1 #undef _HAVE_CURSES
2 
3 #if _CURSES_LIB == 0
4 #elif _CURSES_LIB == 1
5 #include <curses.h>
6 #ifndef NCURSES_VERSION
7 #warning "_CURSES_LIB=1 but NCURSES_VERSION not defined; tview is NOT compiled"
8 #else
9 #define _HAVE_CURSES
10 #endif
11 #elif _CURSES_LIB == 2
12 #include <xcurses.h>
13 #define _HAVE_CURSES
14 #else
15 #warning "_CURSES_LIB is not 0, 1 or 2; tview is NOT compiled"
16 #endif
17 
18 #ifdef _HAVE_CURSES
19 #include <ctype.h>
20 #include <assert.h>
21 #include <string.h>
22 #include <math.h>
23 #include "bam.h"
24 #include "faidx.h"
25 #include "bam2bcf.h"
26 
27 char bam_aux_getCEi(bam1_t *b, int i);
28 char bam_aux_getCSi(bam1_t *b, int i);
29 char bam_aux_getCQi(bam1_t *b, int i);
30 
31 #define TV_MIN_ALNROW 2
32 #define TV_MAX_GOTO  40
33 #define TV_LOW_MAPQ  10
34 
35 #define TV_COLOR_MAPQ   0
36 #define TV_COLOR_BASEQ  1
37 #define TV_COLOR_NUCL   2
38 #define TV_COLOR_COL    3
39 #define TV_COLOR_COLQ   4
40 
41 #define TV_BASE_NUCL 0
42 #define TV_BASE_COLOR_SPACE 1
43 
44 typedef struct {
45 	int mrow, mcol;
46 	WINDOW *wgoto, *whelp;
47 
48 	bam_index_t *idx;
49 	bam_lplbuf_t *lplbuf;
50 	bam_header_t *header;
51 	bamFile fp;
52 	int curr_tid, left_pos;
53 	faidx_t *fai;
54 	bcf_callaux_t *bca;
55 
56 	int ccol, last_pos, row_shift, base_for, color_for, is_dot, l_ref, ins, no_skip, show_name;
57 	char *ref;
58 } tview_t;
59 
tv_pl_func(uint32_t tid,uint32_t pos,int n,const bam_pileup1_t * pl,void * data)60 int tv_pl_func(uint32_t tid, uint32_t pos, int n, const bam_pileup1_t *pl, void *data)
61 {
62 	extern unsigned char bam_nt16_table[256];
63 	tview_t *tv = (tview_t*)data;
64 	int i, j, c, rb, attr, max_ins = 0;
65 	uint32_t call = 0;
66 	if (pos < tv->left_pos || tv->ccol > tv->mcol) return 0; // out of screen
67 	// print referece
68 	rb = (tv->ref && pos - tv->left_pos < tv->l_ref)? tv->ref[pos - tv->left_pos] : 'N';
69 	for (i = tv->last_pos + 1; i < pos; ++i) {
70 		if (i%10 == 0 && tv->mcol - tv->ccol >= 10) mvprintw(0, tv->ccol, "%-d", i+1);
71 		c = tv->ref? tv->ref[i - tv->left_pos] : 'N';
72 		mvaddch(1, tv->ccol++, c);
73 	}
74 	if (pos%10 == 0 && tv->mcol - tv->ccol >= 10) mvprintw(0, tv->ccol, "%-d", pos+1);
75 	{ // call consensus
76 		bcf_callret1_t bcr;
77 		int qsum[4], a1, a2, tmp;
78 		double p[3], prior = 30;
79 		bcf_call_glfgen(n, pl, bam_nt16_table[rb], tv->bca, &bcr);
80 		for (i = 0; i < 4; ++i) qsum[i] = bcr.qsum[i]<<2 | i;
81 		for (i = 1; i < 4; ++i) // insertion sort
82 			for (j = i; j > 0 && qsum[j] > qsum[j-1]; --j)
83 				tmp = qsum[j], qsum[j] = qsum[j-1], qsum[j-1] = tmp;
84 		a1 = qsum[0]&3; a2 = qsum[1]&3;
85 		p[0] = bcr.p[a1*5+a1]; p[1] = bcr.p[a1*5+a2] + prior; p[2] = bcr.p[a2*5+a2];
86 		if ("ACGT"[a1] != toupper(rb)) p[0] += prior + 3;
87 		if ("ACGT"[a2] != toupper(rb)) p[2] += prior + 3;
88 		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);
89 		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);
90 		else call = (1<<a1|1<<a2)<<16 | (int)((p[0]<p[2]?p[0]:p[2]) - p[1] + .499);
91 	}
92 	attr = A_UNDERLINE;
93 	c = ",ACMGRSVTWYHKDBN"[call>>16&0xf];
94 	i = (call&0xffff)/10+1;
95 	if (i > 4) i = 4;
96 	attr |= COLOR_PAIR(i);
97 	if (c == toupper(rb)) c = '.';
98 	attron(attr);
99 	mvaddch(2, tv->ccol, c);
100 	attroff(attr);
101 	if(tv->ins) {
102 		// calculate maximum insert
103 		for (i = 0; i < n; ++i) {
104 			const bam_pileup1_t *p = pl + i;
105 			if (p->indel > 0 && max_ins < p->indel) max_ins = p->indel;
106 		}
107 	}
108 	// core loop
109 	for (j = 0; j <= max_ins; ++j) {
110 		for (i = 0; i < n; ++i) {
111 			const bam_pileup1_t *p = pl + i;
112 			int row = TV_MIN_ALNROW + p->level - tv->row_shift;
113 			if (j == 0) {
114 				if (!p->is_del) {
115 					if (tv->base_for == TV_BASE_COLOR_SPACE &&
116 							(c = bam_aux_getCSi(p->b, p->qpos))) {
117 						c = bam_aux_getCSi(p->b, p->qpos);
118 						// assume that if we found one color, we will be able to get the color error
119 						if (tv->is_dot && '-' == bam_aux_getCEi(p->b, p->qpos)) c = bam1_strand(p->b)? ',' : '.';
120 					} else {
121 						if (tv->show_name) {
122 							char *name = bam1_qname(p->b);
123 							c = (p->qpos + 1 >= p->b->core.l_qname)? ' ' : name[p->qpos];
124 						} else {
125 							c = bam_nt16_rev_table[bam1_seqi(bam1_seq(p->b), p->qpos)];
126 							if (tv->is_dot && toupper(c) == toupper(rb)) c = bam1_strand(p->b)? ',' : '.';
127 						}
128 					}
129 				} else c = p->is_refskip? (bam1_strand(p->b)? '<' : '>') : '*';
130 			} else { // padding
131 				if (j > p->indel) c = '*';
132 				else { // insertion
133 					if (tv->base_for ==  TV_BASE_NUCL) {
134 						if (tv->show_name) {
135 							char *name = bam1_qname(p->b);
136 							c = (p->qpos + j + 1 >= p->b->core.l_qname)? ' ' : name[p->qpos + j];
137 						} else {
138 							c = bam_nt16_rev_table[bam1_seqi(bam1_seq(p->b), p->qpos + j)];
139 							if (j == 0 && tv->is_dot && toupper(c) == toupper(rb)) c = bam1_strand(p->b)? ',' : '.';
140 						}
141 					} else {
142 						c = bam_aux_getCSi(p->b, p->qpos + j);
143 						if (tv->is_dot && '-' == bam_aux_getCEi(p->b, p->qpos + j)) c = bam1_strand(p->b)? ',' : '.';
144 					}
145 				}
146 			}
147 			if (row > TV_MIN_ALNROW && row < tv->mrow) {
148 				int x;
149 				attr = 0;
150 				if (((p->b->core.flag&BAM_FPAIRED) && !(p->b->core.flag&BAM_FPROPER_PAIR))
151 						|| (p->b->core.flag & BAM_FSECONDARY)) attr |= A_UNDERLINE;
152 				if (tv->color_for == TV_COLOR_BASEQ) {
153 					x = bam1_qual(p->b)[p->qpos]/10 + 1;
154 					if (x > 4) x = 4;
155 					attr |= COLOR_PAIR(x);
156 				} else if (tv->color_for == TV_COLOR_MAPQ) {
157 					x = p->b->core.qual/10 + 1;
158 					if (x > 4) x = 4;
159 					attr |= COLOR_PAIR(x);
160 				} else if (tv->color_for == TV_COLOR_NUCL) {
161 					x = bam_nt16_nt4_table[bam1_seqi(bam1_seq(p->b), p->qpos)] + 5;
162 					attr |= COLOR_PAIR(x);
163 				} else if(tv->color_for == TV_COLOR_COL) {
164 					x = 0;
165 					switch(bam_aux_getCSi(p->b, p->qpos)) {
166 						case '0': x = 0; break;
167 						case '1': x = 1; break;
168 						case '2': x = 2; break;
169 						case '3': x = 3; break;
170 						case '4': x = 4; break;
171 						default: x = bam_nt16_nt4_table[bam1_seqi(bam1_seq(p->b), p->qpos)]; break;
172 					}
173 					x+=5;
174 					attr |= COLOR_PAIR(x);
175 				} else if(tv->color_for == TV_COLOR_COLQ) {
176 					x = bam_aux_getCQi(p->b, p->qpos);
177 					if(0 == x) x = bam1_qual(p->b)[p->qpos];
178 					x = x/10 + 1;
179 					if (x > 4) x = 4;
180 					attr |= COLOR_PAIR(x);
181 				}
182 				attron(attr);
183 				mvaddch(row, tv->ccol, bam1_strand(p->b)? tolower(c) : toupper(c));
184 				attroff(attr);
185 			}
186 		}
187 		c = j? '*' : rb;
188 		if (c == '*') {
189 			attr = COLOR_PAIR(8);
190 			attron(attr);
191 			mvaddch(1, tv->ccol++, c);
192 			attroff(attr);
193 		} else mvaddch(1, tv->ccol++, c);
194 	}
195 	tv->last_pos = pos;
196 	return 0;
197 }
198 
tv_init(const char * fn,const char * fn_fa)199 tview_t *tv_init(const char *fn, const char *fn_fa)
200 {
201 	tview_t *tv = (tview_t*)calloc(1, sizeof(tview_t));
202 	tv->is_dot = 1;
203 	tv->fp = bam_open(fn, "r");
204 	bgzf_set_cache_size(tv->fp, 8 * 1024 *1024);
205 	assert(tv->fp);
206 	tv->header = bam_header_read(tv->fp);
207 	tv->idx = bam_index_load(fn);
208 	if (tv->idx == 0) exit(1);
209 	tv->lplbuf = bam_lplbuf_init(tv_pl_func, tv);
210 	if (fn_fa) tv->fai = fai_load(fn_fa);
211 	tv->bca = bcf_call_init(0.83, 13);
212 	tv->ins = 1;
213 
214 	initscr();
215 	keypad(stdscr, TRUE);
216 	clear();
217 	noecho();
218 	cbreak();
219 	tv->mrow = 24; tv->mcol = 80;
220 	getmaxyx(stdscr, tv->mrow, tv->mcol);
221 	tv->wgoto = newwin(3, TV_MAX_GOTO + 10, 10, 5);
222 	tv->whelp = newwin(29, 40, 5, 5);
223 	tv->color_for = TV_COLOR_MAPQ;
224 	start_color();
225 	init_pair(1, COLOR_BLUE, COLOR_BLACK);
226 	init_pair(2, COLOR_GREEN, COLOR_BLACK);
227 	init_pair(3, COLOR_YELLOW, COLOR_BLACK);
228 	init_pair(4, COLOR_WHITE, COLOR_BLACK);
229 	init_pair(5, COLOR_GREEN, COLOR_BLACK);
230 	init_pair(6, COLOR_CYAN, COLOR_BLACK);
231 	init_pair(7, COLOR_YELLOW, COLOR_BLACK);
232 	init_pair(8, COLOR_RED, COLOR_BLACK);
233 	init_pair(9, COLOR_BLUE, COLOR_BLACK);
234 	return tv;
235 }
236 
tv_destroy(tview_t * tv)237 void tv_destroy(tview_t *tv)
238 {
239 	delwin(tv->wgoto); delwin(tv->whelp);
240 	endwin();
241 
242 	bam_lplbuf_destroy(tv->lplbuf);
243 	bcf_call_destroy(tv->bca);
244 	bam_index_destroy(tv->idx);
245 	if (tv->fai) fai_destroy(tv->fai);
246 	free(tv->ref);
247 	bam_header_destroy(tv->header);
248 	bam_close(tv->fp);
249 	free(tv);
250 }
251 
tv_fetch_func(const bam1_t * b,void * data)252 int tv_fetch_func(const bam1_t *b, void *data)
253 {
254 	tview_t *tv = (tview_t*)data;
255 	if (tv->no_skip) {
256 		uint32_t *cigar = bam1_cigar(b); // this is cheating...
257 		int i;
258 		for (i = 0; i <b->core.n_cigar; ++i) {
259 			if ((cigar[i]&0xf) == BAM_CREF_SKIP)
260 				cigar[i] = cigar[i]>>4<<4 | BAM_CDEL;
261 		}
262 	}
263 	bam_lplbuf_push(b, tv->lplbuf);
264 	return 0;
265 }
266 
tv_draw_aln(tview_t * tv,int tid,int pos)267 int tv_draw_aln(tview_t *tv, int tid, int pos)
268 {
269 	// reset
270 	clear();
271 	tv->curr_tid = tid; tv->left_pos = pos;
272 	tv->last_pos = tv->left_pos - 1;
273 	tv->ccol = 0;
274 	// print ref and consensus
275 	if (tv->fai) {
276 		char *str;
277 		if (tv->ref) free(tv->ref);
278 		str = (char*)calloc(strlen(tv->header->target_name[tv->curr_tid]) + 30, 1);
279 		sprintf(str, "%s:%d-%d", tv->header->target_name[tv->curr_tid], tv->left_pos + 1, tv->left_pos + tv->mcol);
280 		tv->ref = fai_fetch(tv->fai, str, &tv->l_ref);
281 		free(str);
282 	}
283 	// draw aln
284 	bam_lplbuf_reset(tv->lplbuf);
285 	bam_fetch(tv->fp, tv->idx, tv->curr_tid, tv->left_pos, tv->left_pos + tv->mcol, tv, tv_fetch_func);
286 	bam_lplbuf_push(0, tv->lplbuf);
287 
288 	while (tv->ccol < tv->mcol) {
289 		int pos = tv->last_pos + 1;
290 		if (pos%10 == 0 && tv->mcol - tv->ccol >= 10) mvprintw(0, tv->ccol, "%-d", pos+1);
291 		mvaddch(1, tv->ccol++, (tv->ref && pos < tv->l_ref)? tv->ref[pos - tv->left_pos] : 'N');
292 		++tv->last_pos;
293 	}
294 	return 0;
295 }
296 
tv_win_goto(tview_t * tv,int * tid,int * pos)297 static void tv_win_goto(tview_t *tv, int *tid, int *pos)
298 {
299 	char str[256], *p;
300 	int i, l = 0;
301 	wborder(tv->wgoto, '|', '|', '-', '-', '+', '+', '+', '+');
302 	mvwprintw(tv->wgoto, 1, 2, "Goto: ");
303 	for (;;) {
304 		int c = wgetch(tv->wgoto);
305 		wrefresh(tv->wgoto);
306 		if (c == KEY_BACKSPACE || c == '\010' || c == '\177') {
307 			--l;
308 		} else if (c == KEY_ENTER || c == '\012' || c == '\015') {
309 			int _tid = -1, _beg, _end;
310 			if (str[0] == '=') {
311 				_beg = strtol(str+1, &p, 10) - 1;
312 				if (_beg > 0) {
313 					*pos = _beg;
314 					return;
315 				}
316 			} else {
317 				bam_parse_region(tv->header, str, &_tid, &_beg, &_end);
318 				if (_tid >= 0) {
319 					*tid = _tid; *pos = _beg;
320 					return;
321 				}
322 			}
323 		} else if (isgraph(c)) {
324 			if (l < TV_MAX_GOTO) str[l++] = c;
325 		} else if (c == '\027') l = 0;
326 		else if (c == '\033') return;
327 		str[l] = '\0';
328 		for (i = 0; i < TV_MAX_GOTO; ++i) mvwaddch(tv->wgoto, 1, 8 + i, ' ');
329 		mvwprintw(tv->wgoto, 1, 8, "%s", str);
330 	}
331 }
332 
tv_win_help(tview_t * tv)333 static void tv_win_help(tview_t *tv) {
334 	int r = 1;
335 	WINDOW *win = tv->whelp;
336 	wborder(win, '|', '|', '-', '-', '+', '+', '+', '+');
337 	mvwprintw(win, r++, 2, "        -=-    Help    -=- ");
338 	r++;
339 	mvwprintw(win, r++, 2, "?          This window");
340 	mvwprintw(win, r++, 2, "Arrows     Small scroll movement");
341 	mvwprintw(win, r++, 2, "h,j,k,l    Small scroll movement");
342 	mvwprintw(win, r++, 2, "H,J,K,L    Large scroll movement");
343 	mvwprintw(win, r++, 2, "ctrl-H     Scroll 1k left");
344 	mvwprintw(win, r++, 2, "ctrl-L     Scroll 1k right");
345 	mvwprintw(win, r++, 2, "space      Scroll one screen");
346 	mvwprintw(win, r++, 2, "backspace  Scroll back one screen");
347 	mvwprintw(win, r++, 2, "g          Go to specific location");
348 	mvwprintw(win, r++, 2, "m          Color for mapping qual");
349 	mvwprintw(win, r++, 2, "n          Color for nucleotide");
350 	mvwprintw(win, r++, 2, "b          Color for base quality");
351 	mvwprintw(win, r++, 2, "c          Color for cs color");
352 	mvwprintw(win, r++, 2, "z          Color for cs qual");
353 	mvwprintw(win, r++, 2, ".          Toggle on/off dot view");
354 	mvwprintw(win, r++, 2, "s          Toggle on/off ref skip");
355 	mvwprintw(win, r++, 2, "r          Toggle on/off rd name");
356 	mvwprintw(win, r++, 2, "N          Turn on nt view");
357 	mvwprintw(win, r++, 2, "C          Turn on cs view");
358 	mvwprintw(win, r++, 2, "i          Toggle on/off ins");
359 	mvwprintw(win, r++, 2, "q          Exit");
360 	r++;
361 	mvwprintw(win, r++, 2, "Underline:      Secondary or orphan");
362 	mvwprintw(win, r++, 2, "Blue:    0-9    Green: 10-19");
363 	mvwprintw(win, r++, 2, "Yellow: 20-29   White: >=30");
364 	wrefresh(win);
365 	wgetch(win);
366 }
367 
tv_loop(tview_t * tv)368 void tv_loop(tview_t *tv)
369 {
370 	int tid, pos;
371 	tid = tv->curr_tid; pos = tv->left_pos;
372 	while (1) {
373 		int c = getch();
374 		switch (c) {
375 			case '?': tv_win_help(tv); break;
376 			case '\033':
377 			case 'q': goto end_loop;
378 			case '/':
379 			case 'g': tv_win_goto(tv, &tid, &pos); break;
380 			case 'm': tv->color_for = TV_COLOR_MAPQ; break;
381 			case 'b': tv->color_for = TV_COLOR_BASEQ; break;
382 			case 'n': tv->color_for = TV_COLOR_NUCL; break;
383 			case 'c': tv->color_for = TV_COLOR_COL; break;
384 			case 'z': tv->color_for = TV_COLOR_COLQ; break;
385 			case 's': tv->no_skip = !tv->no_skip; break;
386 			case 'r': tv->show_name = !tv->show_name; break;
387 			case KEY_LEFT:
388 			case 'h': --pos; break;
389 			case KEY_RIGHT:
390 			case 'l': ++pos; break;
391 			case KEY_SLEFT:
392 			case 'H': pos -= 20; break;
393 			case KEY_SRIGHT:
394 			case 'L': pos += 20; break;
395 			case '.': tv->is_dot = !tv->is_dot; break;
396 			case 'N': tv->base_for = TV_BASE_NUCL; break;
397 			case 'C': tv->base_for = TV_BASE_COLOR_SPACE; break;
398 			case 'i': tv->ins = !tv->ins; break;
399 			case '\010': pos -= 1000; break;
400 			case '\014': pos += 1000; break;
401 			case ' ': pos += tv->mcol; break;
402 			case KEY_UP:
403 			case 'j': --tv->row_shift; break;
404 			case KEY_DOWN:
405 			case 'k': ++tv->row_shift; break;
406 			case KEY_BACKSPACE:
407 			case '\177': pos -= tv->mcol; break;
408 			case KEY_RESIZE: getmaxyx(stdscr, tv->mrow, tv->mcol); break;
409 			default: continue;
410 		}
411 		if (pos < 0) pos = 0;
412 		if (tv->row_shift < 0) tv->row_shift = 0;
413 		tv_draw_aln(tv, tid, pos);
414 	}
415 end_loop:
416 	return;
417 }
418 
bam_tview_main(int argc,char * argv[])419 int bam_tview_main(int argc, char *argv[])
420 {
421 	tview_t *tv;
422 	if (argc == 1) {
423 		fprintf(stderr, "Usage: bamtk tview <aln.bam> [ref.fasta]\n");
424 		return 1;
425 	}
426 	tv = tv_init(argv[1], (argc == 2)? 0 : argv[2]);
427 	tv_draw_aln(tv, 0, 0);
428 	tv_loop(tv);
429 	tv_destroy(tv);
430 	return 0;
431 }
432 #else // #ifdef _HAVE_CURSES
433 #include <stdio.h>
434 #warning "No curses library is available; tview is disabled."
bam_tview_main(int argc,char * argv[])435 int bam_tview_main(int argc, char *argv[])
436 {
437 	fprintf(stderr, "[bam_tview_main] The ncurses library is unavailable; tview is not compiled.\n");
438 	return 1;
439 }
440 #endif // #ifdef _HAVE_CURSES
441