff_file_in))) {
                  break;
              }
          }
          if(0 < names_mem) {
              to_print = 0;
              for(i=0;i<names_num;i++) {
                  if(0 == strcmp(names[i], sff_name(sff))) {
                      to_print = 1;
                      break;
                  }
              }
              // shift down
              if(1 == to_print) { // i < names_num
                  free(names[i]);
                  names[i] = NULL;
                  for(;i<names_num-1;i++) {
                      names[i] = names[i+1];
                      names[i+1] = NULL;
                  }
                  names_num--;
              }
          }
          if(1 == to_print) {
              switch(out_mode) {
                case 0:
                  sff_print(stdout, sff);
                  break;
                case 1:
                  if(fprintf(stdout, "@%s\n%s\n+\n",
                             sff->rheader->name->s,
                             sff->read->bases->s + sff->gheader->key_length) < 0) {
                      ion_error(__func__, "stdout", Exit, WriteFileError);
                  }
                  for(i=sff->gheader->key_length;i<sff->read->quality->l;i++) {
                      if(fputc(QUAL2CHAR(sff->read->quality->s[i]), stdout) < 0) {
                          ion_error(__func__, "stdout", Exit, WriteFileError);
                      }
                  }
                  if(fputc('\n', stdout) < 0) {
                      ion_error(__func__, "stdout", Exit, WriteFileError);
                  }
                  break;
                case 2:
                  sff_write(sff_file_out, sff);
                  break;
              }
          }
          sff_destroy(sff);
      }

      sff_fclose(sff_file_in);
      if(2 == out_mode) {
          sff_fclose(sff_file_out);
      }
      if(-1 != min_row || -1 != max_row || -1 != min_col || -1 != max_col) {
          sff_iter_destroy(sff_iter);
      }

      if(0 != names_num) {
          fprintf(stderr, "** Did not find all the reads with (-R). **\n");
          ion_error(__func__, fn_names, Exit, OutOfRange);
      }

      sff_header_destroy(header);

  }
  if(NULL != names && 0 < names_num) {
      free(names);
  }
  free(fn_names);
  return 0;
}