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 int bcf_call_gap_prep(int n, int *n_plp, bam_pileup1_t **plp, int pos, bcf_callaux_t *bca, const char *ref,
 113                                           const void *rghash)
 114 {
 115         int i, s, j, k, t, n_types, *types, max_rd_len, left, right, max_ins, *score1, *score2, max_ref2;
 116         int N, K, l_run, ref_type, n_alt;
 117         char *inscns = 0, *ref2, *query, **ref_sample;
 118         khash_t(rg) *hash = (khash_t(rg)*)rghash;
 119         if (ref == 0 || bca == 0) return -1;
 120         // mark filtered reads
 121         if (rghash) {
 122                 N = 0;
 123                 for (s = N = 0; s < n; ++s) {
 124                         for (i = 0; i < n_plp[s]; ++i) {
 125                                 bam_pileup1_t *p = plp[s] + i;
 126                                 const uint8_t *rg = bam_aux_get(p->b, "RG");
 127                                 p->aux = 1; // filtered by default
 128                                 if (rg) {
 129                                         khint_t k = kh_get(rg, hash, (const char*)(rg + 1));
 130                                         if (k != kh_end(hash)) p->aux = 0, ++N; // not filtered
 131                                 }
 132                         }
 133                 }
 134                 if (N == 0) return -1; // no reads left
 135         }
 136         // determine if there is a gap
 137         for (s = N = 0; s < n; ++s) {
 138                 for (i = 0; i < n_plp[s]; ++i)
 139                         if (plp[s][i].indel != 0) break;
 140                 if (i < n_plp[s]) break;
 141         }
 142         if (s == n) return -1; // there is no indel at this position.
 143         for (s = N = 0; s < n; ++s) N += n_plp[s]; // N is the total number of reads
 144         { // find out how many types of indels are present
 145                 int m, n_alt = 0, n_tot = 0;
 146                 uint32_t *aux;
 147                 aux = calloc(N + 1, 4);
 148                 m = max_rd_len = 0;
 149                 aux[m++] = MINUS_CONST; // zero indel is always a type
 150                 for (s = 0; s < n; ++s) {
 151                         for (i = 0; i < n_plp[s]; ++i) {
 152                                 const bam_pileup1_t *p = plp[s] + i;
 153                                 if (rghash == 0 || p->aux == 0) {
 154                                         ++n_tot;
 155                                         if (p->indel != 0) {
 156                                                 ++n_alt;
 157                                                 aux[m++] = MINUS_CONST + p->indel;
 158                                         }
 159                                 }
 160                                 j = bam_cigar2qlen(&p->b->core, bam1_cigar(p->b));
 161                                 if (j > max_rd_len) max_rd_len = j;
 162                         }
 163                 }
 164         // To prevent long stretches of N's to be mistaken for indels (sometimes thousands of bases),
 165         //  check the number of N's in the sequence and skip places where half or more reference bases are Ns.
 166         int nN=0; for (i=pos; i-pos<max_rd_len && ref[i]; i++) if ( ref[i]=='N' ) nN++;
 167         if ( nN*2>i ) return -1;
 168
 169                 ks_introsort(uint32_t, m, aux);
 170                 // squeeze out identical types
 171                 for (i = 1, n_types = 1; i < m; ++i)
 172                         if (aux[i] != aux[i-1]) ++n_types;
 173                 if (n_types == 1 || (double)n_alt / n_tot < bca->min_frac || n_alt < bca->min_support) { // then skip
 174                         free(aux); return -1;
 175                 }
 176                 if (n_types >= 64) {
 177                         free(aux);
 178                         if (bam_verbose >= 2)
 179                                 fprintf(stderr, "[%s] excessive INDEL alleles at position %d. Skip the position.\n", __func__, pos + 1);
 180                         return -1;
 181                 }
 182                 types = (int*)calloc(n_types, sizeof(int));
 183                 t = 0;
 184                 types[t++] = aux[0] - MINUS_CONST;
 185                 for (i = 1; i < m; ++i)
 186                         if (aux[i] != aux[i-1])
 187                                 types[t++] = aux[i] - MINUS_CONST;
 188                 free(aux);
 189                 for (t = 0; t < n_types; ++t)
 190                         if (types[t] == 0) break;
 191                 ref_type = t; // the index of the reference type (0)
 192         }
 193         { // calculate left and right boundary
 194                 left = pos > INDEL_WINDOW_SIZE? pos - INDEL_WINDOW_SIZE : 0;
 195                 right = pos + INDEL_WINDOW_SIZE;
 196                 if (types[0] < 0) right -= types[0];
 197                 // in case the alignments stand out the reference
 198                 for (i = pos; i < right; ++i)
 199                         if (ref[i] == 0) break;
 200                 right = i;
 201         }
 202         /* The following block fixes a long-existing flaw in the INDEL
 203          * calling model: the interference of nearby SNPs. However, it also
 204          * reduces the power because sometimes, substitutions caused by
 205          * indels are not distinguishable from true mutations. Multiple
 206          * sequence realignment helps to increase the power.
 207          */
 208         { // construct per-sample consensus
 209                 int L = right - left + 1, max_i, max2_i;
 210                 uint32_t *cns, max, max2;
 211                 char *ref0, *r;
 212                 ref_sample = calloc(n, sizeof(void*));
 213                 cns = calloc(L, 4);
 214                 ref0 = calloc(L, 1);
 215                 for (i = 0; i < right - left; ++i)
 216                         ref0[i] = bam_nt16_table[(int)ref[i+left]];
 217                 for (s = 0; s < n; ++s) {
 218                         r = ref_sample[s] = calloc(L, 1);
 219                         memset(cns, 0, sizeof(int) * L);
 220                         // collect ref and non-ref counts
 221                         for (i = 0; i < n_plp[s]; ++i) {
 222                                 bam_pileup1_t *p = plp[s] + i;
 223                                 bam1_t *b = p->b;
 224                                 uint32_t *cigar = bam1_cigar(b);
 225                                 uint8_t *seq = bam1_seq(b);
 226                                 int x = b->core.pos, y = 0;
 227                                 for (k = 0; k < b->core.n_cigar; ++k) {
 228                                         int op = cigar[k]&0xf;
 229                                         int j, l = cigar[k]>>4;
 230                                         if (op == BAM_CMATCH || op == BAM_CEQUAL || op == BAM_CDIFF) {
 231                                                 for (j = 0; j < l; ++j)
 232                                                         if (x + j >= left && x + j < right)
 233                                                                 cns[x+j-left] += (bam1_seqi(seq, y+j) == ref0[x+j-left])? 1 : 0x10000;
 234                                                 x += l; y += l;
 235                                         } else if (op == BAM_CDEL || op == BAM_CREF_SKIP) x += l;
 236                                         else if (op == BAM_CINS || op == BAM_CSOFT_CLIP) y += l;
 237                                 }
 238                         }
 239                         // determine the consensus
 240                         for (i = 0; i < right - left; ++i) r[i] = ref0[i];
 241                         max = max2 = 0; max_i = max2_i = -1;
 242                         for (i = 0; i < right - left; ++i) {
 243                                 if (cns[i]>>16 >= max>>16) max2 = max, max2_i = max_i, max = cns[i], max_i = i;
 244                                 else if (cns[i]>>16 >= max2>>16) max2 = cns[i], max2_i = i;
 245                         }
 246                         if ((double)(max&0xffff) / ((max&0xffff) + (max>>16)) >= 0.7) max_i = -1;
 247                         if ((double)(max2&0xffff) / ((max2&0xffff) + (max2>>16)) >= 0.7) max2_i = -1;
 248                         if (max_i >= 0) r[max_i] = 15;
 249                         if (max2_i >= 0) r[max2_i] = 15;
 250 //                      for (i = 0; i < right - left; ++i) fputc("=ACMGRSVTWYHKDBN"[(int)r[i]], stderr); fputc('\n', stderr);
 251                 }
 252                 free(ref0); free(cns);
 253         }
 254         { // the length of the homopolymer run around the current position
 255                 int c = bam_nt16_table[(int)ref[pos + 1]];
 256                 if (c == 15) l_run = 1;
 257                 else {
 258                         for (i = pos + 2; ref[i]; ++i)
 259                                 if (bam_nt16_table[(int)ref[i]] != c) break;
 260                         l_run = i;
 261                         for (i = pos; i >= 0; --i)
 262                                 if (bam_nt16_table[(int)ref[i]] != c) break;
 263                         l_run -= i + 1;
 264                 }
 265         }
 266         // construct the consensus sequence
 267         max_ins = types[n_types - 1]; // max_ins is at least 0
 268         if (max_ins > 0) {
 269                 int *inscns_aux = calloc(4 * n_types * max_ins, sizeof(int));
 270                 // count the number of occurrences of each base at each position for each type of insertion
 271                 for (t = 0; t < n_types; ++t) {
 272                         if (types[t] > 0) {
 273                                 for (s = 0; s < n; ++s) {
 274                                         for (i = 0; i < n_plp[s]; ++i) {
 275                                                 bam_pileup1_t *p = plp[s] + i;
 276                                                 if (p->indel == types[t]) {
 277                                                         uint8_t *seq = bam1_seq(p->b);
 278                                                         for (k = 1; k <= p->indel; ++k) {
 279                                                                 int c = bam_nt16_nt4_table[bam1_seqi(seq, p->qpos + k)];
 280                                                                 if (c < 4) ++inscns_aux[(t*max_ins+(k-1))*4 + c];
 281                                                         }
 282                                                 }
 283                                         }
 284                                 }
 285                         }
 286                 }
 287                 // use the majority rule to construct the consensus
 288                 inscns = calloc(n_types * max_ins, 1);
 289                 for (t = 0; t < n_types; ++t) {
 290                         for (j = 0; j < types[t]; ++j) {
 291                                 int max = 0, max_k = -1, *ia = &inscns_aux[(t*max_ins+j)*4];
 292                                 for (k = 0; k < 4; ++k)
 293                                         if (ia[k] > max)
 294                                                 max = ia[k], max_k = k;
 295                                 inscns[t*max_ins + j] = max? max_k : 4;
 296                         }
 297                 }
 298                 free(inscns_aux);
 299         }
 300         // compute the likelihood given each type of indel for each read
 301         max_ref2 = right - left + 2 + 2 * (max_ins > -types[0]? max_ins : -types[0]);
 302         ref2  = calloc(max_ref2, 1);
 303         query = calloc(right - left + max_rd_len + max_ins + 2, 1);
 304         score1 = calloc(N * n_types, sizeof(int));
 305         score2 = calloc(N * n_types, sizeof(int));
 306         bca->indelreg = 0;
 307         for (t = 0; t < n_types; ++t) {
 308                 int l, ir;
 309                 kpa_par_t apf1 = { 1e-4, 1e-2, 10 }, apf2 = { 1e-6, 1e-3, 10 };
 310                 apf1.bw = apf2.bw = abs(types[t]) + 3;
 311                 // compute indelreg
 312                 if (types[t] == 0) ir = 0;
 313                 else if (types[t] > 0) ir = est_indelreg(pos, ref, types[t], &inscns[t*max_ins]);
 314                 else ir = est_indelreg(pos, ref, -types[t], 0);
 315                 if (ir > bca->indelreg) bca->indelreg = ir;
 316 //              fprintf(stderr, "%d, %d, %d\n", pos, types[t], ir);
 317                 // realignment
 318                 for (s = K = 0; s < n; ++s) {
 319                         // write ref2
 320                         for (k = 0, j = left; j <= pos; ++j)
 321                                 ref2[k++] = bam_nt16_nt4_table[(int)ref_sample[s][j-left]];
 322                         if (types[t] <= 0) j += -types[t];
 323                         else for (l = 0; l < types[t]; ++l)
 324                                          ref2[k++] = inscns[t*max_ins + l];
 325                         for (; j < right && ref[j]; ++j)
 326                                 ref2[k++] = bam_nt16_nt4_table[(int)ref_sample[s][j-left]];
 327                         for (; k < max_ref2; ++k) ref2[k] = 4;
 328                         if (j < right) right = j;
 329                         // align each read to ref2
 330                         for (i = 0; i < n_plp[s]; ++i, ++K) {
 331                                 bam_pileup1_t *p = plp[s] + i;
 332                                 int qbeg, qend, tbeg, tend, sc, kk;
 333                                 uint8_t *seq = bam1_seq(p->b);
 334                                 uint32_t *cigar = bam1_cigar(p->b);
 335                                 if (p->b->core.flag&4) continue; // unmapped reads
 336                                 // FIXME: the following loop should be better moved outside; nonetheless, realignment should be much slower anyway.
 337                                 for (kk = 0; kk < p->b->core.n_cigar; ++kk)
 338                                         if ((cigar[kk]&BAM_CIGAR_MASK) == BAM_CREF_SKIP) break;
 339                                 if (kk < p->b->core.n_cigar) continue;
 340                                 // FIXME: the following skips soft clips, but using them may be more sensitive.
 341                                 // determine the start and end of sequences for alignment
 342                                 qbeg = tpos2qpos(&p->b->core, bam1_cigar(p->b), left,  0, &tbeg);
 343                                 qend = tpos2qpos(&p->b->core, bam1_cigar(p->b), right, 1, &tend);
 344                                 if (types[t] < 0) {
 345                                         int l = -types[t];
 346                                         tbeg = tbeg - l > left?  tbeg - l : left;
 347                                 }
 348                                 // write the query sequence
 349                                 for (l = qbeg; l < qend; ++l)
 350                                         query[l - qbeg] = bam_nt16_nt4_table[bam1_seqi(seq, l)];
 351                                 { // do realignment; this is the bottleneck
 352                                         const uint8_t *qual = bam1_qual(p->b), *bq;
 353                                         uint8_t *qq;
 354                                         qq = calloc(qend - qbeg, 1);
 355                                         bq = (uint8_t*)bam_aux_get(p->b, "ZQ");
 356                                         if (bq) ++bq; // skip type
 357                                         for (l = qbeg; l < qend; ++l) {
 358                                                 qq[l - qbeg] = bq? qual[l] + (bq[l] - 64) : qual[l];
 359                                                 if (qq[l - qbeg] > 30) qq[l - qbeg] = 30;
 360                                                 if (qq[l - qbeg] < 7) qq[l - qbeg] = 7;
 361                                         }
 362                                         sc = kpa_glocal((uint8_t*)ref2 + tbeg - left, tend - tbeg + abs(types[t]),
 363                                                                         (uint8_t*)query, qend - qbeg, qq, &apf1, 0, 0);
 364                                         l = (int)(100. * sc / (qend - qbeg) + .499); // used for adjusting indelQ below
 365                                         if (l > 255) l = 255;
 366                                         score1[K*n_types + t] = score2[K*n_types + t] = sc<<8 | l;
 367                                         if (sc > 5) {
 368                                                 sc = kpa_glocal((uint8_t*)ref2 + tbeg - left, tend - tbeg + abs(types[t]),
 369                                                                                 (uint8_t*)query, qend - qbeg, qq, &apf2, 0, 0);
 370                                                 l = (int)(100. * sc / (qend - qbeg) + .499);
 371                                                 if (l > 255) l = 255;
 372                                                 score2[K*n_types + t] = sc<<8 | l;
 373                                         }
 374                                         free(qq);
 375                                 }
 376 /*
 377                                 for (l = 0; l < tend - tbeg + abs(types[t]); ++l)
 378                                         fputc("ACGTN"[(int)ref2[tbeg-left+l]], stderr);
 379                                 fputc('\n', stderr);
 380                                 for (l = 0; l < qend - qbeg; ++l) fputc("ACGTN"[(int)query[l]], stderr);
 381                                 fputc('\n', stderr);
 382                                 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);
 383 */
 384                         }
 385                 }
 386         }
 387         free(ref2); free(query);
 388         { // compute indelQ
 389                 int *sc, tmp, *sumq;
 390                 sc   = alloca(n_types * sizeof(int));
 391                 sumq = alloca(n_types * sizeof(int));
 392                 memset(sumq, 0, sizeof(int) * n_types);
 393                 for (s = K = 0; s < n; ++s) {
 394                         for (i = 0; i < n_plp[s]; ++i, ++K) {
 395                                 bam_pileup1_t *p = plp[s] + i;
 396                                 int *sct = &score1[K*n_types], indelQ1, indelQ2, seqQ, indelQ;
 397                                 for (t = 0; t < n_types; ++t) sc[t] = sct[t]<<6 | t;
 398                                 for (t = 1; t < n_types; ++t) // insertion sort
 399                                         for (j = t; j > 0 && sc[j] < sc[j-1]; --j)
 400                                                 tmp = sc[j], sc[j] = sc[j-1], sc[j-1] = tmp;
 401                                 /* errmod_cal() assumes that if the call is wrong, the
 402                                  * likelihoods of other events are equal. This is about
 403                                  * right for substitutions, but is not desired for
 404                                  * indels. To reuse errmod_cal(), I have to make
 405                                  * compromise for multi-allelic indels.
 406                                  */
 407                                 if ((sc[0]&0x3f) == ref_type) {
 408                                         indelQ1 = (sc[1]>>14) - (sc[0]>>14);
 409                                         seqQ = est_seqQ(bca, types[sc[1]&0x3f], l_run);
 410                                 } else {
 411                                         for (t = 0; t < n_types; ++t) // look for the reference type
 412                                                 if ((sc[t]&0x3f) == ref_type) break;
 413                                         indelQ1 = (sc[t]>>14) - (sc[0]>>14);
 414                                         seqQ = est_seqQ(bca, types[sc[0]&0x3f], l_run);
 415                                 }
 416                                 tmp = sc[0]>>6 & 0xff;
 417                                 indelQ1 = tmp > 111? 0 : (int)((1. - tmp/111.) * indelQ1 + .499); // reduce indelQ
 418                                 sct = &score2[K*n_types];
 419                                 for (t = 0; t < n_types; ++t) sc[t] = sct[t]<<6 | t;
 420                                 for (t = 1; t < n_types; ++t) // insertion sort
 421                                         for (j = t; j > 0 && sc[j] < sc[j-1]; --j)
 422                                                 tmp = sc[j], sc[j] = sc[j-1], sc[j-1] = tmp;
 423                                 if ((sc[0]&0x3f) == ref_type) {
 424                                         indelQ2 = (sc[1]>>14) - (sc[0]>>14);
 425                                 } else {
 426                                         for (t = 0; t < n_types; ++t) // look for the reference type
 427                                                 if ((sc[t]&0x3f) == ref_type) break;
 428                                         indelQ2 = (sc[t]>>14) - (sc[0]>>14);
 429                                 }
 430                                 tmp = sc[0]>>6 & 0xff;
 431                                 indelQ2 = tmp > 111? 0 : (int)((1. - tmp/111.) * indelQ2 + .499);
 432                                 // pick the smaller between indelQ1 and indelQ2
 433                                 indelQ = indelQ1 < indelQ2? indelQ1 : indelQ2;
 434                                 if (indelQ > 255) indelQ = 255;
 435                                 if (seqQ > 255) seqQ = 255;
 436                                 p->aux = (sc[0]&0x3f)<<16 | seqQ<<8 | indelQ; // use 22 bits in total
 437                                 sumq[sc[0]&0x3f] += indelQ < seqQ? indelQ : seqQ;
 438 //                              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);
 439                         }
 440                 }
 441                 // determine bca->indel_types[] and bca->inscns
 442                 bca->maxins = max_ins;
 443                 bca->inscns = realloc(bca->inscns, bca->maxins * 4);
 444                 for (t = 0; t < n_types; ++t)
 445                         sumq[t] = sumq[t]<<6 | t;
 446                 for (t = 1; t < n_types; ++t) // insertion sort
 447                         for (j = t; j > 0 && sumq[j] > sumq[j-1]; --j)
 448                                 tmp = sumq[j], sumq[j] = sumq[j-1], sumq[j-1] = tmp;
 449                 for (t = 0; t < n_types; ++t) // look for the reference type
 450                         if ((sumq[t]&0x3f) == ref_type) break;
 451                 if (t) { // then move the reference type to the first
 452                         tmp = sumq[t];
 453                         for (; t > 0; --t) sumq[t] = sumq[t-1];
 454                         sumq[0] = tmp;
 455                 }
 456                 for (t = 0; t < 4; ++t) bca->indel_types[t] = B2B_INDEL_NULL;
 457                 for (t = 0; t < 4 && t < n_types; ++t) {
 458                         bca->indel_types[t] = types[sumq[t]&0x3f];
 459                         memcpy(&bca->inscns[t * bca->maxins], &inscns[(sumq[t]&0x3f) * max_ins], bca->maxins);
 460                 }
 461                 // update p->aux
 462                 for (s = n_alt = 0; s < n; ++s) {
 463                         for (i = 0; i < n_plp[s]; ++i) {
 464                                 bam_pileup1_t *p = plp[s] + i;
 465                                 int x = types[p->aux>>16&0x3f];
 466                                 for (j = 0; j < 4; ++j)
 467                                         if (x == bca->indel_types[j]) break;
 468                                 p->aux = j<<16 | (j == 4? 0 : (p->aux&0xffff));
 469                                 if ((p->aux>>16&0x3f) > 0) ++n_alt;
 470 //                              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);
 471                         }
 472                 }
 473         }
 474         free(score1); free(score2);
 475         // free
 476         for (i = 0; i < n; ++i) free(ref_sample[i]);
 477         free(ref_sample);
 478         free(types); free(inscns);
 479         return n_alt > 0? 0 : -1;
 480 }