#include <ctype.h>
#include <assert.h>
#include <string.h>
+#include <math.h>
+#include <unistd.h>
#include "bam.h"
#include "faidx.h"
-#include "bam_maqcns.h"
+#include "bam2bcf.h"
+#include "sam_header.h"
+#include "khash.h"
+
+KHASH_MAP_INIT_STR(kh_rg, const char *)
char bam_aux_getCEi(bam1_t *b, int i);
char bam_aux_getCSi(bam1_t *b, int i);
bamFile fp;
int curr_tid, left_pos;
faidx_t *fai;
- bam_maqcns_t *bmc;
+ bcf_callaux_t *bca;
- int ccol, last_pos, row_shift, base_for, color_for, is_dot, l_ref, ins, no_skip;
+ int ccol, last_pos, row_shift, base_for, color_for, is_dot, l_ref, ins, no_skip, show_name;
char *ref;
+ khash_t(kh_rg) *rg_hash;
} tview_t;
int tv_pl_func(uint32_t tid, uint32_t pos, int n, const bam_pileup1_t *pl, void *data)
{
+ extern unsigned char bam_nt16_table[256];
tview_t *tv = (tview_t*)data;
int i, j, c, rb, attr, max_ins = 0;
uint32_t call = 0;
mvaddch(1, tv->ccol++, c);
}
if (pos%10 == 0 && tv->mcol - tv->ccol >= 10) mvprintw(0, tv->ccol, "%-d", pos+1);
- // print consensus
- call = bam_maqcns_call(n, pl, tv->bmc);
+ { // call consensus
+ bcf_callret1_t bcr;
+ int qsum[4], a1, a2, tmp;
+ double p[3], prior = 30;
+ bcf_call_glfgen(n, pl, bam_nt16_table[rb], tv->bca, &bcr);
+ for (i = 0; i < 4; ++i) qsum[i] = bcr.qsum[i]<<2 | i;
+ for (i = 1; i < 4; ++i) // insertion sort
+ for (j = i; j > 0 && qsum[j] > qsum[j-1]; --j)
+ tmp = qsum[j], qsum[j] = qsum[j-1], qsum[j-1] = tmp;
+ a1 = qsum[0]&3; a2 = qsum[1]&3;
+ p[0] = bcr.p[a1*5+a1]; p[1] = bcr.p[a1*5+a2] + prior; p[2] = bcr.p[a2*5+a2];
+ if ("ACGT"[a1] != toupper(rb)) p[0] += prior + 3;
+ if ("ACGT"[a2] != toupper(rb)) p[2] += prior + 3;
+ if (p[0] < p[1] && p[0] < p[2]) call = (1<<a1)<<16 | (int)((p[1]<p[2]?p[1]:p[2]) - p[0] + .499);
+ else if (p[2] < p[1] && p[2] < p[0]) call = (1<<a2)<<16 | (int)((p[0]<p[1]?p[0]:p[1]) - p[2] + .499);
+ else call = (1<<a1|1<<a2)<<16 | (int)((p[0]<p[2]?p[0]:p[2]) - p[1] + .499);
+ }
attr = A_UNDERLINE;
- c = ",ACMGRSVTWYHKDBN"[call>>28&0xf];
- i = (call>>8&0xff)/10+1;
+ c = ",ACMGRSVTWYHKDBN"[call>>16&0xf];
+ i = (call&0xffff)/10+1;
if (i > 4) i = 4;
attr |= COLOR_PAIR(i);
if (c == toupper(rb)) c = '.';
if (!p->is_del) {
if (tv->base_for == TV_BASE_COLOR_SPACE &&
(c = bam_aux_getCSi(p->b, p->qpos))) {
- c = bam_aux_getCSi(p->b, p->qpos);
// assume that if we found one color, we will be able to get the color error
if (tv->is_dot && '-' == bam_aux_getCEi(p->b, p->qpos)) c = bam1_strand(p->b)? ',' : '.';
+ } else {
+ if (tv->show_name) {
+ char *name = bam1_qname(p->b);
+ c = (p->qpos + 1 >= p->b->core.l_qname)? ' ' : name[p->qpos];
+ } else {
+ c = bam_nt16_rev_table[bam1_seqi(bam1_seq(p->b), p->qpos)];
+ if (tv->is_dot && toupper(c) == toupper(rb)) c = bam1_strand(p->b)? ',' : '.';
+ }
}
- else {
- c = bam_nt16_rev_table[bam1_seqi(bam1_seq(p->b), p->qpos)];
- if (tv->is_dot && toupper(c) == toupper(rb)) c = bam1_strand(p->b)? ',' : '.';
- }
- } else c = '*';
+ } else c = p->is_refskip? (bam1_strand(p->b)? '<' : '>') : '*';
} else { // padding
if (j > p->indel) c = '*';
else { // insertion
if (tv->base_for == TV_BASE_NUCL) {
- c = bam_nt16_rev_table[bam1_seqi(bam1_seq(p->b), p->qpos + j)];
- if (j == 0 && tv->is_dot && toupper(c) == toupper(rb)) c = bam1_strand(p->b)? ',' : '.';
- }
- else {
+ if (tv->show_name) {
+ char *name = bam1_qname(p->b);
+ c = (p->qpos + j + 1 >= p->b->core.l_qname)? ' ' : name[p->qpos + j];
+ } else {
+ c = bam_nt16_rev_table[bam1_seqi(bam1_seq(p->b), p->qpos + j)];
+ if (j == 0 && tv->is_dot && toupper(c) == toupper(rb)) c = bam1_strand(p->b)? ',' : '.';
+ }
+ } else {
c = bam_aux_getCSi(p->b, p->qpos + j);
if (tv->is_dot && '-' == bam_aux_getCEi(p->b, p->qpos + j)) c = bam1_strand(p->b)? ',' : '.';
}
return 0;
}
-tview_t *tv_init(const char *fn, const char *fn_fa)
+tview_t *tv_init(const char *fn, const char *fn_fa, char *samples)
{
tview_t *tv = (tview_t*)calloc(1, sizeof(tview_t));
tv->is_dot = 1;
- tv->idx = bam_index_load(fn);
- if (tv->idx == 0) exit(1);
tv->fp = bam_open(fn, "r");
bgzf_set_cache_size(tv->fp, 8 * 1024 *1024);
assert(tv->fp);
tv->header = bam_header_read(tv->fp);
+ tv->idx = bam_index_load(fn);
+ if (tv->idx == 0) exit(1);
tv->lplbuf = bam_lplbuf_init(tv_pl_func, tv);
if (fn_fa) tv->fai = fai_load(fn_fa);
- tv->bmc = bam_maqcns_init();
+ tv->bca = bcf_call_init(0.83, 13);
tv->ins = 1;
- bam_maqcns_prepare(tv->bmc);
+
+ if ( samples )
+ {
+ if ( !tv->header->dict ) tv->header->dict = sam_header_parse2(tv->header->text);
+ void *iter = tv->header->dict;
+ const char *key, *val;
+ int n = 0;
+ tv->rg_hash = kh_init(kh_rg);
+ while ( (iter = sam_header2key_val(iter, "RG","ID","SM", &key, &val)) )
+ {
+ if ( !strcmp(samples,key) || (val && !strcmp(samples,val)) )
+ {
+ khiter_t k = kh_get(kh_rg, tv->rg_hash, key);
+ if ( k != kh_end(tv->rg_hash) ) continue;
+ int ret;
+ k = kh_put(kh_rg, tv->rg_hash, key, &ret);
+ kh_value(tv->rg_hash, k) = val;
+ n++;
+ }
+ }
+ if ( !n )
+ {
+ fprintf(stderr,"The sample or read group \"%s\" not present.\n", samples);
+ exit(-1);
+ }
+ }
initscr();
keypad(stdscr, TRUE);
tv->mrow = 24; tv->mcol = 80;
getmaxyx(stdscr, tv->mrow, tv->mcol);
tv->wgoto = newwin(3, TV_MAX_GOTO + 10, 10, 5);
- tv->whelp = newwin(28, 40, 5, 5);
+ tv->whelp = newwin(29, 40, 5, 5);
tv->color_for = TV_COLOR_MAPQ;
start_color();
init_pair(1, COLOR_BLUE, COLOR_BLACK);
endwin();
bam_lplbuf_destroy(tv->lplbuf);
- bam_maqcns_destroy(tv->bmc);
+ bcf_call_destroy(tv->bca);
bam_index_destroy(tv->idx);
if (tv->fai) fai_destroy(tv->fai);
free(tv->ref);
int tv_fetch_func(const bam1_t *b, void *data)
{
tview_t *tv = (tview_t*)data;
+ if ( tv->rg_hash )
+ {
+ const uint8_t *rg = bam_aux_get(b, "RG");
+ if ( !rg ) return 0;
+ khiter_t k = kh_get(kh_rg, tv->rg_hash, (const char*)(rg + 1));
+ if ( k == kh_end(tv->rg_hash) ) return 0;
+ }
if (tv->no_skip) {
uint32_t *cigar = bam1_cigar(b); // this is cheating...
int i;
static void tv_win_goto(tview_t *tv, int *tid, int *pos)
{
- char str[256];
+ char str[256], *p;
int i, l = 0;
wborder(tv->wgoto, '|', '|', '-', '-', '+', '+', '+', '+');
mvwprintw(tv->wgoto, 1, 2, "Goto: ");
int c = wgetch(tv->wgoto);
wrefresh(tv->wgoto);
if (c == KEY_BACKSPACE || c == '\010' || c == '\177') {
- --l;
+ if(l > 0) --l;
} else if (c == KEY_ENTER || c == '\012' || c == '\015') {
int _tid = -1, _beg, _end;
- bam_parse_region(tv->header, str, &_tid, &_beg, &_end);
- if (_tid >= 0) {
- *tid = _tid; *pos = _beg;
- return;
+ if (str[0] == '=') {
+ _beg = strtol(str+1, &p, 10) - 1;
+ if (_beg > 0) {
+ *pos = _beg;
+ return;
+ }
+ } else {
+ bam_parse_region(tv->header, str, &_tid, &_beg, &_end);
+ if (_tid >= 0) {
+ *tid = _tid; *pos = _beg;
+ return;
+ }
}
} else if (isgraph(c)) {
if (l < TV_MAX_GOTO) str[l++] = c;
mvwprintw(win, r++, 2, "z Color for cs qual");
mvwprintw(win, r++, 2, ". Toggle on/off dot view");
mvwprintw(win, r++, 2, "s Toggle on/off ref skip");
+ mvwprintw(win, r++, 2, "r Toggle on/off rd name");
mvwprintw(win, r++, 2, "N Turn on nt view");
mvwprintw(win, r++, 2, "C Turn on cs view");
mvwprintw(win, r++, 2, "i Toggle on/off ins");
case '?': tv_win_help(tv); break;
case '\033':
case 'q': goto end_loop;
+ case '/':
case 'g': tv_win_goto(tv, &tid, &pos); break;
case 'm': tv->color_for = TV_COLOR_MAPQ; break;
case 'b': tv->color_for = TV_COLOR_BASEQ; break;
case 'c': tv->color_for = TV_COLOR_COL; break;
case 'z': tv->color_for = TV_COLOR_COLQ; break;
case 's': tv->no_skip = !tv->no_skip; break;
+ case 'r': tv->show_name = !tv->show_name; break;
case KEY_LEFT:
case 'h': --pos; break;
case KEY_RIGHT:
return;
}
+void error(const char *format, ...)
+{
+ if ( !format )
+ {
+ fprintf(stderr, "\n");
+ fprintf(stderr, "Usage: bamtk tview [options] <aln.bam> [ref.fasta]\n");
+ fprintf(stderr, "Options:\n");
+ fprintf(stderr, " -p chr:pos go directly to this position\n");
+ fprintf(stderr, " -s STR display only reads from this sample or grou\n");
+ fprintf(stderr, "\n\n");
+ }
+ else
+ {
+ va_list ap;
+ va_start(ap, format);
+ vfprintf(stderr, format, ap);
+ va_end(ap);
+ }
+ exit(-1);
+}
+
+
int bam_tview_main(int argc, char *argv[])
{
tview_t *tv;
- if (argc == 1) {
- fprintf(stderr, "Usage: bamtk tview <aln.bam> [ref.fasta]\n");
- return 1;
- }
- tv = tv_init(argv[1], (argc == 2)? 0 : argv[2]);
- tv_draw_aln(tv, 0, 0);
+ char *samples=NULL, *position=NULL;
+ int c;
+ while ((c = getopt(argc, argv, "s:p:")) >= 0) {
+ switch (c) {
+ case 's': samples=optarg; break;
+ case 'p': position=optarg; break;
+ default: error(NULL);
+ }
+ }
+ if (argc==optind) error(NULL);
+ tv = tv_init(argv[optind], (optind+1>=argc)? 0 : argv[optind+1], samples);
+ if ( position )
+ {
+ int _tid = -1, _beg, _end;
+ bam_parse_region(tv->header, position, &_tid, &_beg, &_end);
+ if (_tid >= 0) { tv->curr_tid = _tid; tv->left_pos = _beg; }
+ }
+ tv_draw_aln(tv, tv->curr_tid, tv->left_pos);
tv_loop(tv);
tv_destroy(tv);
return 0;
projects / samtools.git / blobdiff