#include "bam_maqcns.h"
#include "khash.h"
#include "glf.h"
-KHASH_SET_INIT_INT64(64)
+#include "kstring.h"
+
+typedef int *indel_list_t;
+KHASH_MAP_INIT_INT64(64, indel_list_t)
#define BAM_PLF_SIMPLE 0x01
#define BAM_PLF_CNS 0x02
faidx_t *fai;
khash_t(64) *hash;
uint32_t format;
- int tid, len;
+ int tid, len, last_pos;
int mask;
char *ref;
glfFile fp; // for glf output only
} pu_data_t;
-char **bam_load_pos(const char *fn, int *_n);
+char **__bam_get_lines(const char *fn, int *_n);
void bam_init_header_hash(bam_header_t *header);
int32_t bam_get_tid(const bam_header_t *header, const char *seq_name);
static khash_t(64) *load_pos(const char *fn, bam_header_t *h)
{
- int n, tmp, i;
- char **list, *s;
- uint64_t x;
+ char **list;
+ int i, j, n, *fields, max_fields;
khash_t(64) *hash;
bam_init_header_hash(h);
- list = bam_load_pos(fn, &n);
+ list = __bam_get_lines(fn, &n);
hash = kh_init(64);
+ max_fields = 0; fields = 0;
for (i = 0; i < n; ++i) {
- x = (uint64_t)bam_get_tid(h, list[i]) << 32;
- s = list[i];
- while (*s++);
- x |= *((uint32_t*)s) - 1;
- kh_put(64, hash, x, &tmp);
- free(list[i]);
+ char *str = list[i];
+ int chr, n_fields, ret;
+ khint_t k;
+ uint64_t x;
+ n_fields = ksplit_core(str, 0, &max_fields, &fields);
+ if (n_fields < 2) continue;
+ chr = bam_get_tid(h, str + fields[0]);
+ if (chr < 0) {
+ fprintf(stderr, "[load_pos] unknown reference sequence name: %s\n", str + fields[0]);
+ continue;
+ }
+ x = (uint64_t)chr << 32 | (atoi(str + fields[1]) - 1);
+ k = kh_put(64, hash, x, &ret);
+ if (ret == 0) {
+ fprintf(stderr, "[load_pos] position %s:%s has been loaded.\n", str+fields[0], str+fields[1]);
+ continue;
+ }
+ kh_val(hash, k) = 0;
+ if (n_fields > 2) {
+ // count
+ for (j = 2; j < n_fields; ++j) {
+ char *s = str + fields[j];
+ if ((*s != '+' && *s != '-') || !isdigit(s[1])) break;
+ }
+ if (j > 2) { // update kh_val()
+ int *q, y, z;
+ q = kh_val(hash, k) = (int*)calloc(j - 1, sizeof(int));
+ q[0] = j - 2; z = j; y = 1;
+ for (j = 2; j < z; ++j)
+ q[y++] = atoi(str + fields[j]);
+ }
+ }
+ free(str);
}
- free(list);
+ free(list); free(fields);
return hash;
}
// an analogy to pileup_func() below
-static int glt_func(uint32_t tid, uint32_t pos, int n, const bam_pileup1_t *pu, void *data)
+static int glt3_func(uint32_t tid, uint32_t pos, int n, const bam_pileup1_t *pu, void *data)
{
pu_data_t *d = (pu_data_t*)data;
bam_maqindel_ret_t *r = 0;
- int rb;
+ int rb, *proposed_indels = 0;
glf1_t *g;
- glf2_t g2;
+ glf3_t *g3;
+
if (d->fai == 0) {
- fprintf(stderr, "[glt_func] reference sequence is required for generating GLT. Abort!\n");
+ fprintf(stderr, "[glt3_func] reference sequence is required for generating GLT. Abort!\n");
exit(1);
}
- if (d->hash && kh_get(64, d->hash, (uint64_t)tid<<32|pos) == kh_end(d->hash)) return 0;
+ if (d->hash) { // only output a list of sites
+ khint_t k = kh_get(64, d->hash, (uint64_t)tid<<32|pos);
+ if (k == kh_end(d->hash)) return 0;
+ proposed_indels = kh_val(d->hash, k);
+ }
+ g3 = glf3_init1();
if (d->fai && (int)tid != d->tid) {
if (d->ref) { // then write the end mark
- memset(&g2, 0, sizeof(glf2_t));
- g2.type = GLF_TYPE_END;
- bgzf_write(d->fp, &g2, sizeof(glf2_t));
+ g3->rtype = GLF3_RTYPE_END;
+ glf3_write1(d->fp, g3);
}
- glf_ref_write(d->fp, d->h->target_name[tid]); // write reference
+ glf3_ref_write(d->fp, d->h->target_name[tid], d->h->target_len[tid]); // write reference
free(d->ref);
d->ref = fai_fetch(d->fai, d->h->target_name[tid], &d->len);
d->tid = tid;
+ d->last_pos = 0;
}
rb = (d->ref && (int)pos < d->len)? d->ref[pos] : 'N';
g = bam_maqcns_glfgen(n, pu, bam_nt16_table[rb], d->c);
- memcpy(&g2, g, sizeof(glf1_t));
- g2.type = GLF_TYPE_NORMAL;
- g2.pos = pos;
- bgzf_write(d->fp, &g2, sizeof(glf2_t));
- r = bam_maqindel(n, pos, d->ido, pu, d->ref);
+ memcpy(g3, g, sizeof(glf1_t));
+ g3->rtype = GLF3_RTYPE_SUB;
+ g3->offset = pos - d->last_pos;
+ d->last_pos = pos;
+ glf3_write1(d->fp, g3);
+ if (proposed_indels)
+ r = bam_maqindel(n, pos, d->ido, pu, d->ref, proposed_indels[0], proposed_indels+1);
+ else r = bam_maqindel(n, pos, d->ido, pu, d->ref, 0, 0);
if (r) { // then write indel line
- int g3 = 3 * n, min;
- min = g3;
+ int het = 3 * n, min;
+ min = het;
if (min > r->gl[0]) min = r->gl[0];
if (min > r->gl[1]) min = r->gl[1];
- g2.ref_base = 0;
- g2.type = GLF_TYPE_INDEL;
- memset(g2.lk, 0, 10);
- g2.lk[0] = r->gl[0] - min < 255? r->gl[0] - min : 255;
- g2.lk[1] = r->gl[1] - min < 255? r->gl[1] - min : 255;
- g2.lk[2] = g3 - min < 255? g3 - min : 255;
- *(int16_t*)(g2.lk + 3) = (int16_t)r->indel1;
- *(int16_t*)(g2.lk + 5) = (int16_t)r->indel2;
- g2.min_lk = min < 255? min : 255;
- bgzf_write(d->fp, &g2, sizeof(glf2_t));
- if (r->indel1) bgzf_write(d->fp, r->s[0]+1, r->indel1>0? r->indel1 : -r->indel1);
- if (r->indel2) bgzf_write(d->fp, r->s[1]+1, r->indel2>0? r->indel2 : -r->indel2);
+ g3->ref_base = 0;
+ g3->rtype = GLF3_RTYPE_INDEL;
+ memset(g3->lk, 0, 10);
+ g3->lk[0] = r->gl[0] - min < 255? r->gl[0] - min : 255;
+ g3->lk[1] = r->gl[1] - min < 255? r->gl[1] - min : 255;
+ g3->lk[2] = het - min < 255? het - min : 255;
+ g3->offset = 0;
+ g3->indel_len[0] = r->indel1;
+ g3->indel_len[1] = r->indel2;
+ g3->min_lk = min < 255? min : 255;
+ g3->max_len = (abs(r->indel1) > abs(r->indel2)? abs(r->indel1) : abs(r->indel2)) + 1;
+ g3->indel_seq[0] = strdup(r->s[0]+1);
+ g3->indel_seq[1] = strdup(r->s[1]+1);
+ glf3_write1(d->fp, g3);
bam_maqindel_ret_destroy(r);
}
free(g);
+ glf3_destroy1(g3);
return 0;
}
{
pu_data_t *d = (pu_data_t*)data;
bam_maqindel_ret_t *r = 0;
- int i, j, rb, max_mapq = 0;
+ int i, j, rb, max_mapq = 0, *proposed_indels = 0;
uint32_t x;
- if (d->hash && kh_get(64, d->hash, (uint64_t)tid<<32|pos) == kh_end(d->hash)) return 0;
- if (d->format & BAM_PLF_GLF) return glt_func(tid, pos, n, pu, data);
- if (d->fai && (int)tid != d->tid) {
+
+ if (d->format & BAM_PLF_GLF) return glt3_func(tid, pos, n, pu, data);
+ if (d->hash) { // only output a list of sites
+ khint_t k = kh_get(64, d->hash, (uint64_t)tid<<32|pos);
+ if (k == kh_end(d->hash)) return 0;
+ proposed_indels = kh_val(d->hash, k);
+ }
+ if (d->fai && (int)tid != d->tid) { // then update d->ref
free(d->ref);
d->ref = fai_fetch(d->fai, d->h->target_name[tid], &d->len);
d->tid = tid;
}
printf("%c\t%d\t%d\t%d\t", bam_nt16_rev_table[x>>28], x>>8&0xff, ref_q, x>>16&0xff);
}
- if (d->format & (BAM_PLF_CNS|BAM_PLF_INDEL_ONLY)) {
- if (d->ref) // indel calling
- r = bam_maqindel(n, pos, d->ido, pu, d->ref);
+ if ((d->format & (BAM_PLF_CNS|BAM_PLF_INDEL_ONLY)) && d->ref) {
+ if (proposed_indels)
+ r = bam_maqindel(n, pos, d->ido, pu, d->ref, proposed_indels[0], proposed_indels+1);
+ else r = bam_maqindel(n, pos, d->ido, pu, d->ref, 0, 0);
}
// pileup strings
printf("%d\t", n);
if (!p->is_del) {
int c = bam_nt16_rev_table[bam1_seqi(bam1_seq(p->b), p->qpos)];
if (toupper(c) == toupper(rb)) c = bam1_strand(p->b)? ',' : '.';
- else bam1_strand(p->b)? tolower(c) : toupper(c);
+ else c = bam1_strand(p->b)? tolower(c) : toupper(c);
putchar(c);
if (p->indel > 0) {
printf("+%d", p->indel);
free(fn_fa);
bam_maqcns_prepare(d->c);
if (d->format & BAM_PLF_GLF) {
- glf_header_t *h;
- h = glf_header_init();
+ glf3_header_t *h;
+ h = glf3_header_init();
d->fp = bgzf_fdopen(fileno(stdout), "w");
- glf_header_write(d->fp, h);
- glf_header_destroy(h);
+ glf3_header_write(d->fp, h);
+ glf3_header_destroy(h);
}
- if (fn_list) {
+ if (d->fai == 0)
+ fprintf(stderr, "[bam_pileup] indels will not be called when -f is absent.\n");
+ if (fn_list) { // the input is SAM
tamFile fp;
bam1_t *b;
int ret;
bam_plbuf_destroy(buf);
bam_destroy1(b);
sam_close(fp);
- } else {
+ } else { // the input is BAM
bamFile fp;
fp = (strcmp(argv[optind], "-") == 0)? bam_dopen(fileno(stdin), "r") : bam_open(argv[optind], "r");
d->h = bam_header_read(fp);
bam_close(fp);
}
if (d->format & BAM_PLF_GLF) bgzf_close(d->fp);
+ if (fn_pos) { // free the hash table
+ khint_t k;
+ for (k = kh_begin(d->hash); k < kh_end(d->hash); ++k)
+ if (kh_exist(d->hash, k)) free(kh_val(d->hash, k));
+ kh_destroy(64, d->hash);
+ }
free(fn_pos); free(fn_list);
- kh_destroy(64, d->hash);
bam_header_destroy(d->h);
if (d->fai) fai_destroy(d->fai);
bam_maqcns_destroy(d->c);