6 #ifndef NCURSES_VERSION
7 #warning "_CURSES_LIB=1 but NCURSES_VERSION not defined; tview is NOT compiled"
11 #elif _CURSES_LIB == 2
15 #warning "_CURSES_LIB is not 0, 1 or 2; tview is NOT compiled"
26 #include "sam_header.h"
28 char bam_aux_getCEi(bam1_t *b, int i);
29 char bam_aux_getCSi(bam1_t *b, int i);
30 char bam_aux_getCQi(bam1_t *b, int i);
32 #define TV_MIN_ALNROW 2
33 #define TV_MAX_GOTO 40
34 #define TV_LOW_MAPQ 10
36 #define TV_COLOR_MAPQ 0
37 #define TV_COLOR_BASEQ 1
38 #define TV_COLOR_NUCL 2
39 #define TV_COLOR_COL 3
40 #define TV_COLOR_COLQ 4
42 #define TV_BASE_NUCL 0
43 #define TV_BASE_COLOR_SPACE 1
47 WINDOW *wgoto, *whelp;
53 int curr_tid, left_pos;
57 int ccol, last_pos, row_shift, base_for, color_for, is_dot, l_ref, ins, no_skip, show_name;
59 char *sample; //TODO: multiple samples and read groups
63 int tv_pl_func(uint32_t tid, uint32_t pos, int n, const bam_pileup1_t *pl, void *data)
65 extern unsigned char bam_nt16_table[256];
66 tview_t *tv = (tview_t*)data;
67 int i, j, c, rb, attr, max_ins = 0;
69 if (pos < tv->left_pos || tv->ccol > tv->mcol) return 0; // out of screen
71 rb = (tv->ref && pos - tv->left_pos < tv->l_ref)? tv->ref[pos - tv->left_pos] : 'N';
72 for (i = tv->last_pos + 1; i < pos; ++i) {
73 if (i%10 == 0 && tv->mcol - tv->ccol >= 10) mvprintw(0, tv->ccol, "%-d", i+1);
74 c = tv->ref? tv->ref[i - tv->left_pos] : 'N';
75 mvaddch(1, tv->ccol++, c);
77 if (pos%10 == 0 && tv->mcol - tv->ccol >= 10) mvprintw(0, tv->ccol, "%-d", pos+1);
80 int qsum[4], a1, a2, tmp;
81 double p[3], prior = 30;
82 bcf_call_glfgen(n, pl, bam_nt16_table[rb], tv->bca, &bcr);
83 for (i = 0; i < 4; ++i) qsum[i] = bcr.qsum[i]<<2 | i;
84 for (i = 1; i < 4; ++i) // insertion sort
85 for (j = i; j > 0 && qsum[j] > qsum[j-1]; --j)
86 tmp = qsum[j], qsum[j] = qsum[j-1], qsum[j-1] = tmp;
87 a1 = qsum[0]&3; a2 = qsum[1]&3;
88 p[0] = bcr.p[a1*5+a1]; p[1] = bcr.p[a1*5+a2] + prior; p[2] = bcr.p[a2*5+a2];
89 if ("ACGT"[a1] != toupper(rb)) p[0] += prior + 3;
90 if ("ACGT"[a2] != toupper(rb)) p[2] += prior + 3;
91 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);
92 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);
93 else call = (1<<a1|1<<a2)<<16 | (int)((p[0]<p[2]?p[0]:p[2]) - p[1] + .499);
96 c = ",ACMGRSVTWYHKDBN"[call>>16&0xf];
97 i = (call&0xffff)/10+1;
99 attr |= COLOR_PAIR(i);
100 if (c == toupper(rb)) c = '.';
102 mvaddch(2, tv->ccol, c);
105 // calculate maximum insert
106 for (i = 0; i < n; ++i) {
107 const bam_pileup1_t *p = pl + i;
108 if (p->indel > 0 && max_ins < p->indel) max_ins = p->indel;
112 for (j = 0; j <= max_ins; ++j) {
113 for (i = 0; i < n; ++i) {
114 const bam_pileup1_t *p = pl + i;
115 int row = TV_MIN_ALNROW + p->level - tv->row_shift;
118 if (tv->base_for == TV_BASE_COLOR_SPACE &&
119 (c = bam_aux_getCSi(p->b, p->qpos))) {
120 // assume that if we found one color, we will be able to get the color error
121 if (tv->is_dot && '-' == bam_aux_getCEi(p->b, p->qpos)) c = bam1_strand(p->b)? ',' : '.';
124 char *name = bam1_qname(p->b);
125 c = (p->qpos + 1 >= p->b->core.l_qname)? ' ' : name[p->qpos];
127 c = bam_nt16_rev_table[bam1_seqi(bam1_seq(p->b), p->qpos)];
128 if (tv->is_dot && toupper(c) == toupper(rb)) c = bam1_strand(p->b)? ',' : '.';
131 } else c = p->is_refskip? (bam1_strand(p->b)? '<' : '>') : '*';
133 if (j > p->indel) c = '*';
135 if (tv->base_for == TV_BASE_NUCL) {
137 char *name = bam1_qname(p->b);
138 c = (p->qpos + j + 1 >= p->b->core.l_qname)? ' ' : name[p->qpos + j];
140 c = bam_nt16_rev_table[bam1_seqi(bam1_seq(p->b), p->qpos + j)];
141 if (j == 0 && tv->is_dot && toupper(c) == toupper(rb)) c = bam1_strand(p->b)? ',' : '.';
144 c = bam_aux_getCSi(p->b, p->qpos + j);
145 if (tv->is_dot && '-' == bam_aux_getCEi(p->b, p->qpos + j)) c = bam1_strand(p->b)? ',' : '.';
149 if (row > TV_MIN_ALNROW && row < tv->mrow) {
152 if (((p->b->core.flag&BAM_FPAIRED) && !(p->b->core.flag&BAM_FPROPER_PAIR))
153 || (p->b->core.flag & BAM_FSECONDARY)) attr |= A_UNDERLINE;
154 if (tv->color_for == TV_COLOR_BASEQ) {
155 x = bam1_qual(p->b)[p->qpos]/10 + 1;
157 attr |= COLOR_PAIR(x);
158 } else if (tv->color_for == TV_COLOR_MAPQ) {
159 x = p->b->core.qual/10 + 1;
161 attr |= COLOR_PAIR(x);
162 } else if (tv->color_for == TV_COLOR_NUCL) {
163 x = bam_nt16_nt4_table[bam1_seqi(bam1_seq(p->b), p->qpos)] + 5;
164 attr |= COLOR_PAIR(x);
165 } else if(tv->color_for == TV_COLOR_COL) {
167 switch(bam_aux_getCSi(p->b, p->qpos)) {
168 case '0': x = 0; break;
169 case '1': x = 1; break;
170 case '2': x = 2; break;
171 case '3': x = 3; break;
172 case '4': x = 4; break;
173 default: x = bam_nt16_nt4_table[bam1_seqi(bam1_seq(p->b), p->qpos)]; break;
176 attr |= COLOR_PAIR(x);
177 } else if(tv->color_for == TV_COLOR_COLQ) {
178 x = bam_aux_getCQi(p->b, p->qpos);
179 if(0 == x) x = bam1_qual(p->b)[p->qpos];
182 attr |= COLOR_PAIR(x);
185 mvaddch(row, tv->ccol, bam1_strand(p->b)? tolower(c) : toupper(c));
191 attr = COLOR_PAIR(8);
193 mvaddch(1, tv->ccol++, c);
195 } else mvaddch(1, tv->ccol++, c);
201 tview_t *tv_init(const char *fn, const char *fn_fa, char *samples)
203 tview_t *tv = (tview_t*)calloc(1, sizeof(tview_t));
205 tv->fp = bam_open(fn, "r");
206 bgzf_set_cache_size(tv->fp, 8 * 1024 *1024);
208 tv->header = bam_header_read(tv->fp);
209 tv->idx = bam_index_load(fn);
210 if (tv->idx == 0) exit(1);
211 tv->lplbuf = bam_lplbuf_init(tv_pl_func, tv);
212 if (fn_fa) tv->fai = fai_load(fn_fa);
213 tv->bca = bcf_call_init(0.83, 13);
218 tv->sample = samples;
219 tv->header->dict = sam_header_parse2(tv->header->text);
220 tv->rg2sm = sam_header2tbl(tv->header->dict, "RG", "ID", "SM");
224 keypad(stdscr, TRUE);
228 tv->mrow = 24; tv->mcol = 80;
229 getmaxyx(stdscr, tv->mrow, tv->mcol);
230 tv->wgoto = newwin(3, TV_MAX_GOTO + 10, 10, 5);
231 tv->whelp = newwin(29, 40, 5, 5);
232 tv->color_for = TV_COLOR_MAPQ;
234 init_pair(1, COLOR_BLUE, COLOR_BLACK);
235 init_pair(2, COLOR_GREEN, COLOR_BLACK);
236 init_pair(3, COLOR_YELLOW, COLOR_BLACK);
237 init_pair(4, COLOR_WHITE, COLOR_BLACK);
238 init_pair(5, COLOR_GREEN, COLOR_BLACK);
239 init_pair(6, COLOR_CYAN, COLOR_BLACK);
240 init_pair(7, COLOR_YELLOW, COLOR_BLACK);
241 init_pair(8, COLOR_RED, COLOR_BLACK);
242 init_pair(9, COLOR_BLUE, COLOR_BLACK);
246 void tv_destroy(tview_t *tv)
248 delwin(tv->wgoto); delwin(tv->whelp);
251 bam_lplbuf_destroy(tv->lplbuf);
252 bcf_call_destroy(tv->bca);
253 bam_index_destroy(tv->idx);
254 if (tv->fai) fai_destroy(tv->fai);
256 bam_header_destroy(tv->header);
261 int tv_fetch_func(const bam1_t *b, void *data)
263 tview_t *tv = (tview_t*)data;
266 const uint8_t *rg = bam_aux_get(b, "RG");
268 const char *sm = sam_tbl_get(tv->rg2sm, (const char*)(rg + 1));
270 if ( strcmp(sm,tv->sample) ) return 0;
273 uint32_t *cigar = bam1_cigar(b); // this is cheating...
275 for (i = 0; i <b->core.n_cigar; ++i) {
276 if ((cigar[i]&0xf) == BAM_CREF_SKIP)
277 cigar[i] = cigar[i]>>4<<4 | BAM_CDEL;
280 bam_lplbuf_push(b, tv->lplbuf);
284 int tv_draw_aln(tview_t *tv, int tid, int pos)
288 tv->curr_tid = tid; tv->left_pos = pos;
289 tv->last_pos = tv->left_pos - 1;
291 // print ref and consensus
294 if (tv->ref) free(tv->ref);
295 str = (char*)calloc(strlen(tv->header->target_name[tv->curr_tid]) + 30, 1);
296 sprintf(str, "%s:%d-%d", tv->header->target_name[tv->curr_tid], tv->left_pos + 1, tv->left_pos + tv->mcol);
297 tv->ref = fai_fetch(tv->fai, str, &tv->l_ref);
301 bam_lplbuf_reset(tv->lplbuf);
302 bam_fetch(tv->fp, tv->idx, tv->curr_tid, tv->left_pos, tv->left_pos + tv->mcol, tv, tv_fetch_func);
303 bam_lplbuf_push(0, tv->lplbuf);
305 while (tv->ccol < tv->mcol) {
306 int pos = tv->last_pos + 1;
307 if (pos%10 == 0 && tv->mcol - tv->ccol >= 10) mvprintw(0, tv->ccol, "%-d", pos+1);
308 mvaddch(1, tv->ccol++, (tv->ref && pos < tv->l_ref)? tv->ref[pos - tv->left_pos] : 'N');
314 static void tv_win_goto(tview_t *tv, int *tid, int *pos)
318 wborder(tv->wgoto, '|', '|', '-', '-', '+', '+', '+', '+');
319 mvwprintw(tv->wgoto, 1, 2, "Goto: ");
321 int c = wgetch(tv->wgoto);
323 if (c == KEY_BACKSPACE || c == '\010' || c == '\177') {
325 } else if (c == KEY_ENTER || c == '\012' || c == '\015') {
326 int _tid = -1, _beg, _end;
328 _beg = strtol(str+1, &p, 10) - 1;
334 bam_parse_region(tv->header, str, &_tid, &_beg, &_end);
336 *tid = _tid; *pos = _beg;
340 } else if (isgraph(c)) {
341 if (l < TV_MAX_GOTO) str[l++] = c;
342 } else if (c == '\027') l = 0;
343 else if (c == '\033') return;
345 for (i = 0; i < TV_MAX_GOTO; ++i) mvwaddch(tv->wgoto, 1, 8 + i, ' ');
346 mvwprintw(tv->wgoto, 1, 8, "%s", str);
350 static void tv_win_help(tview_t *tv) {
352 WINDOW *win = tv->whelp;
353 wborder(win, '|', '|', '-', '-', '+', '+', '+', '+');
354 mvwprintw(win, r++, 2, " -=- Help -=- ");
356 mvwprintw(win, r++, 2, "? This window");
357 mvwprintw(win, r++, 2, "Arrows Small scroll movement");
358 mvwprintw(win, r++, 2, "h,j,k,l Small scroll movement");
359 mvwprintw(win, r++, 2, "H,J,K,L Large scroll movement");
360 mvwprintw(win, r++, 2, "ctrl-H Scroll 1k left");
361 mvwprintw(win, r++, 2, "ctrl-L Scroll 1k right");
362 mvwprintw(win, r++, 2, "space Scroll one screen");
363 mvwprintw(win, r++, 2, "backspace Scroll back one screen");
364 mvwprintw(win, r++, 2, "g Go to specific location");
365 mvwprintw(win, r++, 2, "m Color for mapping qual");
366 mvwprintw(win, r++, 2, "n Color for nucleotide");
367 mvwprintw(win, r++, 2, "b Color for base quality");
368 mvwprintw(win, r++, 2, "c Color for cs color");
369 mvwprintw(win, r++, 2, "z Color for cs qual");
370 mvwprintw(win, r++, 2, ". Toggle on/off dot view");
371 mvwprintw(win, r++, 2, "s Toggle on/off ref skip");
372 mvwprintw(win, r++, 2, "r Toggle on/off rd name");
373 mvwprintw(win, r++, 2, "N Turn on nt view");
374 mvwprintw(win, r++, 2, "C Turn on cs view");
375 mvwprintw(win, r++, 2, "i Toggle on/off ins");
376 mvwprintw(win, r++, 2, "q Exit");
378 mvwprintw(win, r++, 2, "Underline: Secondary or orphan");
379 mvwprintw(win, r++, 2, "Blue: 0-9 Green: 10-19");
380 mvwprintw(win, r++, 2, "Yellow: 20-29 White: >=30");
385 void tv_loop(tview_t *tv)
388 tid = tv->curr_tid; pos = tv->left_pos;
392 case '?': tv_win_help(tv); break;
394 case 'q': goto end_loop;
396 case 'g': tv_win_goto(tv, &tid, &pos); break;
397 case 'm': tv->color_for = TV_COLOR_MAPQ; break;
398 case 'b': tv->color_for = TV_COLOR_BASEQ; break;
399 case 'n': tv->color_for = TV_COLOR_NUCL; break;
400 case 'c': tv->color_for = TV_COLOR_COL; break;
401 case 'z': tv->color_for = TV_COLOR_COLQ; break;
402 case 's': tv->no_skip = !tv->no_skip; break;
403 case 'r': tv->show_name = !tv->show_name; break;
405 case 'h': --pos; break;
407 case 'l': ++pos; break;
409 case 'H': pos -= 20; break;
411 case 'L': pos += 20; break;
412 case '.': tv->is_dot = !tv->is_dot; break;
413 case 'N': tv->base_for = TV_BASE_NUCL; break;
414 case 'C': tv->base_for = TV_BASE_COLOR_SPACE; break;
415 case 'i': tv->ins = !tv->ins; break;
416 case '\010': pos -= 1000; break;
417 case '\014': pos += 1000; break;
418 case ' ': pos += tv->mcol; break;
420 case 'j': --tv->row_shift; break;
422 case 'k': ++tv->row_shift; break;
424 case '\177': pos -= tv->mcol; break;
425 case KEY_RESIZE: getmaxyx(stdscr, tv->mrow, tv->mcol); break;
428 if (pos < 0) pos = 0;
429 if (tv->row_shift < 0) tv->row_shift = 0;
430 tv_draw_aln(tv, tid, pos);
436 void error(const char *format, ...)
440 fprintf(stderr, "\n");
441 fprintf(stderr, "Usage: bamtk tview [options] <aln.bam> [ref.fasta]\n");
442 fprintf(stderr, "Options:\n");
443 fprintf(stderr, " -s STR display only reads from this sample\n");
444 fprintf(stderr, "\n\n");
449 va_start(ap, format);
450 vfprintf(stderr, format, ap);
457 int bam_tview_main(int argc, char *argv[])
462 while ((c = getopt(argc, argv, "s:")) >= 0) {
464 case 's': samples=optarg; break;
465 default: error(NULL);
468 if (argc==optind) error(NULL);
469 tv = tv_init(argv[optind], (optind+1>=argc)? 0 : argv[optind+1], samples);
470 tv_draw_aln(tv, 0, 0);
475 #else // #ifdef _HAVE_CURSES
477 #warning "No curses library is available; tview is disabled."
478 int bam_tview_main(int argc, char *argv[])
480 fprintf(stderr, "[bam_tview_main] The ncurses library is unavailable; tview is not compiled.\n");
483 #endif // #ifdef _HAVE_CURSES