bam_tview.c

   1 #include <assert.h>
   2 #include "bam_tview.h"
   3
   4 int base_tv_init(tview_t* tv,const char *fn, const char *fn_fa, const char *samples)
   5         {
   6         assert(tv!=NULL);
   7         assert(fn!=NULL);
   8         tv->mrow = 24; tv->mcol = 80;
   9         tv->color_for = TV_COLOR_MAPQ;
  10         tv->is_dot = 1;
  11         fprintf(stderr,"Opening '%s'\n",fn);
  12         tv->fp = bam_open(fn, "r");
  13         if(tv->fp==0)
  14                 {
  15                 fprintf(stderr,"bam_open %s. %s\n", fn,fn_fa);
  16                 exit(EXIT_FAILURE);
  17                 }
  18         bgzf_set_cache_size(tv->fp, 8 * 1024 *1024);
  19         assert(tv->fp);
  20         fprintf(stderr,"Opening header '%s'\n",fn);
  21         tv->header = bam_header_read(tv->fp);
  22         if(tv->header==0)
  23                 {
  24                 fprintf(stderr,"Cannot read '%s'.\n", fn);
  25                 exit(EXIT_FAILURE);
  26                 }
  27         tv->idx = bam_index_load(fn);
  28         if (tv->idx == 0) exit(1);
  29         tv->lplbuf = bam_lplbuf_init(tv_pl_func, tv);
  30         if (fn_fa) tv->fai = fai_load(fn_fa);
  31         tv->bca = bcf_call_init(0.83, 13);
  32         tv->ins = 1;
  33
  34     if ( samples )
  35     {
  36         if ( !tv->header->dict ) tv->header->dict = sam_header_parse2(tv->header->text);
  37         void *iter = tv->header->dict;
  38         const char *key, *val;
  39         int n = 0;
  40         tv->rg_hash = kh_init(kh_rg);
  41         while ( (iter = sam_header2key_val(iter, "RG","ID","SM", &key, &val)) )
  42         {
  43             if ( !strcmp(samples,key) || (val && !strcmp(samples,val)) )
  44             {
  45                 khiter_t k = kh_get(kh_rg, tv->rg_hash, key);
  46                 if ( k != kh_end(tv->rg_hash) ) continue;
  47                 int ret;
  48                 k = kh_put(kh_rg, tv->rg_hash, key, &ret);
  49                 kh_value(tv->rg_hash, k) = val;
  50                 n++;
  51             }
  52         }
  53         if ( !n )
  54         {
  55             fprintf(stderr,"The sample or read group \"%s\" not present.\n", samples);
  56             exit(EXIT_FAILURE);
  57         }
  58     }
  59
  60         return 0;
  61         }
  62
  63
  64 void base_tv_destroy(tview_t* tv)
  65         {
  66         bam_lplbuf_destroy(tv->lplbuf);
  67         bcf_call_destroy(tv->bca);
  68         bam_index_destroy(tv->idx);
  69         if (tv->fai) fai_destroy(tv->fai);
  70         free(tv->ref);
  71         bam_header_destroy(tv->header);
  72         bam_close(tv->fp);
  73         }
  74
  75
  76 int tv_pl_func(uint32_t tid, uint32_t pos, int n, const bam_pileup1_t *pl, void *data)
  77 {
  78         extern unsigned char bam_nt16_table[256];
  79         tview_t *tv = (tview_t*)data;
  80         int i, j, c, rb, attr, max_ins = 0;
  81         uint32_t call = 0;
  82         if (pos < tv->left_pos || tv->ccol > tv->mcol) return 0; // out of screen
  83         // print referece
  84         rb = (tv->ref && pos - tv->left_pos < tv->l_ref)? tv->ref[pos - tv->left_pos] : 'N';
  85         for (i = tv->last_pos + 1; i < pos; ++i) {
  86                 if (i%10 == 0 && tv->mcol - tv->ccol >= 10) tv->my_mvprintw(tv,0, tv->ccol, "%-d", i+1);
  87                 c = tv->ref? tv->ref[i - tv->left_pos] : 'N';
  88                 tv->my_mvaddch(tv,1, tv->ccol++, c);
  89         }
  90         if (pos%10 == 0 && tv->mcol - tv->ccol >= 10) tv->my_mvprintw(tv,0, tv->ccol, "%-d", pos+1);
  91         { // call consensus
  92                 bcf_callret1_t bcr;
  93                 int qsum[4], a1, a2, tmp;
  94                 double p[3], prior = 30;
  95                 bcf_call_glfgen(n, pl, bam_nt16_table[rb], tv->bca, &bcr);
  96                 for (i = 0; i < 4; ++i) qsum[i] = bcr.qsum[i]<<2 | i;
  97                 for (i = 1; i < 4; ++i) // insertion sort
  98                         for (j = i; j > 0 && qsum[j] > qsum[j-1]; --j)
  99                                 tmp = qsum[j], qsum[j] = qsum[j-1], qsum[j-1] = tmp;
 100                 a1 = qsum[0]&3; a2 = qsum[1]&3;
 101                 p[0] = bcr.p[a1*5+a1]; p[1] = bcr.p[a1*5+a2] + prior; p[2] = bcr.p[a2*5+a2];
 102                 if ("ACGT"[a1] != toupper(rb)) p[0] += prior + 3;
 103                 if ("ACGT"[a2] != toupper(rb)) p[2] += prior + 3;
 104                 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);
 105                 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);
 106                 else call = (1<<a1|1<<a2)<<16 | (int)((p[0]<p[2]?p[0]:p[2]) - p[1] + .499);
 107         }
 108         attr = BAM_TVIEW_UNDERLINE;
 109         c = ",ACMGRSVTWYHKDBN"[call>>16&0xf];
 110         i = (call&0xffff)/10+1;
 111         if (i > 4) i = 4;
 112         attr |= tv->my_colorpair(tv,i);
 113         if (c == toupper(rb)) c = '.';
 114         tv->my_attron(tv,attr);
 115         tv->my_mvaddch(tv,2, tv->ccol, c);
 116         tv->my_attroff(tv,attr);
 117         if(tv->ins) {
 118                 // calculate maximum insert
 119                 for (i = 0; i < n; ++i) {
 120                         const bam_pileup1_t *p = pl + i;
 121                         if (p->indel > 0 && max_ins < p->indel) max_ins = p->indel;
 122                 }
 123         }
 124         // core loop
 125         for (j = 0; j <= max_ins; ++j) {
 126                 for (i = 0; i < n; ++i) {
 127                         const bam_pileup1_t *p = pl + i;
 128                         int row = TV_MIN_ALNROW + p->level - tv->row_shift;
 129                         if (j == 0) {
 130                                 if (!p->is_del) {
 131                                         if (tv->base_for == TV_BASE_COLOR_SPACE &&
 132                                                         (c = bam_aux_getCSi(p->b, p->qpos))) {
 133                                                 // assume that if we found one color, we will be able to get the color error
 134                                                 if (tv->is_dot && '-' == bam_aux_getCEi(p->b, p->qpos)) c = bam1_strand(p->b)? ',' : '.';
 135                                         } else {
 136                                                 if (tv->show_name) {
 137                                                         char *name = bam1_qname(p->b);
 138                                                         c = (p->qpos + 1 >= p->b->core.l_qname)? ' ' : name[p->qpos];
 139                                                 } else {
 140                                                         c = bam_nt16_rev_table[bam1_seqi(bam1_seq(p->b), p->qpos)];
 141                                                         if (tv->is_dot && toupper(c) == toupper(rb)) c = bam1_strand(p->b)? ',' : '.';
 142                                                 }
 143                                         }
 144                                 } else c = p->is_refskip? (bam1_strand(p->b)? '<' : '>') : '*';
 145                         } else { // padding
 146                                 if (j > p->indel) c = '*';
 147                                 else { // insertion
 148                                         if (tv->base_for ==  TV_BASE_NUCL) {
 149                                                 if (tv->show_name) {
 150                                                         char *name = bam1_qname(p->b);
 151                                                         c = (p->qpos + j + 1 >= p->b->core.l_qname)? ' ' : name[p->qpos + j];
 152                                                 } else {
 153                                                         c = bam_nt16_rev_table[bam1_seqi(bam1_seq(p->b), p->qpos + j)];
 154                                                         if (j == 0 && tv->is_dot && toupper(c) == toupper(rb)) c = bam1_strand(p->b)? ',' : '.';
 155                                                 }
 156                                         } else {
 157                                                 c = bam_aux_getCSi(p->b, p->qpos + j);
 158                                                 if (tv->is_dot && '-' == bam_aux_getCEi(p->b, p->qpos + j)) c = bam1_strand(p->b)? ',' : '.';
 159                                         }
 160                                 }
 161                         }
 162                         if (row > TV_MIN_ALNROW && row < tv->mrow) {
 163                                 int x;
 164                                 attr = 0;
 165                                 if (((p->b->core.flag&BAM_FPAIRED) && !(p->b->core.flag&BAM_FPROPER_PAIR))
 166                                                 || (p->b->core.flag & BAM_FSECONDARY)) attr |= BAM_TVIEW_UNDERLINE;
 167                                 if (tv->color_for == TV_COLOR_BASEQ) {
 168                                         x = bam1_qual(p->b)[p->qpos]/10 + 1;
 169                                         if (x > 4) x = 4;
 170                                         attr |= COLOR_PAIR(x);
 171                                 } else if (tv->color_for == TV_COLOR_MAPQ) {
 172                                         x = p->b->core.qual/10 + 1;
 173                                         if (x > 4) x = 4;
 174                                         attr |= COLOR_PAIR(x);
 175                                 } else if (tv->color_for == TV_COLOR_NUCL) {
 176                                         x = bam_nt16_nt4_table[bam1_seqi(bam1_seq(p->b), p->qpos)] + 5;
 177                                         attr |= COLOR_PAIR(x);
 178                                 } else if(tv->color_for == TV_COLOR_COL) {
 179                                         x = 0;
 180                                         switch(bam_aux_getCSi(p->b, p->qpos)) {
 181                                                 case '0': x = 0; break;
 182                                                 case '1': x = 1; break;
 183                                                 case '2': x = 2; break;
 184                                                 case '3': x = 3; break;
 185                                                 case '4': x = 4; break;
 186                                                 default: x = bam_nt16_nt4_table[bam1_seqi(bam1_seq(p->b), p->qpos)]; break;
 187                                         }
 188                                         x+=5;
 189                                         attr |= COLOR_PAIR(x);
 190                                 } else if(tv->color_for == TV_COLOR_COLQ) {
 191                                         x = bam_aux_getCQi(p->b, p->qpos);
 192                                         if(0 == x) x = bam1_qual(p->b)[p->qpos];
 193                                         x = x/10 + 1;
 194                                         if (x > 4) x = 4;
 195                                         attr |= COLOR_PAIR(x);
 196                                 }
 197                                 tv->my_attron(tv,attr);
 198                                 tv->my_mvaddch(tv,row, tv->ccol, bam1_strand(p->b)? tolower(c) : toupper(c));
 199                                 tv->my_attroff(tv,attr);
 200                         }
 201                 }
 202                 c = j? '*' : rb;
 203                 if (c == '*') {
 204                         attr = COLOR_PAIR(8);
 205                         tv->my_attron(tv,attr);
 206                         tv->my_mvaddch(tv,1, tv->ccol++, c);
 207                         tv->my_attroff(tv,attr);
 208                 } else tv->my_mvaddch(tv,1, tv->ccol++, c);
 209         }
 210         tv->last_pos = pos;
 211         return 0;
 212 }
 213
 214
 215
 216
 217 int tv_fetch_func(const bam1_t *b, void *data)
 218 {
 219         tview_t *tv = (tview_t*)data;
 220     if ( tv->rg_hash )
 221     {
 222         const uint8_t *rg = bam_aux_get(b, "RG");
 223         if ( !rg ) return 0;
 224         khiter_t k = kh_get(kh_rg, tv->rg_hash, (const char*)(rg + 1));
 225         if ( k == kh_end(tv->rg_hash) ) return 0;
 226     }
 227         if (tv->no_skip) {
 228                 uint32_t *cigar = bam1_cigar(b); // this is cheating...
 229                 int i;
 230                 for (i = 0; i <b->core.n_cigar; ++i) {
 231                         if ((cigar[i]&0xf) == BAM_CREF_SKIP)
 232                                 cigar[i] = cigar[i]>>4<<4 | BAM_CDEL;
 233                 }
 234         }
 235         bam_lplbuf_push(b, tv->lplbuf);
 236         return 0;
 237 }
 238
 239 int tv_draw_aln(tview_t *tv, int tid, int pos)
 240         {
 241         assert(tv!=NULL);
 242         // reset
 243         tv->my_clear(tv);
 244         tv->curr_tid = tid; tv->left_pos = pos;
 245         tv->last_pos = tv->left_pos - 1;
 246         tv->ccol = 0;
 247         // print ref and consensus
 248         if (tv->fai) {
 249                 char *str;
 250                 if (tv->ref) free(tv->ref);
 251                 assert(tv->curr_tid>=0);
 252
 253                 str = (char*)calloc(strlen(tv->header->target_name[tv->curr_tid]) + 30, 1);
 254                 assert(str!=NULL);
 255                 sprintf(str, "%s:%d-%d", tv->header->target_name[tv->curr_tid], tv->left_pos + 1, tv->left_pos + tv->mcol);
 256                 tv->ref = fai_fetch(tv->fai, str, &tv->l_ref);
 257                 free(str);
 258         }
 259         // draw aln
 260         bam_lplbuf_reset(tv->lplbuf);
 261         bam_fetch(tv->fp, tv->idx, tv->curr_tid, tv->left_pos, tv->left_pos + tv->mcol, tv, tv_fetch_func);
 262         bam_lplbuf_push(0, tv->lplbuf);
 263
 264         while (tv->ccol < tv->mcol) {
 265                 int pos = tv->last_pos + 1;
 266                 if (pos%10 == 0 && tv->mcol - tv->ccol >= 10) tv->my_mvprintw(tv,0, tv->ccol, "%-d", pos+1);
 267                 tv->my_mvaddch(tv,1, tv->ccol++, (tv->ref && pos < tv->l_ref)? tv->ref[pos - tv->left_pos] : 'N');
 268                 ++tv->last_pos;
 269         }
 270         return 0;
 271 }
 272