bam2bcf_indel.c

   1 #include <assert.h>
   2 #include <ctype.h>
   3 #include <string.h>
   4 #include "bam.h"
   5 #include "bam2bcf.h"
   6 #include "kaln.h"
   7 #include "kprobaln.h"
   8 #include "khash.h"
   9 KHASH_SET_INIT_STR(rg)
  10
  11 #include "ksort.h"
  12 KSORT_INIT_GENERIC(uint32_t)
  13
  14 #define MINUS_CONST 0x10000000
  15 #define INDEL_WINDOW_SIZE 50
  16
  17 void *bcf_call_add_rg(void *_hash, const char *hdtext, const char *list)
  18 {
  19         const char *s, *p, *q, *r, *t;
  20         khash_t(rg) *hash;
  21         if (list == 0 || hdtext == 0) return _hash;
  22         if (_hash == 0) _hash = kh_init(rg);
  23         hash = (khash_t(rg)*)_hash;
  24         if ((s = strstr(hdtext, "@RG\t")) == 0) return hash;
  25         do {
  26                 t = strstr(s + 4, "@RG\t"); // the next @RG
  27                 if ((p = strstr(s, "\tID:")) != 0) p += 4;
  28                 if ((q = strstr(s, "\tPL:")) != 0) q += 4;
  29                 if (p && q && (t == 0 || (p < t && q < t))) { // ID and PL are both present
  30                         int lp, lq;
  31                         char *x;
  32                         for (r = p; *r && *r != '\t' && *r != '\n'; ++r); lp = r - p;
  33                         for (r = q; *r && *r != '\t' && *r != '\n'; ++r); lq = r - q;
  34                         x = calloc((lp > lq? lp : lq) + 1, 1);
  35                         for (r = q; *r && *r != '\t' && *r != '\n'; ++r) x[r-q] = *r;
  36                         if (strstr(list, x)) { // insert ID to the hash table
  37                                 khint_t k;
  38                                 int ret;
  39                                 for (r = p; *r && *r != '\t' && *r != '\n'; ++r) x[r-p] = *r;
  40                                 x[r-p] = 0;
  41                                 k = kh_get(rg, hash, x);
  42                                 if (k == kh_end(hash)) k = kh_put(rg, hash, x, &ret);
  43                                 else free(x);
  44                         } else free(x);
  45                 }
  46                 s = t;
  47         } while (s);
  48         return hash;
  49 }
  50
  51 void bcf_call_del_rghash(void *_hash)
  52 {
  53         khint_t k;
  54         khash_t(rg) *hash = (khash_t(rg)*)_hash;
  55         if (hash == 0) return;
  56         for (k = kh_begin(hash); k < kh_end(hash); ++k)
  57                 if (kh_exist(hash, k))
  58                         free((char*)kh_key(hash, k));
  59         kh_destroy(rg, hash);
  60 }
  61
  62 static int tpos2qpos(const bam1_core_t *c, const uint32_t *cigar, int32_t tpos, int is_left, int32_t *_tpos)
  63 {
  64         int k, x = c->pos, y = 0, last_y = 0;
  65         *_tpos = c->pos;
  66         for (k = 0; k < c->n_cigar; ++k) {
  67                 int op = cigar[k] & BAM_CIGAR_MASK;
  68                 int l = cigar[k] >> BAM_CIGAR_SHIFT;
  69                 if (op == BAM_CMATCH || op == BAM_CEQUAL || op == BAM_CDIFF) {
  70                         if (c->pos > tpos) return y;
  71                         if (x + l > tpos) {
  72                                 *_tpos = tpos;
  73                                 return y + (tpos - x);
  74                         }
  75                         x += l; y += l;
  76                         last_y = y;
  77                 } else if (op == BAM_CINS || op == BAM_CSOFT_CLIP) y += l;
  78                 else if (op == BAM_CDEL || op == BAM_CREF_SKIP) {
  79                         if (x + l > tpos) {
  80                                 *_tpos = is_left? x : x + l;
  81                                 return y;
  82                         }
  83                         x += l;
  84                 }
  85         }
  86         *_tpos = x;
  87         return last_y;
  88 }
  89 // FIXME: check if the inserted sequence is consistent with the homopolymer run
  90 // l is the relative gap length and l_run is the length of the homopolymer on the reference
  91 static inline int est_seqQ(const bcf_callaux_t *bca, int l, int l_run)
  92 {
  93         int q, qh;
  94         q = bca->openQ + bca->extQ * (abs(l) - 1);
  95         qh = l_run >= 3? (int)(bca->tandemQ * (double)abs(l) / l_run + .499) : 1000;
  96         return q < qh? q : qh;
  97 }
  98
  99 static inline int est_indelreg(int pos, const char *ref, int l, char *ins4)
 100 {
 101         int i, j, max = 0, max_i = pos, score = 0;
 102         l = abs(l);
 103         for (i = pos + 1, j = 0; ref[i]; ++i, ++j) {
 104                 if (ins4) score += (toupper(ref[i]) != "ACGTN"[(int)ins4[j%l]])? -10 : 1;
 105                 else score += (toupper(ref[i]) != toupper(ref[pos+1+j%l]))? -10 : 1;
 106                 if (score < 0) break;
 107                 if (max < score) max = score, max_i = i;
 108         }
 109         return max_i - pos;
 110 }
 111
 112 /*
 113  *  @n:  number of samples
 114  */
 115 int bcf_call_gap_prep(int n, int *n_plp, bam_pileup1_t **plp, int pos, bcf_callaux_t *bca, const char *ref,
 116                                           const void *rghash)
 117 {
 118         int i, s, j, k, t, n_types, *types, max_rd_len, left, right, max_ins, *score1, *score2, max_ref2;
 119         int N, K, l_run, ref_type, n_alt;
 120         char *inscns = 0, *ref2, *query, **ref_sample;
 121         khash_t(rg) *hash = (khash_t(rg)*)rghash;
 122         if (ref == 0 || bca == 0) return -1;
 123         // mark filtered reads
 124         if (rghash) {
 125                 N = 0;
 126                 for (s = N = 0; s < n; ++s) {
 127                         for (i = 0; i < n_plp[s]; ++i) {
 128                                 bam_pileup1_t *p = plp[s] + i;
 129                                 const uint8_t *rg = bam_aux_get(p->b, "RG");
 130                                 p->aux = 1; // filtered by default
 131                                 if (rg) {
 132                                         khint_t k = kh_get(rg, hash, (const char*)(rg + 1));
 133                                         if (k != kh_end(hash)) p->aux = 0, ++N; // not filtered
 134                                 }
 135                         }
 136                 }
 137                 if (N == 0) return -1; // no reads left
 138         }
 139         // determine if there is a gap
 140         for (s = N = 0; s < n; ++s) {
 141                 for (i = 0; i < n_plp[s]; ++i)
 142                         if (plp[s][i].indel != 0) break;
 143                 if (i < n_plp[s]) break;
 144         }
 145         if (s == n) return -1; // there is no indel at this position.
 146         for (s = N = 0; s < n; ++s) N += n_plp[s]; // N is the total number of reads
 147         { // find out how many types of indels are present
 148         bca->max_support = bca->max_frac = 0;
 149                 int m, n_alt = 0, n_tot = 0, indel_support_ok = 0;
 150                 uint32_t *aux;
 151                 aux = calloc(N + 1, 4);
 152                 m = max_rd_len = 0;
 153                 aux[m++] = MINUS_CONST; // zero indel is always a type
 154                 for (s = 0; s < n; ++s) {
 155             int na = 0, nt = 0;
 156                         for (i = 0; i < n_plp[s]; ++i) {
 157                                 const bam_pileup1_t *p = plp[s] + i;
 158                                 if (rghash == 0 || p->aux == 0) {
 159                                         ++nt;
 160                                         if (p->indel != 0) {
 161                                                 ++na;
 162                                                 aux[m++] = MINUS_CONST + p->indel;
 163                                         }
 164                                 }
 165                                 j = bam_cigar2qlen(&p->b->core, bam1_cigar(p->b));
 166                                 if (j > max_rd_len) max_rd_len = j;
 167                         }
 168             float frac = (float)na/nt;
 169             if ( !indel_support_ok && na >= bca->min_support && frac >= bca->min_frac )
 170                 indel_support_ok = 1;
 171             if ( na > bca->max_support && frac > 0 ) bca->max_support = na, bca->max_frac = frac;
 172             n_alt += na;
 173             n_tot += nt;
 174                 }
 175         // To prevent long stretches of N's to be mistaken for indels (sometimes thousands of bases),
 176         //  check the number of N's in the sequence and skip places where half or more reference bases are Ns.
 177         int nN=0; for (i=pos; i-pos<max_rd_len && ref[i]; i++) if ( ref[i]=='N' ) nN++;
 178         if ( nN*2>i ) { free(aux); return -1; }
 179
 180                 ks_introsort(uint32_t, m, aux);
 181                 // squeeze out identical types
 182                 for (i = 1, n_types = 1; i < m; ++i)
 183                         if (aux[i] != aux[i-1]) ++n_types;
 184         // Taking totals makes it hard to call rare indels
 185         if ( !bca->per_sample_flt )
 186             indel_support_ok = ( (float)n_alt / n_tot < bca->min_frac || n_alt < bca->min_support ) ? 0 : 1;
 187         if ( n_types == 1 || !indel_support_ok ) { // then skip
 188             free(aux); return -1;
 189         }
 190                 if (n_types >= 64) {
 191                         free(aux);
 192                         if (bam_verbose >= 2)
 193                                 fprintf(stderr, "[%s] excessive INDEL alleles at position %d. Skip the position.\n", __func__, pos + 1);
 194                         return -1;
 195                 }
 196                 types = (int*)calloc(n_types, sizeof(int));
 197                 t = 0;
 198                 types[t++] = aux[0] - MINUS_CONST;
 199                 for (i = 1; i < m; ++i)
 200                         if (aux[i] != aux[i-1])
 201                                 types[t++] = aux[i] - MINUS_CONST;
 202                 free(aux);
 203                 for (t = 0; t < n_types; ++t)
 204                         if (types[t] == 0) break;
 205                 ref_type = t; // the index of the reference type (0)
 206         }
 207         { // calculate left and right boundary
 208                 left = pos > INDEL_WINDOW_SIZE? pos - INDEL_WINDOW_SIZE : 0;
 209                 right = pos + INDEL_WINDOW_SIZE;
 210                 if (types[0] < 0) right -= types[0];
 211                 // in case the alignments stand out the reference
 212                 for (i = pos; i < right; ++i)
 213                         if (ref[i] == 0) break;
 214                 right = i;
 215         }
 216         /* The following block fixes a long-existing flaw in the INDEL
 217          * calling model: the interference of nearby SNPs. However, it also
 218          * reduces the power because sometimes, substitutions caused by
 219          * indels are not distinguishable from true mutations. Multiple
 220          * sequence realignment helps to increase the power.
 221          */
 222         { // construct per-sample consensus
 223                 int L = right - left + 1, max_i, max2_i;
 224                 uint32_t *cns, max, max2;
 225                 char *ref0, *r;
 226                 ref_sample = calloc(n, sizeof(void*));
 227                 cns = calloc(L, 4);
 228                 ref0 = calloc(L, 1);
 229                 for (i = 0; i < right - left; ++i)
 230                         ref0[i] = bam_nt16_table[(int)ref[i+left]];
 231                 for (s = 0; s < n; ++s) {
 232                         r = ref_sample[s] = calloc(L, 1);
 233                         memset(cns, 0, sizeof(int) * L);
 234                         // collect ref and non-ref counts
 235                         for (i = 0; i < n_plp[s]; ++i) {
 236                                 bam_pileup1_t *p = plp[s] + i;
 237                                 bam1_t *b = p->b;
 238                                 uint32_t *cigar = bam1_cigar(b);
 239                                 uint8_t *seq = bam1_seq(b);
 240                                 int x = b->core.pos, y = 0;
 241                                 for (k = 0; k < b->core.n_cigar; ++k) {
 242                                         int op = cigar[k]&0xf;
 243                                         int j, l = cigar[k]>>4;
 244                                         if (op == BAM_CMATCH || op == BAM_CEQUAL || op == BAM_CDIFF) {
 245                                                 for (j = 0; j < l; ++j)
 246                                                         if (x + j >= left && x + j < right)
 247                                                                 cns[x+j-left] += (bam1_seqi(seq, y+j) == ref0[x+j-left])? 1 : 0x10000;
 248                                                 x += l; y += l;
 249                                         } else if (op == BAM_CDEL || op == BAM_CREF_SKIP) x += l;
 250                                         else if (op == BAM_CINS || op == BAM_CSOFT_CLIP) y += l;
 251                                 }
 252                         }
 253                         // determine the consensus
 254                         for (i = 0; i < right - left; ++i) r[i] = ref0[i];
 255                         max = max2 = 0; max_i = max2_i = -1;
 256                         for (i = 0; i < right - left; ++i) {
 257                                 if (cns[i]>>16 >= max>>16) max2 = max, max2_i = max_i, max = cns[i], max_i = i;
 258                                 else if (cns[i]>>16 >= max2>>16) max2 = cns[i], max2_i = i;
 259                         }
 260                         if ((double)(max&0xffff) / ((max&0xffff) + (max>>16)) >= 0.7) max_i = -1;
 261                         if ((double)(max2&0xffff) / ((max2&0xffff) + (max2>>16)) >= 0.7) max2_i = -1;
 262                         if (max_i >= 0) r[max_i] = 15;
 263                         if (max2_i >= 0) r[max2_i] = 15;
 264 //                      for (i = 0; i < right - left; ++i) fputc("=ACMGRSVTWYHKDBN"[(int)r[i]], stderr); fputc('\n', stderr);
 265                 }
 266                 free(ref0); free(cns);
 267         }
 268         { // the length of the homopolymer run around the current position
 269                 int c = bam_nt16_table[(int)ref[pos + 1]];
 270                 if (c == 15) l_run = 1;
 271                 else {
 272                         for (i = pos + 2; ref[i]; ++i)
 273                                 if (bam_nt16_table[(int)ref[i]] != c) break;
 274                         l_run = i;
 275                         for (i = pos; i >= 0; --i)
 276                                 if (bam_nt16_table[(int)ref[i]] != c) break;
 277                         l_run -= i + 1;
 278                 }
 279         }
 280         // construct the consensus sequence
 281         max_ins = types[n_types - 1]; // max_ins is at least 0
 282         if (max_ins > 0) {
 283                 int *inscns_aux = calloc(4 * n_types * max_ins, sizeof(int));
 284                 // count the number of occurrences of each base at each position for each type of insertion
 285                 for (t = 0; t < n_types; ++t) {
 286                         if (types[t] > 0) {
 287                                 for (s = 0; s < n; ++s) {
 288                                         for (i = 0; i < n_plp[s]; ++i) {
 289                                                 bam_pileup1_t *p = plp[s] + i;
 290                                                 if (p->indel == types[t]) {
 291                                                         uint8_t *seq = bam1_seq(p->b);
 292                                                         for (k = 1; k <= p->indel; ++k) {
 293                                                                 int c = bam_nt16_nt4_table[bam1_seqi(seq, p->qpos + k)];
 294                                                                 if (c < 4) ++inscns_aux[(t*max_ins+(k-1))*4 + c];
 295                                                         }
 296                                                 }
 297                                         }
 298                                 }
 299                         }
 300                 }
 301                 // use the majority rule to construct the consensus
 302                 inscns = calloc(n_types * max_ins, 1);
 303                 for (t = 0; t < n_types; ++t) {
 304                         for (j = 0; j < types[t]; ++j) {
 305                                 int max = 0, max_k = -1, *ia = &inscns_aux[(t*max_ins+j)*4];
 306                                 for (k = 0; k < 4; ++k)
 307                                         if (ia[k] > max)
 308                                                 max = ia[k], max_k = k;
 309                                 inscns[t*max_ins + j] = max? max_k : 4;
 310                         }
 311                 }
 312                 free(inscns_aux);
 313         }
 314         // compute the likelihood given each type of indel for each read
 315         max_ref2 = right - left + 2 + 2 * (max_ins > -types[0]? max_ins : -types[0]);
 316         ref2  = calloc(max_ref2, 1);
 317         query = calloc(right - left + max_rd_len + max_ins + 2, 1);
 318         score1 = calloc(N * n_types, sizeof(int));
 319         score2 = calloc(N * n_types, sizeof(int));
 320         bca->indelreg = 0;
 321         for (t = 0; t < n_types; ++t) {
 322                 int l, ir;
 323                 kpa_par_t apf1 = { 1e-4, 1e-2, 10 }, apf2 = { 1e-6, 1e-3, 10 };
 324                 apf1.bw = apf2.bw = abs(types[t]) + 3;
 325                 // compute indelreg
 326                 if (types[t] == 0) ir = 0;
 327                 else if (types[t] > 0) ir = est_indelreg(pos, ref, types[t], &inscns[t*max_ins]);
 328                 else ir = est_indelreg(pos, ref, -types[t], 0);
 329                 if (ir > bca->indelreg) bca->indelreg = ir;
 330 //              fprintf(stderr, "%d, %d, %d\n", pos, types[t], ir);
 331                 // realignment
 332                 for (s = K = 0; s < n; ++s) {
 333                         // write ref2
 334                         for (k = 0, j = left; j <= pos; ++j)
 335                                 ref2[k++] = bam_nt16_nt4_table[(int)ref_sample[s][j-left]];
 336                         if (types[t] <= 0) j += -types[t];
 337                         else for (l = 0; l < types[t]; ++l)
 338                                          ref2[k++] = inscns[t*max_ins + l];
 339                         for (; j < right && ref[j]; ++j)
 340                                 ref2[k++] = bam_nt16_nt4_table[(int)ref_sample[s][j-left]];
 341                         for (; k < max_ref2; ++k) ref2[k] = 4;
 342                         if (j < right) right = j;
 343                         // align each read to ref2
 344                         for (i = 0; i < n_plp[s]; ++i, ++K) {
 345                                 bam_pileup1_t *p = plp[s] + i;
 346                                 int qbeg, qend, tbeg, tend, sc, kk;
 347                                 uint8_t *seq = bam1_seq(p->b);
 348                                 uint32_t *cigar = bam1_cigar(p->b);
 349                                 if (p->b->core.flag&4) continue; // unmapped reads
 350                                 // FIXME: the following loop should be better moved outside; nonetheless, realignment should be much slower anyway.
 351                                 for (kk = 0; kk < p->b->core.n_cigar; ++kk)
 352                                         if ((cigar[kk]&BAM_CIGAR_MASK) == BAM_CREF_SKIP) break;
 353                                 if (kk < p->b->core.n_cigar) continue;
 354                                 // FIXME: the following skips soft clips, but using them may be more sensitive.
 355                                 // determine the start and end of sequences for alignment
 356                                 qbeg = tpos2qpos(&p->b->core, bam1_cigar(p->b), left,  0, &tbeg);
 357                                 qend = tpos2qpos(&p->b->core, bam1_cigar(p->b), right, 1, &tend);
 358                                 if (types[t] < 0) {
 359                                         int l = -types[t];
 360                                         tbeg = tbeg - l > left?  tbeg - l : left;
 361                                 }
 362                                 // write the query sequence
 363                                 for (l = qbeg; l < qend; ++l)
 364                                         query[l - qbeg] = bam_nt16_nt4_table[bam1_seqi(seq, l)];
 365                                 { // do realignment; this is the bottleneck
 366                                         const uint8_t *qual = bam1_qual(p->b), *bq;
 367                                         uint8_t *qq;
 368                                         qq = calloc(qend - qbeg, 1);
 369                                         bq = (uint8_t*)bam_aux_get(p->b, "ZQ");
 370                                         if (bq) ++bq; // skip type
 371                                         for (l = qbeg; l < qend; ++l) {
 372                                                 qq[l - qbeg] = bq? qual[l] + (bq[l] - 64) : qual[l];
 373                                                 if (qq[l - qbeg] > 30) qq[l - qbeg] = 30;
 374                                                 if (qq[l - qbeg] < 7) qq[l - qbeg] = 7;
 375                                         }
 376                                         sc = kpa_glocal((uint8_t*)ref2 + tbeg - left, tend - tbeg + abs(types[t]),
 377                                                                         (uint8_t*)query, qend - qbeg, qq, &apf1, 0, 0);
 378                                         l = (int)(100. * sc / (qend - qbeg) + .499); // used for adjusting indelQ below
 379                                         if (l > 255) l = 255;
 380                                         score1[K*n_types + t] = score2[K*n_types + t] = sc<<8 | l;
 381                                         if (sc > 5) {
 382                                                 sc = kpa_glocal((uint8_t*)ref2 + tbeg - left, tend - tbeg + abs(types[t]),
 383                                                                                 (uint8_t*)query, qend - qbeg, qq, &apf2, 0, 0);
 384                                                 l = (int)(100. * sc / (qend - qbeg) + .499);
 385                                                 if (l > 255) l = 255;
 386                                                 score2[K*n_types + t] = sc<<8 | l;
 387                                         }
 388                                         free(qq);
 389                                 }
 390 /*
 391                                 for (l = 0; l < tend - tbeg + abs(types[t]); ++l)
 392                                         fputc("ACGTN"[(int)ref2[tbeg-left+l]], stderr);
 393                                 fputc('\n', stderr);
 394                                 for (l = 0; l < qend - qbeg; ++l) fputc("ACGTN"[(int)query[l]], stderr);
 395                                 fputc('\n', stderr);
 396                                 fprintf(stderr, "pos=%d type=%d read=%d:%d name=%s qbeg=%d tbeg=%d score=%d\n", pos, types[t], s, i, bam1_qname(p->b), qbeg, tbeg, sc);
 397 */
 398                         }
 399                 }
 400         }
 401         free(ref2); free(query);
 402         { // compute indelQ
 403                 int *sc, tmp, *sumq;
 404                 sc   = alloca(n_types * sizeof(int));
 405                 sumq = alloca(n_types * sizeof(int));
 406                 memset(sumq, 0, sizeof(int) * n_types);
 407                 for (s = K = 0; s < n; ++s) {
 408                         for (i = 0; i < n_plp[s]; ++i, ++K) {
 409                                 bam_pileup1_t *p = plp[s] + i;
 410                                 int *sct = &score1[K*n_types], indelQ1, indelQ2, seqQ, indelQ;
 411                                 for (t = 0; t < n_types; ++t) sc[t] = sct[t]<<6 | t;
 412                                 for (t = 1; t < n_types; ++t) // insertion sort
 413                                         for (j = t; j > 0 && sc[j] < sc[j-1]; --j)
 414                                                 tmp = sc[j], sc[j] = sc[j-1], sc[j-1] = tmp;
 415                                 /* errmod_cal() assumes that if the call is wrong, the
 416                                  * likelihoods of other events are equal. This is about
 417                                  * right for substitutions, but is not desired for
 418                                  * indels. To reuse errmod_cal(), I have to make
 419                                  * compromise for multi-allelic indels.
 420                                  */
 421                                 if ((sc[0]&0x3f) == ref_type) {
 422                                         indelQ1 = (sc[1]>>14) - (sc[0]>>14);
 423                                         seqQ = est_seqQ(bca, types[sc[1]&0x3f], l_run);
 424                                 } else {
 425                                         for (t = 0; t < n_types; ++t) // look for the reference type
 426                                                 if ((sc[t]&0x3f) == ref_type) break;
 427                                         indelQ1 = (sc[t]>>14) - (sc[0]>>14);
 428                                         seqQ = est_seqQ(bca, types[sc[0]&0x3f], l_run);
 429                                 }
 430                                 tmp = sc[0]>>6 & 0xff;
 431                                 indelQ1 = tmp > 111? 0 : (int)((1. - tmp/111.) * indelQ1 + .499); // reduce indelQ
 432                                 sct = &score2[K*n_types];
 433                                 for (t = 0; t < n_types; ++t) sc[t] = sct[t]<<6 | t;
 434                                 for (t = 1; t < n_types; ++t) // insertion sort
 435                                         for (j = t; j > 0 && sc[j] < sc[j-1]; --j)
 436                                                 tmp = sc[j], sc[j] = sc[j-1], sc[j-1] = tmp;
 437                                 if ((sc[0]&0x3f) == ref_type) {
 438                                         indelQ2 = (sc[1]>>14) - (sc[0]>>14);
 439                                 } else {
 440                                         for (t = 0; t < n_types; ++t) // look for the reference type
 441                                                 if ((sc[t]&0x3f) == ref_type) break;
 442                                         indelQ2 = (sc[t]>>14) - (sc[0]>>14);
 443                                 }
 444                                 tmp = sc[0]>>6 & 0xff;
 445                                 indelQ2 = tmp > 111? 0 : (int)((1. - tmp/111.) * indelQ2 + .499);
 446                                 // pick the smaller between indelQ1 and indelQ2
 447                                 indelQ = indelQ1 < indelQ2? indelQ1 : indelQ2;
 448                                 if (indelQ > 255) indelQ = 255;
 449                                 if (seqQ > 255) seqQ = 255;
 450                                 p->aux = (sc[0]&0x3f)<<16 | seqQ<<8 | indelQ; // use 22 bits in total
 451                                 sumq[sc[0]&0x3f] += indelQ < seqQ? indelQ : seqQ;
 452 //                              fprintf(stderr, "pos=%d read=%d:%d name=%s call=%d indelQ=%d seqQ=%d\n", pos, s, i, bam1_qname(p->b), types[sc[0]&0x3f], indelQ, seqQ);
 453                         }
 454                 }
 455                 // determine bca->indel_types[] and bca->inscns
 456                 bca->maxins = max_ins;
 457                 bca->inscns = realloc(bca->inscns, bca->maxins * 4);
 458                 for (t = 0; t < n_types; ++t)
 459                         sumq[t] = sumq[t]<<6 | t;
 460                 for (t = 1; t < n_types; ++t) // insertion sort
 461                         for (j = t; j > 0 && sumq[j] > sumq[j-1]; --j)
 462                                 tmp = sumq[j], sumq[j] = sumq[j-1], sumq[j-1] = tmp;
 463                 for (t = 0; t < n_types; ++t) // look for the reference type
 464                         if ((sumq[t]&0x3f) == ref_type) break;
 465                 if (t) { // then move the reference type to the first
 466                         tmp = sumq[t];
 467                         for (; t > 0; --t) sumq[t] = sumq[t-1];
 468                         sumq[0] = tmp;
 469                 }
 470                 for (t = 0; t < 4; ++t) bca->indel_types[t] = B2B_INDEL_NULL;
 471                 for (t = 0; t < 4 && t < n_types; ++t) {
 472                         bca->indel_types[t] = types[sumq[t]&0x3f];
 473                         memcpy(&bca->inscns[t * bca->maxins], &inscns[(sumq[t]&0x3f) * max_ins], bca->maxins);
 474                 }
 475                 // update p->aux
 476                 for (s = n_alt = 0; s < n; ++s) {
 477                         for (i = 0; i < n_plp[s]; ++i) {
 478                                 bam_pileup1_t *p = plp[s] + i;
 479                                 int x = types[p->aux>>16&0x3f];
 480                                 for (j = 0; j < 4; ++j)
 481                                         if (x == bca->indel_types[j]) break;
 482                                 p->aux = j<<16 | (j == 4? 0 : (p->aux&0xffff));
 483                                 if ((p->aux>>16&0x3f) > 0) ++n_alt;
 484 //                              fprintf(stderr, "X pos=%d read=%d:%d name=%s call=%d type=%d q=%d seqQ=%d\n", pos, s, i, bam1_qname(p->b), p->aux>>16&63, bca->indel_types[p->aux>>16&63], p->aux&0xff, p->aux>>8&0xff);
 485                         }
 486                 }
 487         }
 488         free(score1); free(score2);
 489         // free
 490         for (i = 0; i < n; ++i) free(ref_sample[i]);
 491         free(ref_sample);
 492         free(types); free(inscns);
 493         return n_alt > 0? 0 : -1;
 494 }