r1187: set stage for the ds tag; not added yet

This commit is contained in:
Heng Li
2024-03-10 14:12:56 -04:00
parent a12673611f
commit 23d2674c39
3 changed files with 14 additions and 11 deletions

View File

@@ -139,10 +139,10 @@ int mm_write_sam_hdr(const mm_idx_t *idx, const char *rg, const char *ver, int a
return ret;
}
static void write_cs_core(kstring_t *s, const uint8_t *tseq, const uint8_t *qseq, const mm_reg1_t *r, char *tmp, int no_iden, int write_tag)
static void write_cs_ds_core(kstring_t *s, const uint8_t *tseq, const uint8_t *qseq, const mm_reg1_t *r, char *tmp, int no_iden, int is_ds, int write_tag)
{
int i, q_off, t_off;
if (write_tag) mm_sprintf_lite(s, "\tcs:Z:");
if (write_tag) mm_sprintf_lite(s, "\t%cs:Z:", is_ds? 'd' : 'c');
for (i = q_off = t_off = 0; i < (int)r->p->n_cigar; ++i) {
int j, op = r->p->cigar[i]&0xf, len = r->p->cigar[i]>>4;
assert((op >= MM_CIGAR_MATCH && op <= MM_CIGAR_N_SKIP) || op == MM_CIGAR_EQ_MATCH || op == MM_CIGAR_X_MISMATCH);
@@ -218,7 +218,7 @@ static void write_MD_core(kstring_t *s, const uint8_t *tseq, const uint8_t *qseq
assert(t_off == r->re - r->rs && q_off == r->qe - r->qs);
}
static void write_cs_or_MD(void *km, kstring_t *s, const mm_idx_t *mi, const mm_bseq1_t *t, const mm_reg1_t *r, int no_iden, int is_MD, int write_tag, int is_qstrand)
static void write_cs_ds_or_MD(void *km, kstring_t *s, const mm_idx_t *mi, const mm_bseq1_t *t, const mm_reg1_t *r, int no_iden, int is_MD, int is_ds, int write_tag, int is_qstrand)
{
extern unsigned char seq_nt4_table[256];
int i;
@@ -244,8 +244,8 @@ static void write_cs_or_MD(void *km, kstring_t *s, const mm_idx_t *mi, const mm_
}
}
}
if (is_MD) write_MD_core(s, tseq, qseq, r, tmp, write_tag);
else write_cs_core(s, tseq, qseq, r, tmp, no_iden, write_tag);
if (is_MD == 1) write_MD_core(s, tseq, qseq, r, tmp, write_tag);
else write_cs_ds_core(s, tseq, qseq, r, tmp, no_iden, is_ds, write_tag);
kfree(km, qseq); kfree(km, tseq); kfree(km, tmp);
}
@@ -256,7 +256,7 @@ int mm_gen_cs_or_MD(void *km, char **buf, int *max_len, const mm_idx_t *mi, cons
str.s = *buf, str.l = 0, str.m = *max_len;
t.l_seq = strlen(seq);
t.seq = (char*)seq;
write_cs_or_MD(km, &str, mi, &t, r, no_iden, is_MD, 0, is_qstrand);
write_cs_ds_or_MD(km, &str, mi, &t, r, no_iden, is_MD, 0, 0, is_qstrand);
*max_len = str.m;
*buf = str.s;
return str.l;
@@ -326,8 +326,8 @@ void mm_write_paf3(kstring_t *s, const mm_idx_t *mi, const mm_bseq1_t *t, const
for (k = 0; k < r->p->n_cigar; ++k)
mm_sprintf_lite(s, "%d%c", r->p->cigar[k]>>4, MM_CIGAR_STR[r->p->cigar[k]&0xf]);
}
if (r->p && (opt_flag & (MM_F_OUT_CS|MM_F_OUT_MD)))
write_cs_or_MD(km, s, mi, t, r, !(opt_flag&MM_F_OUT_CS_LONG), opt_flag&MM_F_OUT_MD, 1, !!(opt_flag&MM_F_QSTRAND));
if (r->p && (opt_flag & (MM_F_OUT_CS|MM_F_OUT_DS|MM_F_OUT_MD)))
write_cs_ds_or_MD(km, s, mi, t, r, !(opt_flag&MM_F_OUT_CS_LONG), opt_flag&MM_F_OUT_MD, !!(opt_flag&MM_F_OUT_DS), 1, !!(opt_flag&MM_F_QSTRAND));
if ((opt_flag & MM_F_COPY_COMMENT) && t->comment)
mm_sprintf_lite(s, "\t%s", t->comment);
}
@@ -535,8 +535,8 @@ void mm_write_sam3(kstring_t *s, const mm_idx_t *mi, const mm_bseq1_t *t, int se
}
}
}
if (r->p && (opt_flag & (MM_F_OUT_CS|MM_F_OUT_MD)))
write_cs_or_MD(km, s, mi, t, r, !(opt_flag&MM_F_OUT_CS_LONG), opt_flag&MM_F_OUT_MD, 1, 0);
if (r->p && (opt_flag & (MM_F_OUT_CS|MM_F_OUT_DS|MM_F_OUT_MD)))
write_cs_ds_or_MD(km, s, mi, t, r, !(opt_flag&MM_F_OUT_CS_LONG), opt_flag&MM_F_OUT_MD, !!(opt_flag&MM_F_OUT_DS), 1, 0);
if (cigar_in_tag)
write_sam_cigar(s, flag, 1, t->l_seq, r, opt_flag);
}

2
main.c
View File

@@ -77,6 +77,7 @@ static ko_longopt_t long_options[] = {
{ "print-chains", ko_no_argument, 352 },
{ "no-hash-name", ko_no_argument, 353 },
{ "secondary-seq", ko_no_argument, 354 },
{ "ds", ko_no_argument, 355 },
{ "help", ko_no_argument, 'h' },
{ "max-intron-len", ko_required_argument, 'G' },
{ "version", ko_no_argument, 'V' },
@@ -243,6 +244,7 @@ int main(int argc, char *argv[])
else if (c == 352) mm_dbg_flag |= MM_DBG_PRINT_CHAIN; // --print-chains
else if (c == 353) opt.flag |= MM_F_NO_HASH_NAME; // --no-hash-name
else if (c == 354) opt.flag |= MM_F_SECONDARY_SEQ; // --secondary-seq
else if (c == 355) opt.flag |= MM_F_OUT_DS; // --ds
else if (c == 330) {
fprintf(stderr, "[WARNING] \033[1;31m --lj-min-ratio has been deprecated.\033[0m\n");
} else if (c == 314) { // --frag

View File

@@ -5,7 +5,7 @@
#include <stdio.h>
#include <sys/types.h>
#define MM_VERSION "2.26-r1183-dirty"
#define MM_VERSION "2.26-r1187-dirty"
#define MM_F_NO_DIAG (0x001LL) // no exact diagonal hit
#define MM_F_NO_DUAL (0x002LL) // skip pairs where query name is lexicographically larger than target name
@@ -44,6 +44,7 @@
#define MM_F_NO_HASH_NAME (0x400000000LL)
#define MM_F_SPLICE_OLD (0x800000000LL)
#define MM_F_SECONDARY_SEQ (0x1000000000LL) //output SEQ field for seqondary alignments using hard clipping
#define MM_F_OUT_DS (0x2000000000LL)
#define MM_I_HPC 0x1
#define MM_I_NO_SEQ 0x2