// Returns 0 on success, -1 on failure.
static int sync_mq_mc(bam1_t* src, bam1_t* dest)
{
if ( (src->core.flag & BAM_FUNMAP) == 0 ) { // If mapped
// Copy Mate Mapping Quality
uint32_t mq = src->core.qual;
uint8_t* data;
if ((data = bam_aux_get(dest,"MQ")) != NULL) {
bam_aux_del(dest, data);
}
bam_aux_append(dest, "MQ", 'i', sizeof(uint32_t), (uint8_t*)&mq);
}
// Copy mate cigar if either read is mapped
if ( (src->core.flag & BAM_FUNMAP) == 0 || (dest->core.flag & BAM_FUNMAP) == 0 ) {
uint8_t* data_mc;
if ((data_mc = bam_aux_get(dest,"MC")) != NULL) {
bam_aux_del(dest, data_mc);
}
// Convert cigar to string
kstring_t mc = { 0, 0, NULL };
if (bam_format_cigar(src, &mc) < 0) return -1;
bam_aux_append(dest, "MC", 'Z', ks_len(&mc)+1, (uint8_t*)ks_str(&mc));
free(mc.s);
}
return 0;
}
/*
* This function calculates ct tag for two bams, it assumes they are from the same template and
* writes the tag to the first read in position terms.
*/
static void bam_template_cigar(bam1_t *b1, bam1_t *b2, kstring_t *str)
{
bam1_t *swap;
int i, end;
uint32_t *cigar;
str->l = 0;
if (b1->core.tid != b2->core.tid || b1->core.tid < 0 || b1->core.pos < 0 || b2->core.pos < 0 || b1->core.flag&BAM_FUNMAP || b2->core.flag&BAM_FUNMAP) return; // coordinateless or not on the same chr; skip
if (b1->core.pos > b2->core.pos) swap = b1, b1 = b2, b2 = swap; // make sure b1 has a smaller coordinate
kputc((b1->core.flag & BAM_FREAD1)? '1' : '2', str); // segment index
kputc((b1->core.flag & BAM_FREVERSE)? 'R' : 'F', str); // strand
for (i = 0, cigar = bam_get_cigar(b1); i < b1->core.n_cigar; ++i) {
kputw(bam_cigar_oplen(cigar[i]), str);
kputc(bam_cigar_opchr(cigar[i]), str);
}
end = bam_endpos(b1);
kputw(b2->core.pos - end, str);
kputc('T', str);
kputc((b2->core.flag & BAM_FREAD1)? '1' : '2', str); // segment index
kputc((b2->core.flag & BAM_FREVERSE)? 'R' : 'F', str); // strand
for (i = 0, cigar = bam_get_cigar(b2); i < b2->core.n_cigar; ++i) {
kputw(bam_cigar_oplen(cigar[i]), str);
kputc(bam_cigar_opchr(cigar[i]), str);
}
uint8_t* data;
if ((data = bam_aux_get(b1,"ct")) != NULL) bam_aux_del(b1, data);
if ((data = bam_aux_get(b2,"ct")) != NULL) bam_aux_del(b2, data);
bam_aux_append(b1, "ct", 'Z', str->l+1, (uint8_t*)str->s);
}
static void sync_mq(bam1_t* src, bam1_t* dest)
{
if ( (src->core.flag & BAM_FUNMAP) == 0 ) { // If mapped
uint32_t mq = src->core.qual;
uint8_t* data;
if ((data = bam_aux_get(dest,"MQ")) != NULL) {
bam_aux_del(dest, data);
}
bam_aux_append(dest, "MQ", 'i', sizeof(uint32_t), (uint8_t*)&mq);
}
}
static void overwrite_all_func(const state_t* state, bam1_t* file_read)
{
uint8_t* data = (uint8_t*)strdup(state->rg_id);
int len = strlen(state->rg_id)+1;
// If the old exists delete it
uint8_t* old = bam_aux_get(file_read, "RG");
if (old != NULL) {
bam_aux_del(file_read, old);
}
bam_aux_append(file_read, "RG", 'Z', len, data);
free(data);
}
static int uniform_fetch_func(bam1_t *b, void *data)
{
uint8_t *to_delete;
data_t_uniform *tmp = (data_t_uniform*)data;
bam1_core_t *c = &b->core;
char *iq;
char *dq;
iq = malloc((c->l_qseq+1) * sizeof(char));
memset(iq, tmp->iq, c->l_qseq);
iq[c->l_qseq] = '\0';
to_delete = bam_aux_get(b, BI_TAG);
if (to_delete) {
bam_aux_del(b, to_delete);
}
bam_aux_append(b, BI_TAG, 'Z', c->l_qseq+1, (uint8_t*) iq);
dq = malloc((c->l_qseq+1) * sizeof(char));
memset(dq, tmp->dq, c->l_qseq);
dq[c->l_qseq] = '\0';
to_delete = bam_aux_get(b, BD_TAG);
if (to_delete) {
bam_aux_del(b, to_delete);
}
bam_aux_append(b, BD_TAG, 'Z', c->l_qseq+1, (uint8_t*) dq);
bam_write1(tmp->out, b);
free(iq);
free(dq);
return 0;
}
static void bam_translate(bam1_t* b, trans_tbl_t* tbl)
{
// Update target id if not unmapped tid
if ( b->core.tid >= 0 ) { b->core.tid = tbl->tid_trans[b->core.tid]; }
if ( b->core.mtid >= 0 ) { b->core.mtid = tbl->tid_trans[b->core.mtid]; }
// If we have a RG update it
uint8_t *rg = bam_aux_get(b, "RG");
if (rg) {
char* decoded_rg = bam_aux2Z(rg);
khiter_t k = kh_get(c2c, tbl->rg_trans, decoded_rg);
if (k != kh_end(tbl->rg_trans)) {
char* translate_rg = kh_value(tbl->rg_trans,k);
bam_aux_del(b, rg);
bam_aux_append(b, "RG", 'Z', strlen(translate_rg) + 1, (uint8_t*)translate_rg);
} else {
fprintf(pysamerr, "[bam_translate] RG tag \"%s\" on read \"%s\" encountered with no corresponding entry in header, tag lost\n",decoded_rg, bam_get_qname(b));
bam_aux_del(b, rg);
}
}
// If we have a PG update it
uint8_t *pg = bam_aux_get(b, "PG");
if (pg) {
char* decoded_pg = bam_aux2Z(pg);
khiter_t k = kh_get(c2c, tbl->pg_trans, decoded_pg);
if (k != kh_end(tbl->pg_trans)) {
char* translate_pg = kh_value(tbl->pg_trans,k);
bam_aux_del(b, pg);
bam_aux_append(b, "PG", 'Z', strlen(translate_pg) + 1, (uint8_t*)translate_pg);
} else {
fprintf(pysamerr, "[bam_translate] PG tag \"%s\" on read \"%s\" encountered with no corresponding entry in header, tag lost\n",decoded_pg, bam_get_qname(b));
bam_aux_del(b, pg);
}
}
}
static int add_mate_score(bam1_t *src, bam1_t *dest)
{
uint8_t *data_ms;
uint32_t mate_score = calc_mate_score(src);
if ((data_ms = bam_aux_get(dest, "ms")) != NULL) {
bam_aux_del(dest, data_ms);
}
if (bam_aux_append(dest, "ms", 'i', sizeof(uint32_t), (uint8_t*)&mate_score) == -1) {
return -1;
}
return 0;
}
// Returns 0 to indicate read should be output 1 otherwise
static int process_aln(const bam_hdr_t *h, bam1_t *b, samview_settings_t* settings)
{
if (settings->remove_B) bam_remove_B(b);
if (settings->min_qlen > 0) {
int k, qlen = 0;
uint32_t *cigar = bam_get_cigar(b);
for (k = 0; k < b->core.n_cigar; ++k)
if ((bam_cigar_type(bam_cigar_op(cigar[k]))&1) || bam_cigar_op(cigar[k]) == BAM_CHARD_CLIP)
qlen += bam_cigar_oplen(cigar[k]);
if (qlen < settings->min_qlen) return 1;
}
if (b->core.qual < settings->min_mapQ || ((b->core.flag & settings->flag_on) != settings->flag_on) || (b->core.flag & settings->flag_off))
return 1;
if (settings->bed && (b->core.tid < 0 || !bed_overlap(settings->bed, h->target_name[b->core.tid], b->core.pos, bam_endpos(b))))
return 1;
if (settings->subsam_frac > 0.) {
uint32_t k = __ac_Wang_hash(__ac_X31_hash_string(bam_get_qname(b)) ^ settings->subsam_seed);
if ((double)(k&0xffffff) / 0x1000000 >= settings->subsam_frac) return 1;
}
if (settings->rghash) {
uint8_t *s = bam_aux_get(b, "RG");
if (s) {
khint_t k = kh_get(rg, settings->rghash, (char*)(s + 1));
if (k == kh_end(settings->rghash)) return 1;
}
}
if (settings->library) {
const char *p = bam_get_library((bam_hdr_t*)h, b);
if (!p || strcmp(p, settings->library) != 0) return 1;
}
if (settings->remove_aux_len) {
size_t i;
for (i = 0; i < settings->remove_aux_len; ++i) {
uint8_t *s = bam_aux_get(b, settings->remove_aux[i]);
if (s) {
bam_aux_del(b, s);
}
}
}
return 0;
}
int bam_merge_core2(int by_qname, const char *out, const char *headers, int n, char * const *fn, int flag, const char *reg, int level)
#endif
{
bamFile fpout, *fp;
heap1_t *heap;
bam_header_t *hout = 0;
bam_header_t *hheaders = NULL;
int i, j, *RG_len = 0;
uint64_t idx = 0;
char **RG = 0, mode[8];
bam_iter_t *iter = 0;
if (headers) {
tamFile fpheaders = sam_open(headers);
if (fpheaders == 0) {
const char *message = strerror(errno);
fprintf(stderr, "[bam_merge_core] cannot open '%s': %s\n", headers, message);
return -1;
}
hheaders = sam_header_read(fpheaders);
sam_close(fpheaders);
}
g_is_by_qname = by_qname;
fp = (bamFile*)calloc(n, sizeof(bamFile));
heap = (heap1_t*)calloc(n, sizeof(heap1_t));
iter = (bam_iter_t*)calloc(n, sizeof(bam_iter_t));
// prepare RG tag
if (flag & MERGE_RG) {
RG = (char**)calloc(n, sizeof(void*));
RG_len = (int*)calloc(n, sizeof(int));
for (i = 0; i != n; ++i) {
int l = strlen(fn[i]);
const char *s = fn[i];
if (l > 4 && strcmp(s + l - 4, ".bam") == 0) l -= 4;
for (j = l - 1; j >= 0; --j) if (s[j] == '/') break;
++j; l -= j;
RG[i] = calloc(l + 1, 1);
RG_len[i] = l;
strncpy(RG[i], s + j, l);
}
}
// read the first
for (i = 0; i != n; ++i) {
bam_header_t *hin;
fp[i] = bam_open(fn[i], "r");
if (fp[i] == 0) {
int j;
fprintf(stderr, "[bam_merge_core] fail to open file %s\n", fn[i]);
for (j = 0; j < i; ++j) bam_close(fp[j]);
free(fp); free(heap);
// FIXME: possible memory leak
return -1;
}
hin = bam_header_read(fp[i]);
if (i == 0) { // the first BAM
hout = hin;
} else { // validate multiple baf
int min_n_targets = hout->n_targets;
if (hin->n_targets < min_n_targets) min_n_targets = hin->n_targets;
for (j = 0; j < min_n_targets; ++j)
if (strcmp(hout->target_name[j], hin->target_name[j]) != 0) {
fprintf(stderr, "[bam_merge_core] different target sequence name: '%s' != '%s' in file '%s'\n",
hout->target_name[j], hin->target_name[j], fn[i]);
return -1;
}
// If this input file has additional target reference sequences,
// add them to the headers to be output
if (hin->n_targets > hout->n_targets) {
swap_header_targets(hout, hin);
// FIXME Possibly we should also create @SQ text headers
// for the newly added reference sequences
}
bam_header_destroy(hin);
}
}
if (hheaders) {
// If the text headers to be swapped in include any @SQ headers,
// check that they are consistent with the existing binary list
// of reference information.
if (hheaders->n_targets > 0) {
if (hout->n_targets != hheaders->n_targets) {
fprintf(stderr, "[bam_merge_core] number of @SQ headers in '%s' differs from number of target sequences\n", headers);
if (!reg) return -1;
}
for (j = 0; j < hout->n_targets; ++j)
if (strcmp(hout->target_name[j], hheaders->target_name[j]) != 0) {
fprintf(stderr, "[bam_merge_core] @SQ header '%s' in '%s' differs from target sequence\n", hheaders->target_name[j], headers);
if (!reg) return -1;
}
}
swap_header_text(hout, hheaders);
bam_header_destroy(hheaders);
}
if (reg) {
int tid, beg, end;
if (bam_parse_region(hout, reg, &tid, &beg, &end) < 0) {
fprintf(stderr, "[%s] Malformated region string or undefined reference name\n", __func__);
return -1;
}
for (i = 0; i < n; ++i) {
bam_index_t *idx;
idx = bam_index_load(fn[i]);
iter[i] = bam_iter_query(idx, tid, beg, end);
bam_index_destroy(idx);
}
}
for (i = 0; i < n; ++i) {
heap1_t *h = heap + i;
h->i = i;
h->b = (bam1_t*)calloc(1, sizeof(bam1_t));
if (bam_iter_read(fp[i], iter[i], h->b) >= 0) {
h->pos = ((uint64_t)h->b->core.tid<<32) | (uint32_t)((int32_t)h->b->core.pos+1)<<1 | bam1_strand(h->b);
h->idx = idx++;
}
else h->pos = HEAP_EMPTY;
}
if (flag & MERGE_UNCOMP) level = 0;
else if (flag & MERGE_LEVEL1) level = 1;
strcpy(mode, "w");
if (level >= 0) sprintf(mode + 1, "%d", level < 9? level : 9);
if ((fpout = strcmp(out, "-")? bam_open(out, "w") : bam_dopen(fileno(stdout), "w")) == 0) {
fprintf(stderr, "[%s] fail to create the output file.\n", __func__);
return -1;
}
bam_header_write(fpout, hout);
bam_header_destroy(hout);
#ifndef _PBGZF_USE
if (!(flag & MERGE_UNCOMP)) bgzf_mt(fpout, n_threads, 256);
#endif
ks_heapmake(heap, n, heap);
while (heap->pos != HEAP_EMPTY) {
bam1_t *b = heap->b;
if (flag & MERGE_RG) {
uint8_t *rg = bam_aux_get(b, "RG");
if (rg) bam_aux_del(b, rg);
bam_aux_append(b, "RG", 'Z', RG_len[heap->i] + 1, (uint8_t*)RG[heap->i]);
}
bam_write1_core(fpout, &b->core, b->data_len, b->data);
if ((j = bam_iter_read(fp[heap->i], iter[heap->i], b)) >= 0) {
heap->pos = ((uint64_t)b->core.tid<<32) | (uint32_t)((int)b->core.pos+1)<<1 | bam1_strand(b);
heap->idx = idx++;
} else if (j == -1) {
heap->pos = HEAP_EMPTY;
free(heap->b->data); free(heap->b);
heap->b = 0;
} else fprintf(stderr, "[bam_merge_core] '%s' is truncated. Continue anyway.\n", fn[heap->i]);
ks_heapadjust(heap, 0, n, heap);
}
if (flag & MERGE_RG) {
for (i = 0; i != n; ++i) free(RG[i]);
free(RG); free(RG_len);
}
for (i = 0; i != n; ++i) {
bam_iter_destroy(iter[i]);
bam_close(fp[i]);
}
bam_close(fpout);
free(fp); free(heap); free(iter);
return 0;
}
static int dindel_fetch_func(bam1_t *b, void *data)
{
data_t_dindel *tmp = (data_t_dindel*)data;
bam1_core_t *c = &b->core;
int rlen;
uint8_t *to_delete;
/* don't change reads failing default mask: BAM_FUNMAP | BAM_FSECONDARY | BAM_FQCFAIL | BAM_FDUP */
if (c->flag & BAM_DEF_MASK) {
/* fprintf(stderr, "skipping read: %s at pos %d\n", bam1_qname(b), c->pos); */
bam_write1(tmp->out, b);
return 0;
}
/* get the reference sequence and compute homopolymer array */
if (tmp->tid != c->tid) {
/*fprintf(stderr, "fetching reference sequence %s\n",
tmp->in->header->target_name[c->tid]); */
char *ref = fai_fetch(tmp->fai, tmp->in->header->target_name[c->tid], &rlen);
strtoupper(ref);/* safeguard */
int rlen = strlen(ref);
tmp->tid = c->tid;
if (tmp->hpcount) free(tmp->hpcount);
tmp->hpcount = (int*)malloc(rlen*sizeof(int));
find_homopolymers(ref, tmp->hpcount, rlen);
free(ref);
tmp->rlen = rlen;
/* fprintf(stderr, "fetched reference sequence\n");*/
}
/* parse the cigar string */
uint32_t *cigar = bam1_cigar(b);
uint8_t indelq[c->l_qseq+1];
/* fprintf(stderr, "l_qseq:%d\n", c->l_qseq); */
int i;
int x = c->pos; /* coordinate on reference */
int y = 0; /* coordinate on query */
for (i = 0; i < c->n_cigar; ++i) {
int j, oplen = cigar[i]>>4, op = cigar[i]&0xf;
if (op == BAM_CMATCH || op == BAM_CEQUAL || op == BAM_CDIFF) {
for (j = 0; j < oplen; j++) {
/*fprintf(stderr, "query:%d, ref:%d, count:%d\n",
y, x, tmp->hpcount[x+1]); */
/* FIXME clang complains: The left operand of '>' is a garbage value */
indelq[y] = (x > tmp->rlen-2) ? DINDELQ[0] : (tmp->hpcount[x+1]>18 ?
DINDELQ[0] : DINDELQ[tmp->hpcount[x+1]]);
x++;
y++;
}
} else if (op == BAM_CHARD_CLIP) { /* do nothing */
} else if (op == BAM_CDEL) {
x += oplen;
} else if (op == BAM_CINS || op == BAM_CSOFT_CLIP) {
for (j = 0; j < oplen; j++) {
/* fprintf(stderr, "query:%d, ref:%d\n", y, x); */
indelq[y] = DINDELQ[0];
y++;
}
} else {
LOG_FATAL("unknown op %d for read %s\n", op, bam1_qname(b));/* FIXME skip? seen this somewhere else properly handled */
exit(1);
}
}
indelq[y] = '\0';
to_delete = bam_aux_get(b, BI_TAG);
if (to_delete) {
bam_aux_del(b, to_delete);
}
bam_aux_append(b, BI_TAG, 'Z', c->l_qseq+1, indelq);
to_delete = bam_aux_get(b, BD_TAG);
if (to_delete) {
bam_aux_del(b, to_delete);
}
bam_aux_append(b, BD_TAG, 'Z', c->l_qseq+1, indelq);
bam_write1(tmp->out, b);
return 0;
}
// from bam_md.c in SAMtools
// modified not fill in the NM tag, and not to start the reference a c->pos
static void
tmap_sam_md1_core(bam1_t *b, char *ref)
{
uint8_t *seq = bam1_seq(b);
uint32_t *cigar = bam1_cigar(b);
bam1_core_t *c = &b->core;
int i, x, y, u = 0;
kstring_t *str;
uint8_t *old_md, *old_nm;
int32_t old_nm_i=-1, nm=0;
str = (kstring_t*)calloc(1, sizeof(kstring_t));
for (i = y = x = 0; i < c->n_cigar; ++i) {
int j, l = cigar[i]>>4, op = cigar[i]&0xf;
if (op == BAM_CMATCH) {
for (j = 0; j < l; ++j) {
int z = y + j;
int c1 = bam1_seqi(seq, z), c2 = bam_nt16_table[(int)ref[x+j]];
if (ref[x+j] == 0) break; // out of boundary
if ((c1 == c2 && c1 != 15 && c2 != 15) || c1 == 0) { // a match
++u;
} else {
ksprintf(str, "%d", u);
kputc(ref[x+j], str);
u = 0;
nm++;
}
}
if (j < l) break;
x += l; y += l;
} else if (op == BAM_CDEL) {
ksprintf(str, "%d", u);
kputc('^', str);
for (j = 0; j < l; ++j) {
if (ref[x+j] == 0) break;
kputc(ref[x+j], str);
}
u = 0;
if (j < l) break;
x += l;
nm += l;
} else if (op == BAM_CINS || op == BAM_CSOFT_CLIP) {
y += l;
if (op == BAM_CINS) nm += l;
} else if (op == BAM_CREF_SKIP) {
x += l;
}
}
ksprintf(str, "%d", u);
// update MD
old_md = bam_aux_get(b, "MD");
if(NULL == old_md) {
bam_aux_append(b, "MD", 'Z', str->l + 1, (uint8_t*)str->s);
}
else {
int is_diff = 0;
if(strlen((char*)old_md+1) == str->l) {
for(i = 0; i < str->l; ++i) {
if(toupper(old_md[i+1]) != toupper(str->s[i])) {
break;
}
}
if(i < str->l) {
is_diff = 1;
}
}
else {
is_diff = 1;
}
if(1 == is_diff) {
bam_aux_del(b, old_md);
bam_aux_append(b, "MD", 'Z', str->l + 1, (uint8_t*)str->s);
}
}
// update NM
old_nm = bam_aux_get(b, "NM");
if(NULL != old_nm) {
old_nm_i = bam_aux2i(old_nm);
if(old_nm_i != nm) {
bam_aux_del(b, old_nm);
bam_aux_append(b, "NM", 'i', 4, (uint8_t*)&nm);
}
}
free(str->s); free(str);
}
int main(int argc, char *argv[])
{
short out2stdout=0;
hashtable ht=new_hashtable(HASHSIZE);
bamFile in,in2;
bamFile out;
int paired;//1 if not paired or pair read 1, 2 otherwise
index_mem=sizeof(hashtable)*sizeof(hashnode**)*HASHSIZE*2;
if (argc != 3) {
fprintf(stderr, "Usage: bam_fix_NH <in.bam> <out.bam or - for stdout>\n");
return 1;
}
// Open file and exit if error
in = bam_open(argv[1], "rb");
out2stdout = strcmp(argv[2], "-")? 0 : 1;
out = strcmp(argv[2], "-")? bam_open(argv[2], "w") : bam_dopen(fileno(stdout), "w");
if (in == 0 ) {
fprintf(stderr, "ERROR: Fail to open BAM file %s\n", argv[1]);
return 1;
}
if (out == 0) {
fprintf(stderr, "ERROR: Fail to open BAM file %s\n", argv[2]);
return 1;
}
unsigned long num_alns=0;
int ref;
// ***********
// Copy header
bam_header_t *header;
header = bam_header_read(in);
bam_header_write(out,header);
// sorted by name?
// Should not rely on the value in SO
bam1_t *aln=bam_init1();
bam1_t *prev=bam_init1();
if (!out2stdout) {
fprintf(stderr,"bam_fix_NH version %s\n",VERSION);
fprintf(stderr,"Processing %s\n",argv[1]);
fprintf(stderr,"Hashing...\n");fflush(stderr);
}
while(bam_read1(in,aln)>=0) { // read alignment
if (aln->core.tid < 0) continue;//ignore unaligned reads
if (aln->core.flag & BAM_FUNMAP) continue;
if (aln->core.flag & BAM_FREAD2) paired=2;
else paired=1;
++num_alns;
new_read_aln(ht,fix_read_name(bam1_qname(aln),paired));
if(!out2stdout) PRINT_ALNS_PROCESSED(num_alns);
}
bam_close(in);
if(!out2stdout) {
fprintf(stderr,"%s%lu\n",BACKLINE,num_alns);
fprintf(stderr,"Hashing complete (%lu alignments)\n",num_alns);
fprintf(stderr,"Memory used: %ld MB\n",index_mem/1024/1024);
fprintf(stderr,"Updating entries with NH and printing BAM...\n");
fflush(stderr);
}
// reopen
in2 = bam_open(argv[1], "rb");
if (in2 == 0 ) {
fprintf(stderr, "ERROR: Fail to open BAM file %s\n", argv[1]);
return 1;
}
header = bam_header_read(in2);
num_alns=0;
while(bam_read1(in2,aln)>=0) { // read alignment
paired=1;
if (aln->core.tid < 0) continue;//ignore unaligned reads
if (aln->core.flag & BAM_FUNMAP) continue;
if (aln->core.flag & BAM_FREAD2) paired=2;
++num_alns;
READ_ALN *r=get_read_aln(ht,fix_read_name(bam1_qname(aln),paired));
assert(r!=NULL);
// update the NH field
uint8_t *old_nh = bam_aux_get(aln, "NH");
int32_t nh=r->ctr;
if (old_nh) {
if (nh!=bam_aux2i(old_nh)) {
fprintf(stderr,"warning: value mismatch! replacing>%s %d->%d\n",bam1_qname(aln),bam_aux2i(old_nh),nh);
}
bam_aux_del(aln, old_nh);
bam_aux_append(aln, "NH", 'i', 4, (uint8_t*)&nh);
#ifdef DEBUG
// printf("!>%s %d\n",bam1_qname(aln),r->ctr);
#endif
}
if (!old_nh) { // add NH
bam_aux_append(aln, "NH", 'i', 4, (uint8_t*)&nh);
#ifdef DEBUG
fprintf(stderr,"!>%s %d\n",bam1_qname(aln),bam_aux2i(old_nh));
#endif
}
bam_write1(out,aln);
if(!out2stdout) PRINT_ALNS_PROCESSED(num_alns);
}
//
bam_destroy1(aln);
bam_close(in2);
bam_close(out);
if(!out2stdout) {
fprintf(stderr,"%s%lu\n",BACKLINE,num_alns);
fprintf(stderr,"Done.\n");
}
return 0;
}
/*!
@abstract Merge multiple sorted BAM.
@param is_by_qname whether to sort by query name
@param out output BAM file name
@param mode sam_open() mode to be used to create the final output file
(overrides level settings from UNCOMP and LEVEL1 flags)
@param headers name of SAM file from which to copy '@' header lines,
or NULL to copy them from the first file to be merged
@param n number of files to be merged
@param fn names of files to be merged
@param flag flags that control how the merge is undertaken
@param reg region to merge
@param n_threads number of threads to use (passed to htslib)
@discussion Padding information may NOT correctly maintained. This
function is NOT thread safe.
*/
int bam_merge_core2(int by_qname, const char *out, const char *mode, const char *headers, int n, char * const *fn, int flag, const char *reg, int n_threads)
{
samFile *fpout, **fp;
heap1_t *heap;
bam_hdr_t *hout = NULL;
int i, j, *RG_len = NULL;
uint64_t idx = 0;
char **RG = NULL;
hts_itr_t **iter = NULL;
bam_hdr_t **hdr = NULL;
trans_tbl_t *translation_tbl = NULL;
// Is there a specified pre-prepared header to use for output?
if (headers) {
samFile* fpheaders = sam_open(headers, "r");
if (fpheaders == NULL) {
const char *message = strerror(errno);
fprintf(pysamerr, "[bam_merge_core] cannot open '%s': %s\n", headers, message);
return -1;
}
hout = sam_hdr_read(fpheaders);
sam_close(fpheaders);
}
g_is_by_qname = by_qname;
fp = (samFile**)calloc(n, sizeof(samFile*));
heap = (heap1_t*)calloc(n, sizeof(heap1_t));
iter = (hts_itr_t**)calloc(n, sizeof(hts_itr_t*));
hdr = (bam_hdr_t**)calloc(n, sizeof(bam_hdr_t*));
translation_tbl = (trans_tbl_t*)calloc(n, sizeof(trans_tbl_t));
// prepare RG tag from file names
if (flag & MERGE_RG) {
RG = (char**)calloc(n, sizeof(char*));
RG_len = (int*)calloc(n, sizeof(int));
for (i = 0; i != n; ++i) {
int l = strlen(fn[i]);
const char *s = fn[i];
if (l > 4 && strcmp(s + l - 4, ".bam") == 0) l -= 4;
for (j = l - 1; j >= 0; --j) if (s[j] == '/') break;
++j; l -= j;
RG[i] = (char*)calloc(l + 1, 1);
RG_len[i] = l;
strncpy(RG[i], s + j, l);
}
}
// open and read the header from each file
for (i = 0; i < n; ++i) {
bam_hdr_t *hin;
fp[i] = sam_open(fn[i], "r");
if (fp[i] == NULL) {
int j;
fprintf(pysamerr, "[bam_merge_core] fail to open file %s\n", fn[i]);
for (j = 0; j < i; ++j) sam_close(fp[j]);
free(fp); free(heap);
// FIXME: possible memory leak
return -1;
}
hin = sam_hdr_read(fp[i]);
if (hout)
trans_tbl_init(hout, hin, translation_tbl+i, flag & MERGE_COMBINE_RG, flag & MERGE_COMBINE_PG);
else {
// As yet, no headers to merge into...
hout = bam_hdr_dup(hin);
// ...so no need to translate header into itself
trans_tbl_init(hout, hin, translation_tbl+i, true, true);
}
// TODO sam_itr_next() doesn't yet work for SAM files,
// so for those keep the headers around for use with sam_read1()
if (hts_get_format(fp[i])->format == sam) hdr[i] = hin;
else { bam_hdr_destroy(hin); hdr[i] = NULL; }
if ((translation_tbl+i)->lost_coord_sort && !by_qname) {
fprintf(pysamerr, "[bam_merge_core] Order of targets in file %s caused coordinate sort to be lost\n", fn[i]);
}
}
// Transform the header into standard form
pretty_header(&hout->text,hout->l_text);
// If we're only merging a specified region move our iters to start at that point
if (reg) {
int* rtrans = rtrans_build(n, hout->n_targets, translation_tbl);
int tid, beg, end;
const char *name_lim = hts_parse_reg(reg, &beg, &end);
char *name = malloc(name_lim - reg + 1);
memcpy(name, reg, name_lim - reg);
name[name_lim - reg] = '\0';
tid = bam_name2id(hout, name);
free(name);
if (tid < 0) {
fprintf(pysamerr, "[%s] Malformated region string or undefined reference name\n", __func__);
return -1;
}
for (i = 0; i < n; ++i) {
hts_idx_t *idx = sam_index_load(fp[i], fn[i]);
// (rtrans[i*n+tid]) Look up what hout tid translates to in input tid space
int mapped_tid = rtrans[i*hout->n_targets+tid];
if (mapped_tid != INT32_MIN) {
iter[i] = sam_itr_queryi(idx, mapped_tid, beg, end);
} else {
iter[i] = sam_itr_queryi(idx, HTS_IDX_NONE, 0, 0);
}
hts_idx_destroy(idx);
if (iter[i] == NULL) break;
}
free(rtrans);
} else {
for (i = 0; i < n; ++i) {
if (hdr[i] == NULL) {
iter[i] = sam_itr_queryi(NULL, HTS_IDX_REST, 0, 0);
if (iter[i] == NULL) break;
}
else iter[i] = NULL;
}
}
if (i < n) {
fprintf(pysamerr, "[%s] Memory allocation failed\n", __func__);
return -1;
}
// Load the first read from each file into the heap
for (i = 0; i < n; ++i) {
heap1_t *h = heap + i;
h->i = i;
h->b = bam_init1();
if ((iter[i]? sam_itr_next(fp[i], iter[i], h->b) : sam_read1(fp[i], hdr[i], h->b)) >= 0) {
bam_translate(h->b, translation_tbl + i);
h->pos = ((uint64_t)h->b->core.tid<<32) | (uint32_t)((int32_t)h->b->core.pos+1)<<1 | bam_is_rev(h->b);
h->idx = idx++;
}
else {
h->pos = HEAP_EMPTY;
bam_destroy1(h->b);
h->b = NULL;
}
}
// Open output file and write header
if ((fpout = sam_open(out, mode)) == 0) {
fprintf(pysamerr, "[%s] fail to create the output file.\n", __func__);
return -1;
}
sam_hdr_write(fpout, hout);
if (!(flag & MERGE_UNCOMP)) hts_set_threads(fpout, n_threads);
// Begin the actual merge
ks_heapmake(heap, n, heap);
while (heap->pos != HEAP_EMPTY) {
bam1_t *b = heap->b;
if (flag & MERGE_RG) {
uint8_t *rg = bam_aux_get(b, "RG");
if (rg) bam_aux_del(b, rg);
bam_aux_append(b, "RG", 'Z', RG_len[heap->i] + 1, (uint8_t*)RG[heap->i]);
}
sam_write1(fpout, hout, b);
if ((j = (iter[heap->i]? sam_itr_next(fp[heap->i], iter[heap->i], b) : sam_read1(fp[heap->i], hdr[heap->i], b))) >= 0) {
bam_translate(b, translation_tbl + heap->i);
heap->pos = ((uint64_t)b->core.tid<<32) | (uint32_t)((int)b->core.pos+1)<<1 | bam_is_rev(b);
heap->idx = idx++;
} else if (j == -1) {
heap->pos = HEAP_EMPTY;
bam_destroy1(heap->b);
heap->b = NULL;
} else fprintf(pysamerr, "[bam_merge_core] '%s' is truncated. Continue anyway.\n", fn[heap->i]);
ks_heapadjust(heap, 0, n, heap);
}
// Clean up and close
if (flag & MERGE_RG) {
for (i = 0; i != n; ++i) free(RG[i]);
free(RG); free(RG_len);
}
for (i = 0; i < n; ++i) {
trans_tbl_destroy(translation_tbl + i);
hts_itr_destroy(iter[i]);
bam_hdr_destroy(hdr[i]);
sam_close(fp[i]);
}
bam_hdr_destroy(hout);
sam_close(fpout);
free(translation_tbl); free(fp); free(heap); free(iter); free(hdr);
return 0;
}
int main(int argc, char *argv[])
{
hashtable ht=new_hashtable(HASHSIZE);
bamFile in,in2;
bamFile out;
if (argc != 3) {
fprintf(stderr, "Usage: bam_fix_NH <in.bam> <out.bam>\n");
return 1;
}
// Open file and exit if error
//in = strcmp(argv[1], "-")? bam_open(argv[1], "rb") : bam_dopen(fileno(stdin), "rb");
in = bam_open(argv[1], "rb");
out = strcmp(argv[2], "-")? bam_open(argv[2], "w") : bam_dopen(fileno(stdout), "w");
if (in == 0 ) {
fprintf(stderr, "ERROR: Fail to open BAM file %s\n", argv[1]);
return 1;
}
if (out == 0) {
fprintf(stderr, "ERROR: Fail to open BAM file %s\n", argv[2]);
return 1;
}
unsigned long num_alns=0;
int ref;
// ***********
// Copy header
bam_header_t *header;
header = bam_header_read(in);
bam_header_write(out,header);
// sorted by name?
// Should not rely on the value in SO
bam1_t *aln=bam_init1();
bam1_t *prev=bam_init1();
printf("Hashing...\n");flush(stdout);
while(bam_read1(in,aln)>=0) { // read alignment
if (aln->core.tid < 0) continue;//ignore unaligned reads
++num_alns;
new_read_aln(ht,bam1_qname(aln));
}
bam_close(in);
printf("Hashing complete (%lu alignments)\n",num_alns);
printf("Memory used in the hash: %ld MB\n",index_mem/1024/1024);
flush(stdout);
// reopen
in2 = bam_open(argv[1], "rb");
if (in2 == 0 ) {
fprintf(stderr, "ERROR: Fail to open BAM file %s\n", argv[1]);
return 1;
}
header = bam_header_read(in2);
while(bam_read1(in2,aln)>=0) { // read alignment
if (aln->core.tid < 0) continue;//ignore unaligned reads
++num_alns;
READ_ALN *r=get_read_aln(ht,bam1_qname(aln));
//assert(r!=NULL);
// update the NH field
uint8_t *old_nh = bam_aux_get(aln, "NH");
uint8_t nh=r->ctr;
if (old_nh) {
if (nh!=bam_aux2i(old_nh)) {
fprintf(stderr,"warning: value mismatch! replacing>%s %d->%d\n",bam1_qname(aln),bam_aux2i(old_nh),nh);
}
bam_aux_del(aln, old_nh);
bam_aux_append(aln, "NH", 'i', 4, (uint8_t*)&nh);
}
if (!old_nh) { // add NH
bam_aux_append(aln, "NH", 'i', 4, (uint8_t*)&nh);
#ifdef DEBUG
printf("!>%s %d\n",bam1_qname(aln),bam_aux2i(old_nh));
#endif
}
// in->header
// Also fix the XS:A tag
// BAM_FREAD1
// BAM_FREAD2
// BAM_FREVERSE the read is mapped to the reverse strand
//bam1_cigar(b)
//BAM_CREF_SKIP 3 CIGAR skip on the reference (e.g. spliced alignment)
//BAM_FREVERSE 16 the read is mapped to the reverse strand
if (aln->core.flag & BAM_FSECONDARY) continue; // skip secondary alignments
if (aln->core.flag & ! BAM_FPAIRED) continue; // not paired
if (aln->core.flag & ! BAM_FPROPER_PAIR) continue; // not a proper pair
if (aln->core.flag & ! BAM_FMUNMAP) continue; // the mate is mapped
if (aln->core.flag & BAM_FSECONDARY) continue; // secundary read
if (aln->core.flag & BAM_FREAD2) continue; // only count each pair once
// core.strand == 0 (f/+) 1 r/-
// flag
// bam1_qname(b)
bam_write1(out,aln);
}
//
bam_destroy1(aln);
bam_close(in2);
bam_close(out);
return 0;
/*
uint8_t *old_nm = bam_aux_get(b, "NM");
90 if (c->flag & BAM_FUNMAP) return;
91 if (old_nm) old_nm_i = bam_aux2i(old_nm);
92 if (!old_nm) bam_aux_append(b, "NM", 'i', 4, (uint8_t*)&nm);
93 else if (nm != old_nm_i) {
94 fprintf(stderr, "[bam_fillmd1] different NM for read '%s': %d -> %d\n", bam1_qname(b), old_nm_i, nm);
95 bam_aux_del(b, old_nm);
96 bam_aux_append(b, "NM", 'i', 4, (uint8_t*)&nm);
97 }
*/
}
void bam_fillmd1_core(bam1_t *b, char *ref, int ref_len, int flag, int max_nm)
{
uint8_t *seq = bam_get_seq(b);
uint32_t *cigar = bam_get_cigar(b);
bam1_core_t *c = &b->core;
int i, x, y, u = 0;
kstring_t *str;
int32_t old_nm_i = -1, nm = 0;
str = (kstring_t*)calloc(1, sizeof(kstring_t));
for (i = y = 0, x = c->pos; i < c->n_cigar; ++i) {
int j, l = cigar[i]>>4, op = cigar[i]&0xf;
if (op == BAM_CMATCH || op == BAM_CEQUAL || op == BAM_CDIFF) {
for (j = 0; j < l; ++j) {
int c1, c2, z = y + j;
if (x+j >= ref_len || ref[x+j] == '\0') break; // out of bounds
c1 = bam_seqi(seq, z), c2 = seq_nt16_table[(int)ref[x+j]];
if ((c1 == c2 && c1 != 15 && c2 != 15) || c1 == 0) { // a match
if (flag&USE_EQUAL) seq[z/2] &= (z&1)? 0xf0 : 0x0f;
++u;
} else {
kputw(u, str);
kputc(ref[x+j], str);
u = 0;
++nm;
}
}
if (j < l) break;
x += l;
y += l;
} else if (op == BAM_CDEL) {
kputw(u, str);
kputc('^', str);
for (j = 0; j < l; ++j) {
if (x+j >= ref_len || ref[x+j] == '\0') break;
kputc(ref[x+j], str);
}
u = 0;
x += j;
nm += j;
if (j < l) break;
} else if (op == BAM_CINS || op == BAM_CSOFT_CLIP) {
y += l;
if (op == BAM_CINS) nm += l;
} else if (op == BAM_CREF_SKIP) {
x += l;
}
}
kputw(u, str);
// apply max_nm
if (max_nm > 0 && nm >= max_nm) {
for (i = y = 0, x = c->pos; i < c->n_cigar; ++i) {
int j, l = cigar[i]>>4, op = cigar[i]&0xf;
if (op == BAM_CMATCH || op == BAM_CEQUAL || op == BAM_CDIFF) {
for (j = 0; j < l; ++j) {
int c1, c2, z = y + j;
if (x+j >= ref_len || ref[x+j] == '\0') break; // out of bounds
c1 = bam_seqi(seq, z), c2 = seq_nt16_table[(int)ref[x+j]];
if ((c1 == c2 && c1 != 15 && c2 != 15) || c1 == 0) { // a match
seq[z/2] |= (z&1)? 0x0f : 0xf0;
bam_get_qual(b)[z] = 0;
}
}
if (j < l) break;
x += l;
y += l;
} else if (op == BAM_CDEL || op == BAM_CREF_SKIP) x += l;
else if (op == BAM_CINS || op == BAM_CSOFT_CLIP) y += l;
}
}
// update NM
if ((flag & UPDATE_NM) && !(c->flag & BAM_FUNMAP)) {
uint8_t *old_nm = bam_aux_get(b, "NM");
if (old_nm) old_nm_i = bam_aux2i(old_nm);
if (!old_nm) bam_aux_append(b, "NM", 'i', 4, (uint8_t*)&nm);
else if (nm != old_nm_i) {
fprintf(stderr, "[bam_fillmd1] different NM for read '%s': %d -> %d\n", bam_get_qname(b), old_nm_i, nm);
bam_aux_del(b, old_nm);
bam_aux_append(b, "NM", 'i', 4, (uint8_t*)&nm);
}
}
// update MD
if ((flag & UPDATE_MD) && !(c->flag & BAM_FUNMAP)) {
uint8_t *old_md = bam_aux_get(b, "MD");
if (!old_md) bam_aux_append(b, "MD", 'Z', str->l + 1, (uint8_t*)str->s);
else {
int is_diff = 0;
if (strlen((char*)old_md+1) == str->l) {
for (i = 0; i < str->l; ++i)
if (toupper(old_md[i+1]) != toupper(str->s[i]))
break;
if (i < str->l) is_diff = 1;
} else is_diff = 1;
if (is_diff) {
fprintf(stderr, "[bam_fillmd1] different MD for read '%s': '%s' -> '%s'\n", bam_get_qname(b), old_md+1, str->s);
bam_aux_del(b, old_md);
bam_aux_append(b, "MD", 'Z', str->l + 1, (uint8_t*)str->s);
}
}
}
// drop all tags but RG
if (flag&DROP_TAG) {
uint8_t *q = bam_aux_get(b, "RG");
bam_aux_drop_other(b, q);
}
// reduce the resolution of base quality
if (flag&BIN_QUAL) {
uint8_t *qual = bam_get_qual(b);
for (i = 0; i < b->core.l_qseq; ++i)
if (qual[i] >= 3) qual[i] = qual[i]/10*10 + 7;
}
free(str->s);
free(str);
}
static int trim_ns(bam1_t *b, void *data) {
int ret = 0;
opts_t *op((opts_t *)data);
std::vector<uint8_t> aux(bam_get_aux(b), bam_get_aux(b) + bam_get_l_aux(b));
int tmp;
uint8_t *const seq(bam_get_seq(b));
uint32_t *const cigar(bam_get_cigar(b));
//op->n_cigar = b->core.n_cigar;
op->resize(b->l_data); // Make sure it's big enough to hold everything.
memcpy(op->data, b->data, b->core.l_qname);
// Get #Ns at the beginning
for(tmp = 0; bam_seqi(seq, tmp) == dlib::htseq::HTS_N; ++tmp);
const int n_start(tmp);
if(tmp == b->core.l_qseq - 1) // all bases are N -- garbage read
ret |= op->skip_all_ns;
// Get #Ns at the end
for(tmp = b->core.l_qseq - 1; bam_seqi(seq, tmp) == dlib::htseq::HTS_N; --tmp);
const int n_end(b->core.l_qseq - 1 - tmp);
// Get new length for read
int final_len(b->core.l_qseq - n_end - n_start);
if(final_len < 0) final_len = 0;
if(final_len < op->min_trimmed_len) // Too short.
ret |= 1;
// Copy in qual and all of aux.
if(n_end) {
if((tmp = bam_cigar_oplen(cigar[b->core.n_cigar - 1]) - n_end) == 0) {
LOG_DEBUG("Entire cigar operation is the softclip. Decrease the number of new cigar operations.\n");
--b->core.n_cigar;
} else {
LOG_DEBUG("Updating second cigar operation in-place.\n");
cigar[b->core.n_cigar - 1] = bam_cigar_gen(tmp, BAM_CSOFT_CLIP);
}
}
// Get new n_cigar.
if((tmp = bam_cigar_oplen(*cigar) - n_start) == 0) {
memcpy(op->data + b->core.l_qname, cigar + 1, (--b->core.n_cigar) << 2); // << 2 for 4 bit per cigar op
} else {
if(n_start) *cigar = bam_cigar_gen(tmp, BAM_CSOFT_CLIP);
memcpy(op->data + b->core.l_qname, cigar, b->core.n_cigar << 2);
}
uint8_t *opseq(op->data + b->core.l_qname + (b->core.n_cigar << 2)); // Pointer to the seq region of new data field.
for(tmp = 0; tmp < final_len >> 1; ++tmp)
opseq[tmp] = (bam_seqi(seq, ((tmp << 1) + n_start)) << 4) | (bam_seqi(seq, (tmp << 1) + n_start + 1));
if(final_len & 1)
opseq[tmp] = (bam_seqi(seq, ((tmp << 1) + n_start)) << 4);
tmp = bam_get_l_aux(b);
memcpy(opseq + ((final_len + 1) >> 1), bam_get_qual(b) + n_start, final_len + tmp);
// Switch data strings
std::swap(op->data, b->data);
b->core.l_qseq = final_len;
memcpy(bam_get_aux(b), aux.data(), aux.size());
b->l_data = (bam_get_aux(b) - b->data) + aux.size();
if(n_end) bam_aux_append(b, "NE", 'i', sizeof(int), (uint8_t *)&n_end);
if(n_start) bam_aux_append(b, "NS", 'i', sizeof(int), (uint8_t *)&n_start);
const uint32_t *pvar((uint32_t *)dlib::array_tag(b, "PV"));
tmp = b->core.flag & BAM_FREVERSE ? n_end: n_start;
if(pvar) {
std::vector<uint32_t>pvals(pvar + tmp, pvar + final_len + tmp);
bam_aux_del(b, (uint8_t *)(pvar) - 6);
dlib::bam_aux_array_append(b, "PV", 'I', sizeof(uint32_t), final_len, (uint8_t *)pvals.data());
}
const uint32_t *fvar((uint32_t *)dlib::array_tag(b, "FA"));
if(fvar) {
std::vector<uint32_t>fvals(fvar + tmp, fvar + final_len + tmp);
bam_aux_del(b, (uint8_t *)(fvar) - 6);
dlib::bam_aux_array_append(b, "FA", 'I', sizeof(uint32_t), final_len, (uint8_t *)fvals.data());
}
return ret;
}
static int aux_fields1(void)
{
static const char sam[] = "data:,"
"@SQ\tSN:one\tLN:1000\n"
"@SQ\tSN:two\tLN:500\n"
"r1\t0\tone\t500\t20\t8M\t*\t0\t0\tATGCATGC\tqqqqqqqq\tXA:A:k\tXi:i:37\tXf:f:" xstr(PI) "\tXd:d:" xstr(E) "\tXZ:Z:" HELLO "\tXH:H:" BEEF "\tXB:B:c,-2,0,+2\tB0:B:i,-2147483648,-1,0,1,2147483647\tB1:B:I,0,1,2147483648,4294967295\tB2:B:s,-32768,-1,0,1,32767\tB3:B:S,0,1,32768,65535\tB4:B:c,-128,-1,0,1,127\tB5:B:C,0,1,127,255\tBf:B:f,-3.14159,2.71828\tZZ:i:1000000\tF2:d:2.46801\tY1:i:-2147483648\tY2:i:-2147483647\tY3:i:-1\tY4:i:0\tY5:i:1\tY6:i:2147483647\tY7:i:2147483648\tY8:i:4294967295\n";
// Canonical form of the alignment record above, as output by sam_format1()
static const char r1[] = "r1\t0\tone\t500\t20\t8M\t*\t0\t0\tATGCATGC\tqqqqqqqq\tXi:i:37\tXf:f:3.14159\tXd:d:2.71828\tXZ:Z:" NEW_HELLO "\tXH:H:" BEEF "\tXB:B:c,-2,0,2\tB0:B:i,-2147483648,-1,0,1,2147483647\tB1:B:I,0,1,2147483648,4294967295\tB2:B:s,-32768,-1,0,1,32767\tB3:B:S,0,1,32768,65535\tB4:B:c,-128,-1,0,1,127\tB5:B:C,0,1,127,255\tBf:B:f,-3.14159,2.71828\tZZ:i:1000000\tF2:f:9.8765\tY1:i:-2147483648\tY2:i:-2147483647\tY3:i:-1\tY4:i:0\tY5:i:1\tY6:i:2147483647\tY7:i:2147483648\tY8:i:4294967295\tN0:i:-1234\tN1:i:1234\tN2:i:-2\tN3:i:3\tF1:f:4.5678\tN4:B:S,65535,32768,1,0\tN5:i:4242";
samFile *in = sam_open(sam, "r");
bam_hdr_t *header = sam_hdr_read(in);
bam1_t *aln = bam_init1();
uint8_t *p;
kstring_t ks = { 0, 0, NULL };
int64_t b0vals[5] = { -2147483648LL,-1,0,1,2147483647LL }; // i
int64_t b1vals[4] = { 0,1,2147483648LL,4294967295LL }; // I
int64_t b2vals[5] = { -32768,-1,0,1,32767 }; // s
int64_t b3vals[4] = { 0,1,32768,65535 }; // S
int64_t b4vals[5] = { -128,-1,0,1,127 }; // c
int64_t b5vals[4] = { 0,1,127,255 }; // C
// NB: Floats not doubles below!
// See https://randomascii.wordpress.com/2012/06/26/doubles-are-not-floats-so-dont-compare-them/
float bfvals[2] = { -3.14159f, 2.71828f };
int8_t n4v1[] = { -128, -64, -32, -16, -8, -4, -2, -1,
0, 1, 2, 4, 8, 16, 32, 64, 127 };
uint32_t n4v2[] = { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 1234, 5678, 1U << 31, 0 };
int16_t n4v3[] = { -32768, -1, 0, 1, 32767 };
float n4v4[] = { 0, 1, 2, 10, 20, 30, 1.5, -2.5 };
uint8_t n4v5[] = { 0, 255 };
int32_t n4v6[] = { -2147483647 - 1, 10, -1, 0, 1, 2147483647 };
uint16_t n4v7[] = { 65535, 32768, 1, 0 };
int32_t ival = -1234;
uint32_t uval = 1234;
float f1 = 4.5678;
float f2 = 9.8765;
size_t nvals, i;
if (sam_read1(in, header, aln) >= 0) {
if ((p = check_bam_aux_get(aln, "XA", 'A')) && bam_aux2A(p) != 'k')
fail("XA field is '%c', expected 'k'", bam_aux2A(p));
bam_aux_del(aln,p);
if (bam_aux_get(aln,"XA"))
fail("XA field was not deleted");
if ((p = check_bam_aux_get(aln, "Xi", 'C')) && bam_aux2i(p) != 37)
fail("Xi field is %"PRId64", expected 37", bam_aux2i(p));
if ((p = check_bam_aux_get(aln, "Xf", 'f')) && fabs(bam_aux2f(p) - PI) > 1E-6)
fail("Xf field is %.12f, expected pi", bam_aux2f(p));
if ((p = check_bam_aux_get(aln, "Xd", 'd')) && fabs(bam_aux2f(p) - E) > 1E-6)
fail("Xf field is %.12f, expected e", bam_aux2f(p));
if ((p = check_bam_aux_get(aln, "XZ", 'Z')) && strcmp(bam_aux2Z(p), HELLO) != 0)
fail("XZ field is \"%s\", expected \"%s\"", bam_aux2Z(p), HELLO);
bam_aux_update_str(aln,"XZ",strlen(NEW_HELLO)+1,NEW_HELLO);
if ((p = check_bam_aux_get(aln, "XZ", 'Z')) && strcmp(bam_aux2Z(p), NEW_HELLO) != 0)
fail("XZ field is \"%s\", expected \"%s\"", bam_aux2Z(p), NEW_HELLO);
if ((p = check_bam_aux_get(aln, "XH", 'H')) && strcmp(bam_aux2Z(p), BEEF) != 0)
fail("XH field is \"%s\", expected \"%s\"", bam_aux2Z(p), BEEF);
if ((p = check_bam_aux_get(aln, "XB", 'B'))
&& ! (memcmp(p, "Bc", 2) == 0
&& memcmp(p + 2, "\x03\x00\x00\x00\xfe\x00\x02", 7) == 0))
fail("XB field is %c,..., expected c,-2,0,+2", p[1]);
check_int_B_array(aln, "B0", NELE(b0vals), b0vals);
check_int_B_array(aln, "B1", NELE(b1vals), b1vals);
check_int_B_array(aln, "B2", NELE(b2vals), b2vals);
check_int_B_array(aln, "B3", NELE(b3vals), b3vals);
check_int_B_array(aln, "B4", NELE(b4vals), b4vals);
check_int_B_array(aln, "B5", NELE(b5vals), b5vals);
nvals = NELE(bfvals);
if ((p = check_bam_aux_get(aln, "Bf", 'B')) != NULL) {
if (bam_auxB_len(p) != nvals)
fail("Wrong length reported for Bf field, got %d, expected %zd\n",
bam_auxB_len(p), nvals);
for (i = 0; i < nvals; i++) {
if (bam_auxB2f(p, i) != bfvals[i]) {
fail("Wrong value from bam_auxB2f for Bf field index %zd, "
"got %f expected %f\n",
i, bam_auxB2f(p, i), bfvals[i]);
}
}
}
if ((p = check_bam_aux_get(aln, "ZZ", 'I')) && bam_aux2i(p) != 1000000)
fail("ZZ field is %"PRId64", expected 1000000", bam_aux2i(p));
if ((p = bam_aux_get(aln, "Y1")) && bam_aux2i(p) != -2147483647-1)
fail("Y1 field is %"PRId64", expected -2^31", bam_aux2i(p));
if ((p = bam_aux_get(aln, "Y2")) && bam_aux2i(p) != -2147483647)
fail("Y2 field is %"PRId64", expected -2^31+1", bam_aux2i(p));
if ((p = bam_aux_get(aln, "Y3")) && bam_aux2i(p) != -1)
fail("Y3 field is %"PRId64", expected -1", bam_aux2i(p));
if ((p = bam_aux_get(aln, "Y4")) && bam_aux2i(p) != 0)
fail("Y4 field is %"PRId64", expected 0", bam_aux2i(p));
if ((p = bam_aux_get(aln, "Y5")) && bam_aux2i(p) != 1)
fail("Y5 field is %"PRId64", expected 1", bam_aux2i(p));
if ((p = bam_aux_get(aln, "Y6")) && bam_aux2i(p) != 2147483647)
fail("Y6 field is %"PRId64", expected 2^31-1", bam_aux2i(p));
if ((p = bam_aux_get(aln, "Y7")) && bam_aux2i(p) != 2147483648LL)
fail("Y7 field is %"PRId64", expected 2^31", bam_aux2i(p));
if ((p = bam_aux_get(aln, "Y8")) && bam_aux2i(p) != 4294967295LL)
fail("Y8 field is %"PRId64", expected 2^32-1", bam_aux2i(p));
// Try appending some new tags
if (bam_aux_append(aln, "N0", 'i', sizeof(ival), (uint8_t *) &ival) != 0)
fail("Failed to append N0:i tag");
if ((p = bam_aux_get(aln, "N0")) && bam_aux2i(p) != ival)
fail("N0 field is %"PRId64", expected %d", bam_aux2i(p), ival);
if (bam_aux_append(aln, "N1", 'I', sizeof(uval), (uint8_t *) &uval) != 0)
fail("failed to append N1:I tag");
if ((p = bam_aux_get(aln, "N1")) && bam_aux2i(p) != uval)
fail("N1 field is %"PRId64", expected %u", bam_aux2i(p), uval);
// Append tags with bam_aux_update_int()
if (bam_aux_update_int(aln, "N2", -2) < 0)
fail("failed to append N2:c tag");
if (bam_aux_update_int(aln, "N3", 3) < 0)
fail("failed to append N3:C tag");
p = bam_aux_get(aln, "N2");
if (!p)
fail("failed to retrieve N2 tag");
else if (*p != 'c' || bam_aux2i(p) != -2)
fail("N2 field is %c:%"PRId64", expected c:-2", *p, bam_aux2i(p));
p = bam_aux_get(aln, "N3");
if (!p)
fail("failed to retrieve N3 tag");
else if (*p != 'C' || bam_aux2i(p) != 3)
fail("N3 field is %c:%"PRId64", expected C:3", *p, bam_aux2i(p));
// Try changing values with bam_aux_update_int()
i = test_update_int(aln, "N2", 2, 'C', "N3", 3, 'C');
if (i == 0) test_update_int(aln, "N2", 1234, 'S', "N3", 3, 'C');
if (i == 0) test_update_int(aln, "N2", -1, 's', "N3", 3, 'C');
if (i == 0) test_update_int(aln, "N2", 4294967295U, 'I', "N3", 3, 'C');
if (i == 0) test_update_int(aln, "N2", -2, 'i', "N3", 3, 'C');
// Append a value with bam_aux_update_float()
if (bam_aux_update_float(aln, "F1", f1) < 0)
fail("append F1:f tag");
p = bam_aux_get(aln, "F1");
if (!p)
fail("retrieve F1 tag");
else if (*p != 'f' || bam_aux2f(p) != f1)
fail("F1 field is %c:%e, expected f:%e", *p, bam_aux2f(p), f1);
// Change a double tag to a float
if (bam_aux_update_float(aln, "F2", f2) < 0)
fail("update F2 tag");
p = bam_aux_get(aln, "F2");
if (!p)
fail("retrieve F2 tag");
else if (*p != 'f' || bam_aux2f(p) != f2)
fail("F2 field is %c:%e, expected f:%e", *p, bam_aux2f(p), f2);
// Check the next one is intact too
p = bam_aux_get(aln, "Y1");
if (!p)
fail("retrieve Y1 tag");
else if (*p != 'i' && bam_aux2i(p) != -2147483647-1)
fail("Y1 field is %"PRId64", expected -2^31", bam_aux2i(p));
// bam_aux_update_array tests
// append a new array
i = test_update_array(aln, "N4", 'c', NELE(n4v1), n4v1, "\0\0", 0, 0);
// Add a sentinal to check resizes work
if (i == 0) i = test_update_int(aln, "N5", 4242, 'S', "\0\0", 0, 0);
// alter the array tag a few times
if (i == 0)
i = test_update_array(aln, "N4", 'I', NELE(n4v2), n4v2,
"N5", 4242, 'S');
if (i == 0)
i = test_update_array(aln, "N4", 's', NELE(n4v3), n4v3,
"N5", 4242, 'S');
if (i == 0)
i = test_update_array(aln, "N4", 'f', NELE(n4v4), n4v4,
"N5", 4242, 'S');
if (i == 0)
i = test_update_array(aln, "N4", 'c', NELE(n4v5), n4v5,
"N5", 4242, 'S');
if (i == 0)
i = test_update_array(aln, "N4", 'i', NELE(n4v6), n4v6,
"N5", 4242, 'S');
if (i == 0)
i = test_update_array(aln, "N4", 'S', NELE(n4v7), n4v7,
"N5", 4242, 'S');
if (sam_format1(header, aln, &ks) < 0)
fail("can't format record");
if (strcmp(ks.s, r1) != 0)
fail("record formatted incorrectly: \"%s\"", ks.s);
free(ks.s);
}
else fail("can't read record");
bam_destroy1(aln);
bam_hdr_destroy(header);
sam_close(in);
return 1;
}
