Exemple #1
0
static bwa_seq_t *bwa_read_bam(bwa_seqio_t *bs, int n_needed, int *n, int is_comp, int trim_qual)
{
	bwa_seq_t *seqs, *p;
	int n_seqs, l, i;
	long n_trimmed = 0, n_tot = 0;
	bam1_t *b;
	int res;

	b = bam_init1();
	n_seqs = 0;
	seqs = (bwa_seq_t*)calloc(n_needed, sizeof(bwa_seq_t));
	while ((res = bam_read1(bs->fp, b)) >= 0) {
		uint8_t *s, *q;
		int go = 0;
		if ((bs->which & 1) && (b->core.flag & BAM_FREAD1)) go = 1;
		if ((bs->which & 2) && (b->core.flag & BAM_FREAD2)) go = 1;
		if ((bs->which & 4) && !(b->core.flag& BAM_FREAD1) && !(b->core.flag& BAM_FREAD2))go = 1;
		if (go == 0) continue;
		l = b->core.l_qseq;
		p = &seqs[n_seqs++];
		p->tid = -1; // no assigned to a thread
		p->qual = 0;
		p->full_len = p->clip_len = p->len = l;
		n_tot += p->full_len;
		s = bam1_seq(b); q = bam1_qual(b);
		p->seq = (ubyte_t*)calloc(p->len + 1, 1);
		p->qual = (ubyte_t*)calloc(p->len + 1, 1);
		for (i = 0; i != p->full_len; ++i) {
			p->seq[i] = bam_nt16_nt4_table[(int)bam1_seqi(s, i)];
			p->qual[i] = q[i] + 33 < 126? q[i] + 33 : 126;
		}
		if (bam1_strand(b)) { // then reverse 
			seq_reverse(p->len, p->seq, 1);
			seq_reverse(p->len, p->qual, 0);
		}
		if (trim_qual >= 1) n_trimmed += bwa_trim_read(trim_qual, p);
		p->rseq = (ubyte_t*)calloc(p->full_len, 1);
		memcpy(p->rseq, p->seq, p->len);
		seq_reverse(p->len, p->seq, 0); // *IMPORTANT*: will be reversed back in bwa_refine_gapped()
		seq_reverse(p->len, p->rseq, is_comp);
		p->name = strdup((const char*)bam1_qname(b));
		if (n_seqs == n_needed) break;
	}
	if (res < 0 && res != -1) err_fatal_simple("Error reading bam file");
	*n = n_seqs;
	if (n_seqs && trim_qual >= 1)
		fprintf(stderr, "[bwa_read_seq] %.1f%% bases are trimmed.\n", 100.0f * n_trimmed/n_tot);
	if (n_seqs == 0) {
		free(seqs);
		bam_destroy1(b);
		return 0;
	}
	bam_destroy1(b);
	return seqs;
}
Exemple #2
0
// Mostly stolen from bwa_read_bam.
void bam1_to_seq(bam1_t *raw, bwa_seq_t *p, int is_comp, int trim_qual)
{
    // long n_trimmed = 0;

    uint8_t *s, *q;
    int i, l = raw->core.l_qseq;
    p->tid = -1; // no assigned to a thread
    p->qual = 0;
    p->full_len = p->clip_len = p->len = l;
    // n_tot += p->full_len;
    s = bam1_seq(raw);
    q = bam1_qual(raw);
    p->seq = (ubyte_t*)calloc(p->len + 1, 1);
    p->qual = (ubyte_t*)calloc(p->len + 1, 1);
    for (i = 0; i != p->full_len; ++i) {
        p->seq[i] = bam_nt16_nt4_table[(int)bam1_seqi(s, i)];
        p->qual[i] = q[i] + 33 < 126? q[i] + 33 : 126;
    }
    if (bam1_strand(raw)) { // then reverse
        seq_reverse(p->len, p->seq, 1);
        seq_reverse(p->len, p->qual, 0);
    }
    if (trim_qual >= 1) /* n_trimmed += */ bwa_trim_read(trim_qual, p);
    p->rseq = (ubyte_t*)calloc(p->full_len, 1);
    memcpy(p->rseq, p->seq, p->len);
    seq_reverse(p->len, p->seq, 0); // *IMPORTANT*: will be reversed back in bwa_refine_gapped()
    seq_reverse(p->len, p->rseq, is_comp);
    p->max_entries = 0 ;

    // We don't set a name, it's contained in the original record
    // anyway.
    // p->name = strdup((const char*)bam1_qname(raw));

    // No place to put the tally right now.
    // if (n_seqs && trim_qual >= 1)
    // fprintf(stderr, "[bwa_read_seq] %.1f%% bases are trimmed.\n", 100.0f * n_trimmed/n_tot);
}
Exemple #3
0
bwa_seq_t *bwa_read_seq(bwa_seqio_t *bs, int n_needed, int *n, int mode, int trim_qual)
{
	bwa_seq_t *seqs, *p;
	kseq_t *seq = bs->ks;
	int n_seqs, l, i, is_comp = mode&BWA_MODE_COMPREAD, is_64 = mode&BWA_MODE_IL13, l_bc = mode>>24;
	long n_trimmed = 0, n_tot = 0;

	if (l_bc > BWA_MAX_BCLEN) {
		fprintf(stderr, "[%s] the maximum barcode length is %d.\n", __func__, BWA_MAX_BCLEN);
		return 0;
	}
	if (bs->is_bam) return bwa_read_bam(bs, n_needed, n, is_comp, trim_qual); // l_bc has no effect for BAM input
	n_seqs = 0;
	seqs = (bwa_seq_t*)calloc(n_needed, sizeof(bwa_seq_t));
	while ((l = kseq_read(seq)) >= 0) {
		if ((mode & BWA_MODE_CFY) && (seq->comment.l != 0)) {
			// skip reads that are marked to be filtered by Casava
			char *s = index(seq->comment.s, ':');
			if (s && *(++s) == 'Y') {
				continue;
			}
		}
		if (is_64 && seq->qual.l)
			for (i = 0; i < seq->qual.l; ++i) seq->qual.s[i] -= 31;
		if (seq->seq.l <= l_bc) continue; // sequence length equals or smaller than the barcode length
		p = &seqs[n_seqs++];
		if (l_bc) { // then trim barcode
			for (i = 0; i < l_bc; ++i)
				p->bc[i] = (seq->qual.l && seq->qual.s[i]-33 < BARCODE_LOW_QUAL)? tolower(seq->seq.s[i]) : toupper(seq->seq.s[i]);
			p->bc[i] = 0;
			for (; i < seq->seq.l; ++i)
				seq->seq.s[i - l_bc] = seq->seq.s[i];
			seq->seq.l -= l_bc; seq->seq.s[seq->seq.l] = 0;
			if (seq->qual.l) {
				for (i = l_bc; i < seq->qual.l; ++i)
					seq->qual.s[i - l_bc] = seq->qual.s[i];
				seq->qual.l -= l_bc; seq->qual.s[seq->qual.l] = 0;
			}
			l = seq->seq.l;
		} else p->bc[0] = 0;
		p->tid = -1; // no assigned to a thread
		p->qual = 0;
		p->full_len = p->clip_len = p->len = l;
		n_tot += p->full_len;
		p->seq = (ubyte_t*)calloc(p->len, 1);
		for (i = 0; i != p->full_len; ++i)
			p->seq[i] = nst_nt4_table[(int)seq->seq.s[i]];
		if (seq->qual.l) { // copy quality
			p->qual = (ubyte_t*)strdup((char*)seq->qual.s);
			if (trim_qual >= 1) n_trimmed += bwa_trim_read(trim_qual, p);
		}
		p->rseq = (ubyte_t*)calloc(p->full_len, 1);
		memcpy(p->rseq, p->seq, p->len);
		seq_reverse(p->len, p->seq, 0); // *IMPORTANT*: will be reversed back in bwa_refine_gapped()
		seq_reverse(p->len, p->rseq, is_comp);
		p->name = strdup((const char*)seq->name.s);
		{ // trim /[12]$
			int t = strlen(p->name);
			if (t > 2 && p->name[t-2] == '/' && (p->name[t-1] == '1' || p->name[t-1] == '2')) p->name[t-2] = '\0';
		}
		if (n_seqs == n_needed) break;
	}
	*n = n_seqs;
	if (n_seqs && trim_qual >= 1)
		fprintf(stderr, "[bwa_read_seq] %.1f%% bases are trimmed. \n", 100.0f * n_trimmed/n_tot);
	if (n_seqs == 0) {
		free(seqs);
		return 0;
	}
	return seqs;
}
Exemple #4
0
int bwa_read_seq1(bwa_seqio_t *bs, int iter, int tid, int thrds, bwa_seq_t **_seqs, int *n_avail, int mode, int trim_qual)
{
	bwa_seq_t *p;
   bwa_seq_t *seqs = *_seqs;
	kseq_t *seq = bs->ks;
	int n_seqs, l, i, is_comp = mode&BWA_MODE_COMPREAD, is_64 = mode&BWA_MODE_IL13, l_bc = mode>>24;
	long n_trimmed = 0, n_tot = 0;
   bool first;

	if (l_bc > BWA_MAX_BCLEN) {
		fprintf(stderr, "[%s] the maximum barcode length is %d.\n", __func__, BWA_MAX_BCLEN);
		return 0;
	}
	if (bs->is_bam) {
      fprintf (stderr, "IS BAM! --- Port bwa_read_bam function\n");
      bwa_free_read_seq(*n_avail, seqs);
      // return bwa_read_bam(bs, n_needed, n, is_comp, trim_qual); // l_bc has no effect for BAM input
      exit(0);
   }
   if (*n_avail == 0) {
      if (*n_avail) bwa_free_read_seq(*n_avail, seqs);
	   seqs = (bwa_seq_t*)calloc(READ_SEQ_SIZE, sizeof(bwa_seq_t));
      *_seqs = seqs;
      *n_avail = READ_SEQ_SIZE;
   }
	n_seqs = 0;
   first = true;
//err_fwrite("FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF", strlen("FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF"), 1, stdout);
//long cksm = 0;
	while ((l = kseq_read1(seq, iter, tid, thrds, &first)) >= 0) {
		if ((mode & BWA_MODE_CFY) && (seq->comment.l != 0)) {
			// skip reads that are marked to be filtered by Casava
			char *s = index(seq->comment.s, ':');
			if (s && *(++s) == 'Y') {
				continue;
			}
		}
		if (is_64 && seq->qual.l)
			for (i = 0; i < seq->qual.l; ++i) seq->qual.s[i] -= 31;
		if (seq->seq.l <= l_bc) continue; // sequence length equals or smaller than the barcode length
		p = &seqs[n_seqs++];
      if(n_seqs > READ_SEQ_SIZE) {
         fprintf (stderr, "READ_SEQ_SIZE not big enough\n");
         abort();
      }
      init_bwa_seq_t(p);
		if (l_bc) { // then trim barcode
			for (i = 0; i < l_bc; ++i)
				p->bc[i] = (seq->qual.l && seq->qual.s[i]-33 < BARCODE_LOW_QUAL)? tolower(seq->seq.s[i]) : toupper(seq->seq.s[i]);
			p->bc[i] = 0;
			for (; i < seq->seq.l; ++i)
				seq->seq.s[i - l_bc] = seq->seq.s[i];
			seq->seq.l -= l_bc; seq->seq.s[seq->seq.l] = 0;
			if (seq->qual.l) {
				for (i = l_bc; i < seq->qual.l; ++i)
					seq->qual.s[i - l_bc] = seq->qual.s[i];
				seq->qual.l -= l_bc; seq->qual.s[seq->qual.l] = 0;
			}
			l = seq->seq.l;
		} else p->bc[0] = 0;
//ComputeChecksum(seq->seq.s,seq->seq.l,&cksm);
		p->tid = -1; // no assigned to a thread
		p->full_len = p->clip_len = p->len = l;
		n_tot += p->full_len;
      if (p->llen < p->len) {
         if(p->seq) free(p->seq);
         p->llen = p->len;
		   p->seq = (ubyte_t*)calloc(p->len, 1);
		   p->rseq = (ubyte_t*)calloc(p->full_len, 1);
      }
		for (i = 0; i != p->full_len; ++i)
			p->seq[i] = nst_nt4_table[(int)seq->seq.s[i]];
		if (seq->qual.l) { // copy quality
         if(p->qual) free(p->qual);
			p->qual = (ubyte_t*)strdup((char*)seq->qual.s);
			if (trim_qual >= 1) n_trimmed += bwa_trim_read(trim_qual, p);
		}
		memcpy(p->rseq, p->seq, p->len);
		seq_reverse(p->len, p->seq, 0); // *IMPORTANT*: will be reversed back in bwa_refine_gapped()
		seq_reverse(p->len, p->rseq, is_comp);
      if(p->name) free(p->name);
		p->name = strdup((const char*)seq->name.s);
		{ // trim /[12]$
			int t = strlen(p->name);
			if (t > 2 && p->name[t-2] == '/' && (p->name[t-1] == '1' || p->name[t-1] == '2')) p->name[t-2] = '\0';
		}
		//if (n_seqs == n_needed) break;
		if(kseq_end(seq)) break;
	}
	if (n_seqs && trim_qual >= 1)
		fprintf(stderr, "[bwa_read_seq1] %.1f%% bases are trimmed.\n", 100.0f * n_trimmed/n_tot);
//fprintf(stderr, "%d tid: %d sequences: %d cksum: %lx\n", iter, tid, n_seqs, cksm);
	return n_seqs;
}
Exemple #5
0
static bwa_seq_t *bwa_read_bam(bwa_seqio_t *bs, int n_needed, int *n, int is_comp, int trim_qual)
{
	bwa_seq_t *seqs, *p;
	int n_seqs, l, i;
	long n_trimmed = 0, n_tot = 0;
	bam1_t *b;
	int res;

	b = bam_init1();
	n_seqs = 0;
	seqs = (bwa_seq_t*)calloc(n_needed, sizeof(bwa_seq_t));
#ifdef USE_HTSLIB
	while ((res = sam_read1(bs->fp, bs->h, b)) >= 0) {
#else
	while ((res = bam_read1(bs->fp, b)) >= 0) {
#endif
		uint8_t *s, *q;
		int go = 0;
		if ((bs->which & 1) && (b->core.flag & BAM_FREAD1)) go = 1;
		if ((bs->which & 2) && (b->core.flag & BAM_FREAD2)) go = 1;
		if ((bs->which & 4) && !(b->core.flag& BAM_FREAD1) && !(b->core.flag& BAM_FREAD2))go = 1;
		if (go == 0) continue;
		l = b->core.l_qseq;
		p = &seqs[n_seqs++];
		p->tid = -1; // no assigned to a thread
		p->qual = 0;
		p->full_len = p->clip_len = p->len = l;
		n_tot += p->full_len;
#ifdef USE_HTSLIB
		s = bam_get_seq(b); q = bam_get_qual(b);
#else
		s = bam1_seq(b); q = bam1_qual(b);
#endif
		p->seq = (ubyte_t*)calloc(p->len + 1, 1);
		p->qual = (ubyte_t*)calloc(p->len + 1, 1);
		for (i = 0; i != p->full_len; ++i) {
#ifdef USE_HTSLIB
			p->seq[i] = bam_nt16_nt4_table[(int)bam_seqi(s, i)];
#else
			p->seq[i] = bam_nt16_nt4_table[(int)bam1_seqi(s, i)];
#endif
			p->qual[i] = q[i] + 33 < 126? q[i] + 33 : 126;
		}
#ifdef USE_HTSLIB
		if (bam_is_rev(b)) { // then reverse 
#else
		if (bam1_strand(b)) { // then reverse 
#endif
			seq_reverse(p->len, p->seq, 1);
			seq_reverse(p->len, p->qual, 0);
		}
		if (trim_qual >= 1) n_trimmed += bwa_trim_read(trim_qual, p);
		p->rseq = (ubyte_t*)calloc(p->full_len, 1);
		memcpy(p->rseq, p->seq, p->len);
		seq_reverse(p->len, p->seq, 0); // *IMPORTANT*: will be reversed back in bwa_refine_gapped()
		seq_reverse(p->len, p->rseq, is_comp);
#ifdef USE_HTSLIB
		p->name = strdup((const char*)bam_get_qname(b));
#else
		p->name = strdup((const char*)bam1_qname(b));
#endif
		if (n_seqs == n_needed) break;
	}
	if (res < 0 && res != -1) err_fatal_simple("Error reading bam file");
	*n = n_seqs;
	if (n_seqs && trim_qual >= 1)
		fprintf(stderr, "[bwa_read_seq] %.1f%% bases are trimmed.\n", 100.0f * n_trimmed/n_tot);
	if (n_seqs == 0) {
		free(seqs);
		bam_destroy1(b);
		return 0;
	}
	bam_destroy1(b);
	return seqs;
}

#define BARCODE_LOW_QUAL 13

bwa_seq_t *bwa_read_seq(bwa_seqio_t *bs, int n_needed, int *n, int mode, int trim_qual)
{
	bwa_seq_t *seqs, *p;
	kseq_t *seq = bs->ks;
	int n_seqs, l, i, is_comp = mode&BWA_MODE_COMPREAD, is_64 = mode&BWA_MODE_IL13, l_bc = mode>>24;
	long n_trimmed = 0, n_tot = 0;

	if (l_bc > BWA_MAX_BCLEN) {
		fprintf(stderr, "[%s] the maximum barcode length is %d.\n", __func__, BWA_MAX_BCLEN);
		return 0;
	}
	if (bs->is_bam) return bwa_read_bam(bs, n_needed, n, is_comp, trim_qual); // l_bc has no effect for BAM input
	n_seqs = 0;
	seqs = (bwa_seq_t*)calloc(n_needed, sizeof(bwa_seq_t));
	while ((l = kseq_read(seq)) >= 0) {
		if ((mode & BWA_MODE_CFY) && (seq->comment.l != 0)) {
			// skip reads that are marked to be filtered by Casava
			char *s = index(seq->comment.s, ':');
			if (s && *(++s) == 'Y') {
				continue;
			}
		}
		if (is_64 && seq->qual.l)
			for (i = 0; i < seq->qual.l; ++i) seq->qual.s[i] -= 31;
		if (seq->seq.l <= l_bc) continue; // sequence length equals or smaller than the barcode length
		p = &seqs[n_seqs++];
		if (l_bc) { // then trim barcode
			for (i = 0; i < l_bc; ++i)
				p->bc[i] = (seq->qual.l && seq->qual.s[i]-33 < BARCODE_LOW_QUAL)? tolower(seq->seq.s[i]) : toupper(seq->seq.s[i]);
			p->bc[i] = 0;
			for (; i < seq->seq.l; ++i)
				seq->seq.s[i - l_bc] = seq->seq.s[i];
			seq->seq.l -= l_bc; seq->seq.s[seq->seq.l] = 0;
			if (seq->qual.l) {
				for (i = l_bc; i < seq->qual.l; ++i)
					seq->qual.s[i - l_bc] = seq->qual.s[i];
				seq->qual.l -= l_bc; seq->qual.s[seq->qual.l] = 0;
			}
			l = seq->seq.l;
		} else p->bc[0] = 0;
		p->tid = -1; // no assigned to a thread
		p->qual = 0;
		p->full_len = p->clip_len = p->len = l;
		n_tot += p->full_len;
		p->seq = (ubyte_t*)calloc(p->full_len, 1);
		for (i = 0; i != p->full_len; ++i)
			p->seq[i] = nst_nt4_table[(int)seq->seq.s[i]];
		if (seq->qual.l) { // copy quality
			p->qual = (ubyte_t*)strdup((char*)seq->qual.s);
			if (trim_qual >= 1) n_trimmed += bwa_trim_read(trim_qual, p);
		}
		p->rseq = (ubyte_t*)calloc(p->full_len, 1);
		memcpy(p->rseq, p->seq, p->len);
		seq_reverse(p->len, p->seq, 0); // *IMPORTANT*: will be reversed back in bwa_refine_gapped()
		seq_reverse(p->len, p->rseq, is_comp);
		p->name = strdup((const char*)seq->name.s);
		{ // trim /[12]$
			int t = strlen(p->name);
			if (t > 2 && p->name[t-2] == '/' && (p->name[t-1] == '1' || p->name[t-1] == '2')) p->name[t-2] = '\0';
		}
		if (n_seqs == n_needed) break;
	}
	*n = n_seqs;
	if (n_seqs && trim_qual >= 1)
		fprintf(stderr, "[bwa_read_seq] %.1f%% bases are trimmed.\n", 100.0f * n_trimmed/n_tot);
	if (n_seqs == 0) {
		free(seqs);
		return 0;
	}
	return seqs;
}

void bwa_free_read_seq(int n_seqs, bwa_seq_t *seqs)
{
	int i, j;
	for (i = 0; i != n_seqs; ++i) {
		bwa_seq_t *p = seqs + i;
		for (j = 0; j < p->n_multi; ++j)
			if (p->multi[j].cigar) free(p->multi[j].cigar);
		free(p->name);
		free(p->seq); free(p->rseq); free(p->qual); free(p->aln); free(p->md); free(p->multi);
		free(p->cigar);
	}
	free(seqs);
}