bam_maqcns.c

   1 #include <math.h>
   2 #include <assert.h>
   3 #include "bam.h"
   4 #include "bam_maqcns.h"
   5 #include "ksort.h"
   6 #include "kaln.h"
   7 KSORT_INIT_GENERIC(uint32_t)
   8
   9 #define INDEL_WINDOW_SIZE 50
  10 #define INDEL_EXT_DEP 0.9
  11
  12 typedef struct __bmc_aux_t {
  13         int max;
  14         uint32_t *info;
  15 } bmc_aux_t;
  16
  17 typedef struct {
  18         float esum[4], fsum[4];
  19         uint32_t c[4];
  20         uint32_t rms_mapQ;
  21 } glf_call_aux_t;
  22
  23 char bam_nt16_nt4_table[] = { 4, 0, 1, 4, 2, 4, 4, 4, 3, 4, 4, 4, 4, 4, 4, 4 };
  24
  25 /*
  26   P(<b1,b2>) = \theta \sum_{i=1}^{N-1} 1/i
  27   P(D|<b1,b2>) = \sum_{k=1}^{N-1} p_k 1/2 [(k/N)^n_2(1-k/N)^n_1 + (k/N)^n1(1-k/N)^n_2]
  28   p_k = 1/k / \sum_{i=1}^{N-1} 1/i
  29  */
  30 static void cal_het(bam_maqcns_t *aa)
  31 {
  32         int k, n1, n2;
  33         double sum_harmo; // harmonic sum
  34         double poly_rate;
  35
  36         free(aa->lhet);
  37         aa->lhet = (double*)calloc(256 * 256, sizeof(double));
  38         sum_harmo = 0.0;
  39         for (k = 1; k <= aa->n_hap - 1; ++k)
  40                 sum_harmo += 1.0 / k;
  41         for (n1 = 0; n1 < 256; ++n1) {
  42                 for (n2 = 0; n2 < 256; ++n2) {
  43                         long double sum = 0.0;
  44                         double lC = aa->is_soap? 0 : lgamma(n1+n2+1) - lgamma(n1+1) - lgamma(n2+1); // \binom{n1+n2}{n1}
  45                         for (k = 1; k <= aa->n_hap - 1; ++k) {
  46                                 double pk = 1.0 / k / sum_harmo;
  47                                 double log1 = log((double)k/aa->n_hap);
  48                                 double log2 = log(1.0 - (double)k/aa->n_hap);
  49                                 sum += pk * 0.5 * (expl(log1*n2) * expl(log2*n1) + expl(log1*n1) * expl(log2*n2));
  50                         }
  51                         aa->lhet[n1<<8|n2] = lC + logl(sum);
  52                 }
  53         }
  54         poly_rate = aa->het_rate * sum_harmo;
  55         aa->q_r = -4.343 * log(2.0 * poly_rate / (1.0 - poly_rate));
  56 }
  57
  58 /** initialize the helper structure */
  59 static void cal_coef(bam_maqcns_t *aa)
  60 {
  61         int k, n, q;
  62         long double sum_a[257], b[256], q_c[256], tmp[256], fk2[256];
  63         double *lC;
  64
  65         // aa->lhet will be allocated and initialized
  66         free(aa->fk); free(aa->coef);
  67         aa->coef = 0;
  68         aa->fk = (double*)calloc(256, sizeof(double));
  69         aa->fk[0] = fk2[0] = 1.0;
  70         for (n = 1; n != 256; ++n) {
  71                 aa->fk[n] = pow(aa->theta, n) * (1.0 - aa->eta) + aa->eta;
  72                 fk2[n] = aa->fk[n>>1]; // this is an approximation, assuming reads equally likely come from both strands
  73         }
  74         if (aa->is_soap) return;
  75         aa->coef = (double*)calloc(256*256*64, sizeof(double));
  76         lC = (double*)calloc(256 * 256, sizeof(double));
  77         for (n = 1; n != 256; ++n)
  78                 for (k = 1; k <= n; ++k)
  79                         lC[n<<8|k] = lgamma(n+1) - lgamma(k+1) - lgamma(n-k+1);
  80         for (q = 1; q != 64; ++q) {
  81                 double e = pow(10.0, -q/10.0);
  82                 double le = log(e);
  83                 double le1 = log(1.0-e);
  84                 for (n = 1; n != 256; ++n) {
  85                         double *coef = aa->coef + (q<<16|n<<8);
  86                         sum_a[n+1] = 0.0;
  87                         for (k = n; k >= 0; --k) { // a_k = \sum_{i=k}^n C^n_k \epsilon^k (1-\epsilon)^{n-k}
  88                                 sum_a[k] = sum_a[k+1] + expl(lC[n<<8|k] + k*le + (n-k)*le1);
  89                                 b[k] = sum_a[k+1] / sum_a[k];
  90                                 if (b[k] > 0.99) b[k] = 0.99;
  91                         }
  92                         for (k = 0; k != n; ++k) // log(\bar\beta_{nk}(\bar\epsilon)^{f_k})
  93                                 q_c[k] = -4.343 * fk2[k] * logl(b[k] / e);
  94                         for (k = 1; k != n; ++k) q_c[k] += q_c[k-1]; // \prod_{i=0}^k c_i
  95                         for (k = 0; k <= n; ++k) { // powl() in 64-bit mode seems broken on my Mac OS X 10.4.9
  96                                 tmp[k] = -4.343 * logl(1.0 - expl(fk2[k] * logl(b[k])));
  97                                 coef[k] = (k? q_c[k-1] : 0) + tmp[k]; // this is the final c_{nk}
  98                         }
  99                 }
 100         }
 101         free(lC);
 102 }
 103
 104 bam_maqcns_t *bam_maqcns_init()
 105 {
 106         bam_maqcns_t *bm;
 107         bm = (bam_maqcns_t*)calloc(1, sizeof(bam_maqcns_t));
 108         bm->aux = (bmc_aux_t*)calloc(1, sizeof(bmc_aux_t));
 109         bm->het_rate = 0.001;
 110         bm->theta = 0.85;
 111         bm->n_hap = 2;
 112         bm->eta = 0.03;
 113         bm->cap_mapQ = 60;
 114         return bm;
 115 }
 116
 117 void bam_maqcns_prepare(bam_maqcns_t *bm)
 118 {
 119         cal_coef(bm); cal_het(bm);
 120 }
 121
 122 void bam_maqcns_destroy(bam_maqcns_t *bm)
 123 {
 124         if (bm == 0) return;
 125         free(bm->lhet); free(bm->fk); free(bm->coef); free(bm->aux->info);
 126         free(bm->aux); free(bm);
 127 }
 128
 129 glf1_t *bam_maqcns_glfgen(int _n, const bam_pileup1_t *pl, uint8_t ref_base, bam_maqcns_t *bm)
 130 {
 131         glf_call_aux_t *b;
 132         int i, j, k, w[8], c, n;
 133         glf1_t *g = (glf1_t*)calloc(1, sizeof(glf1_t));
 134         float p[16], min_p = 1e30;
 135         uint64_t rms;
 136
 137         g->ref_base = ref_base;
 138         if (_n == 0) return g;
 139
 140         // construct aux array
 141         if (bm->aux->max < _n) {
 142                 bm->aux->max = _n;
 143                 kroundup32(bm->aux->max);
 144                 bm->aux->info = (uint32_t*)realloc(bm->aux->info, 4 * bm->aux->max);
 145         }
 146         for (i = n = 0; i < _n; ++i) {
 147                 const bam_pileup1_t *p = pl + i;
 148                 uint32_t q, x = 0, qq;
 149                 if (p->is_del || (p->b->core.flag&BAM_FUNMAP)) continue;
 150                 q = (uint32_t)bam1_qual(p->b)[p->qpos];
 151                 x |= (uint32_t)bam1_strand(p->b) << 18 | q << 8 | p->b->core.qual;
 152                 if (p->b->core.qual < q) q = p->b->core.qual;
 153                 x |= q << 24;
 154                 qq = bam1_seqi(bam1_seq(p->b), p->qpos);
 155                 q = bam_nt16_nt4_table[qq? qq : ref_base];
 156                 if (!p->is_del && q < 4) x |= 1 << 21 | q << 16;
 157                 bm->aux->info[n++] = x;
 158         }
 159         ks_introsort(uint32_t, n, bm->aux->info);
 160         // generate esum and fsum
 161         b = (glf_call_aux_t*)calloc(1, sizeof(glf_call_aux_t));
 162         for (k = 0; k != 8; ++k) w[k] = 0;
 163         rms = 0;
 164         for (j = n - 1; j >= 0; --j) { // calculate esum and fsum
 165                 uint32_t info = bm->aux->info[j];
 166                 int tmp;
 167                 if (info>>24 < 4 && (info>>8&0x3f) != 0) info = 4<<24 | (info&0xffffff);
 168                 k = info>>16&7;
 169                 if (info>>24 > 0) {
 170                         b->esum[k&3] += bm->fk[w[k]] * (info>>24);
 171                         b->fsum[k&3] += bm->fk[w[k]];
 172                         if (w[k] < 0xff) ++w[k];
 173                         ++b->c[k&3];
 174                 }
 175                 tmp = (int)(info&0xff) < bm->cap_mapQ? (int)(info&0xff) : bm->cap_mapQ;
 176                 rms += tmp * tmp;
 177         }
 178         b->rms_mapQ = (uint8_t)(sqrt((double)rms / n) + .499);
 179         // rescale ->c[]
 180         for (j = c = 0; j != 4; ++j) c += b->c[j];
 181         if (c > 255) {
 182                 for (j = 0; j != 4; ++j) b->c[j] = (int)(254.0 * b->c[j] / c + 0.5);
 183                 for (j = c = 0; j != 4; ++j) c += b->c[j];
 184         }
 185         if (!bm->is_soap) {
 186                 // generate likelihood
 187                 for (j = 0; j != 4; ++j) {
 188                         // homozygous
 189                         float tmp1, tmp3;
 190                         int tmp2, bar_e;
 191                         for (k = 0, tmp1 = tmp3 = 0.0, tmp2 = 0; k != 4; ++k) {
 192                                 if (j == k) continue;
 193                                 tmp1 += b->esum[k]; tmp2 += b->c[k]; tmp3 += b->fsum[k];
 194                         }
 195                         if (tmp2) {
 196                                 bar_e = (int)(tmp1 / tmp3 + 0.5);
 197                                 if (bar_e < 4) bar_e = 4; // should not happen
 198                                 if (bar_e > 63) bar_e = 63;
 199                                 p[j<<2|j] = tmp1 + bm->coef[bar_e<<16|c<<8|tmp2];
 200                         } else p[j<<2|j] = 0.0; // all the bases are j
 201                         // heterozygous
 202                         for (k = j + 1; k < 4; ++k) {
 203                                 for (i = 0, tmp2 = 0, tmp1 = tmp3 = 0.0; i != 4; ++i) {
 204                                         if (i == j || i == k) continue;
 205                                         tmp1 += b->esum[i]; tmp2 += b->c[i]; tmp3 += b->fsum[i];
 206                                 }
 207                                 if (tmp2) {
 208                                         bar_e = (int)(tmp1 / tmp3 + 0.5);
 209                                         if (bar_e < 4) bar_e = 4;
 210                                         if (bar_e > 63) bar_e = 63;
 211                                         p[j<<2|k] = p[k<<2|j] = -4.343 * bm->lhet[b->c[j]<<8|b->c[k]] + tmp1 + bm->coef[bar_e<<16|c<<8|tmp2];
 212                                 } else p[j<<2|k] = p[k<<2|j] = -4.343 * bm->lhet[b->c[j]<<8|b->c[k]]; // all the bases are either j or k
 213                         }
 214                         //
 215                         for (k = 0; k != 4; ++k)
 216                                 if (p[j<<2|k] < 0.0) p[j<<2|k] = 0.0;
 217                 }
 218
 219                 { // fix p[k<<2|k]
 220                         float max1, max2, min1, min2;
 221                         int max_k, min_k;
 222                         max_k = min_k = -1;
 223                         max1 = max2 = -1.0; min1 = min2 = 1e30;
 224                         for (k = 0; k < 4; ++k) {
 225                                 if (b->esum[k] > max1) {
 226                                         max2 = max1; max1 = b->esum[k]; max_k = k;
 227                                 } else if (b->esum[k] > max2) max2 = b->esum[k];
 228                         }
 229                         for (k = 0; k < 4; ++k) {
 230                                 if (p[k<<2|k] < min1) {
 231                                         min2 = min1; min1 = p[k<<2|k]; min_k = k;
 232                                 } else if (p[k<<2|k] < min2) min2 = p[k<<2|k];
 233                         }
 234                         if (max1 > max2 && (min_k != max_k || min1 + 1.0 > min2))
 235                                 p[max_k<<2|max_k] = min1 > 1.0? min1 - 1.0 : 0.0;
 236                 }
 237         } else { // apply the SOAP model
 238                 // generate likelihood
 239                 for (j = 0; j != 4; ++j) {
 240                         float tmp;
 241                         // homozygous
 242                         for (k = 0, tmp = 0.0; k != 4; ++k)
 243                                 if (j != k) tmp += b->esum[k];
 244                         p[j<<2|j] = tmp;
 245                         // heterozygous
 246                         for (k = j + 1; k < 4; ++k) {
 247                                 for (i = 0, tmp = 0.0; i != 4; ++i)
 248                                         if (i != j && i != k) tmp += b->esum[i];
 249                                 p[j<<2|k] = p[k<<2|j] = -4.343 * bm->lhet[b->c[j]<<8|b->c[k]] + tmp;
 250                         }
 251                 }
 252         }
 253
 254         // convert necessary information to glf1_t
 255         g->ref_base = ref_base; g->max_mapQ = b->rms_mapQ;
 256         g->depth = n > 16777215? 16777215 : n;
 257         for (j = 0; j != 4; ++j)
 258                 for (k = j; k < 4; ++k)
 259                         if (p[j<<2|k] < min_p) min_p = p[j<<2|k];
 260         g->min_lk = min_p > 255.0? 255 : (int)(min_p + 0.5);
 261         for (j = c = 0; j != 4; ++j)
 262                 for (k = j; k < 4; ++k)
 263                         g->lk[c++] = p[j<<2|k]-min_p > 255.0? 255 : (int)(p[j<<2|k]-min_p + 0.5);
 264
 265         free(b);
 266         return g;
 267 }
 268
 269 uint32_t glf2cns(const glf1_t *g, int q_r)
 270 {
 271         int i, j, k, tmp[16], min = 10000, min2 = 10000, min3 = 10000, min_g = -1, min_g2 = -1;
 272         uint32_t x = 0;
 273         for (i = k = 0; i < 4; ++i)
 274                 for (j = i; j < 4; ++j) {
 275                         tmp[j<<2|i] = -1;
 276                         tmp[i<<2|j] = g->lk[k++] + (i == j? 0 : q_r);
 277                 }
 278         for (i = 0; i < 16; ++i) {
 279                 if (tmp[i] < 0) continue;
 280                 if (tmp[i] < min) {
 281                         min3 = min2; min2 = min; min = tmp[i]; min_g2 = min_g; min_g = i;
 282                 } else if (tmp[i] < min2) {
 283                         min3 = min2; min2 = tmp[i]; min_g2 = i;
 284                 } else if (tmp[i] < min3) min3 = tmp[i];
 285         }
 286         x = min_g >= 0? (1U<<(min_g>>2&3) | 1U<<(min_g&3)) << 28 : 0xf << 28;
 287         x |= min_g2 >= 0? (1U<<(min_g2>>2&3) | 1U<<(min_g2&3)) << 24 : 0xf << 24;
 288         x |= (uint32_t)g->max_mapQ << 16;
 289         x |= min2 < 10000? (min2 - min < 256? min2 - min : 255) << 8 : 0xff << 8;
 290         x |= min2 < 10000 && min3 < 10000? (min3 - min2 < 256? min3 - min2 : 255) : 0xff;
 291         return x;
 292 }
 293
 294 uint32_t bam_maqcns_call(int n, const bam_pileup1_t *pl, bam_maqcns_t *bm)
 295 {
 296         glf1_t *g;
 297         uint32_t x;
 298         if (n) {
 299                 g = bam_maqcns_glfgen(n, pl, 0xf, bm);
 300                 x = glf2cns(g, (int)(bm->q_r + 0.5));
 301                 free(g);
 302         } else x = 0xfU<<28 | 0xfU<<24;
 303         return x;
 304 }
 305
 306 /************** *****************/
 307
 308 bam_maqindel_opt_t *bam_maqindel_opt_init()
 309 {
 310         bam_maqindel_opt_t *mi = (bam_maqindel_opt_t*)calloc(1, sizeof(bam_maqindel_opt_t));
 311         mi->q_indel = 40;
 312         mi->r_indel = 0.00015;
 313         //
 314         mi->mm_penalty = 3;
 315         mi->indel_err = 4;
 316         mi->ambi_thres = 10;
 317         return mi;
 318 }
 319
 320 void bam_maqindel_ret_destroy(bam_maqindel_ret_t *mir)
 321 {
 322         if (mir == 0) return;
 323         free(mir->s[0]); free(mir->s[1]); free(mir);
 324 }
 325
 326 int bam_tpos2qpos(const bam1_core_t *c, const uint32_t *cigar, int32_t tpos, int is_left, int32_t *_tpos)
 327 {
 328         int k, x = c->pos, y = 0, last_y = 0;
 329         *_tpos = c->pos;
 330         for (k = 0; k < c->n_cigar; ++k) {
 331                 int op = cigar[k] & BAM_CIGAR_MASK;
 332                 int l = cigar[k] >> BAM_CIGAR_SHIFT;
 333                 if (op == BAM_CMATCH) {
 334                         if (c->pos > tpos) return y;
 335                         if (x + l > tpos) {
 336                                 *_tpos = tpos;
 337                                 return y + (tpos - x);
 338                         }
 339                         x += l; y += l;
 340                         last_y = y;
 341                 } else if (op == BAM_CINS || op == BAM_CSOFT_CLIP) y += l;
 342                 else if (op == BAM_CDEL || op == BAM_CREF_SKIP) {
 343                         if (x + l > tpos) {
 344                                 *_tpos = is_left? x : x + l;
 345                                 return y;
 346                         }
 347                         x += l;
 348                 }
 349         }
 350         *_tpos = x;
 351         return last_y;
 352 }
 353
 354 #define MINUS_CONST 0x10000000
 355
 356 bam_maqindel_ret_t *bam_maqindel(int n, int pos, const bam_maqindel_opt_t *mi, const bam_pileup1_t *pl, const char *ref,
 357                                                                  int _n_types, int *_types)
 358 {
 359         int i, j, n_types, *types, left, right, max_rd_len = 0;
 360         bam_maqindel_ret_t *ret = 0;
 361         // if there is no proposed indel, check if there is an indel from the alignment
 362         if (_n_types == 0) {
 363                 for (i = 0; i < n; ++i) {
 364                         const bam_pileup1_t *p = pl + i;
 365                         if (!(p->b->core.flag&BAM_FUNMAP) && p->indel != 0) break;
 366                 }
 367                 if (i == n) return 0; // no indel
 368         }
 369         { // calculate how many types of indels are available (set n_types and types)
 370                 int m;
 371                 uint32_t *aux;
 372                 aux = (uint32_t*)calloc(n + _n_types + 1, 4);
 373                 m = 0;
 374                 aux[m++] = MINUS_CONST; // zero indel is always a type
 375                 for (i = 0; i < n; ++i) {
 376                         const bam_pileup1_t *p = pl + i;
 377                         if (!(p->b->core.flag&BAM_FUNMAP) && p->indel != 0)
 378                                 aux[m++] = MINUS_CONST + p->indel;
 379                         j = bam_cigar2qlen(&p->b->core, bam1_cigar(p->b));
 380                         if (j > max_rd_len) max_rd_len = j;
 381                 }
 382                 if (_n_types) // then also add this to aux[]
 383                         for (i = 0; i < _n_types; ++i)
 384                                 if (_types[i]) aux[m++] = MINUS_CONST + _types[i];
 385                 ks_introsort(uint32_t, m, aux);
 386                 // squeeze out identical types
 387                 for (i = 1, n_types = 1; i < m; ++i)
 388                         if (aux[i] != aux[i-1]) ++n_types;
 389                 types = (int*)calloc(n_types, sizeof(int));
 390                 j = 0;
 391                 types[j++] = aux[0] - MINUS_CONST;
 392                 for (i = 1; i < m; ++i) {
 393                         if (aux[i] != aux[i-1])
 394                                 types[j++] = aux[i] - MINUS_CONST;
 395                 }
 396                 free(aux);
 397         }
 398         { // calculate left and right boundary
 399                 left = pos > INDEL_WINDOW_SIZE? pos - INDEL_WINDOW_SIZE : 0;
 400                 right = pos + INDEL_WINDOW_SIZE;
 401                 if (types[0] < 0) right -= types[0];
 402                 // in case the alignments stand out the reference
 403                 for (i = pos; i < right; ++i)
 404                         if (ref[i] == 0) break;
 405                 right = i;
 406         }
 407         { // the core part
 408                 char *ref2, *rs, *inscns = 0;
 409                 int k, l, *score, *pscore, max_ins = types[n_types-1];
 410                 if (max_ins > 0) { // get the consensus of inserted sequences
 411                         int *inscns_aux = (int*)calloc(4 * n_types * max_ins, sizeof(int));
 412                         // count occurrences
 413                         for (i = 0; i < n_types; ++i) {
 414                                 if (types[i] <= 0) continue; // not insertion
 415                                 for (j = 0; j < n; ++j) {
 416                                         const bam_pileup1_t *p = pl + j;
 417                                         if (!(p->b->core.flag&BAM_FUNMAP) && p->indel == types[i]) {
 418                                                 for (k = 1; k <= p->indel; ++k) {
 419                                                         int c = bam_nt16_nt4_table[bam1_seqi(bam1_seq(p->b), p->qpos + k)];
 420                                                         if (c < 4) ++inscns_aux[i*max_ins*4 + (k-1)*4 + c];
 421                                                 }
 422                                         }
 423                                 }
 424                         }
 425                         // construct the consensus of inserted sequence
 426                         inscns = (char*)calloc(n_types * max_ins, sizeof(char));
 427                         for (i = 0; i < n_types; ++i) {
 428                                 for (j = 0; j < types[i]; ++j) {
 429                                         int max = 0, max_k = -1, *ia = inscns_aux + i*max_ins*4 + j*4;
 430                                         for (k = 0; k < 4; ++k) {
 431                                                 if (ia[k] > max) {
 432                                                         max = ia[k];
 433                                                         max_k = k;
 434                                                 }
 435                                         }
 436                                         inscns[i*max_ins + j] = max? 1<<max_k : 15;
 437                                 }
 438                         }
 439                         free(inscns_aux);
 440                 }
 441                 // calculate score
 442                 ref2 = (char*)calloc(right - left + types[n_types-1] + 2, 1);
 443                 rs   = (char*)calloc(right - left + max_rd_len + types[n_types-1] + 2, 1);
 444                 score = (int*)calloc(n_types * n, sizeof(int));
 445                 pscore = (int*)calloc(n_types * n, sizeof(int));
 446                 for (i = 0; i < n_types; ++i) {
 447                         ka_param_t ap = ka_param_blast;
 448                         ap.band_width = 2 * types[n_types - 1] + 2;
 449                         // write ref2
 450                         for (k = 0, j = left; j <= pos; ++j)
 451                                 ref2[k++] = bam_nt16_nt4_table[bam_nt16_table[(int)ref[j]]];
 452                         if (types[i] <= 0) j += -types[i];
 453                         else for (l = 0; l < types[i]; ++l)
 454                                          ref2[k++] = bam_nt16_nt4_table[(int)inscns[i*max_ins + l]];
 455                         for (; j < right && ref[j]; ++j)
 456                                 ref2[k++] = bam_nt16_nt4_table[bam_nt16_table[(int)ref[j]]];
 457                         if (j < right) right = j;
 458                         // calculate score for each read
 459                         for (j = 0; j < n; ++j) {
 460                                 const bam_pileup1_t *p = pl + j;
 461                                 int qbeg, qend, tbeg, tend;
 462                                 if (p->b->core.flag & BAM_FUNMAP) continue;
 463                                 qbeg = bam_tpos2qpos(&p->b->core, bam1_cigar(p->b), left,  0, &tbeg);
 464                                 qend = bam_tpos2qpos(&p->b->core, bam1_cigar(p->b), right, 1, &tend);
 465                                 assert(tbeg >= left);
 466                                 for (l = qbeg; l < qend; ++l)
 467                                         rs[l - qbeg] = bam_nt16_nt4_table[bam1_seqi(bam1_seq(p->b), l)];
 468                                 {
 469                                         int x, y, n_acigar, ps;
 470                                         uint32_t *acigar;
 471                                         ps = 0;
 472                                         if (tend - tbeg + types[i] <= 0) {
 473                                                 score[i*n+j] = -(1<<20);
 474                                                 pscore[i*n+j] = 1<<20;
 475                                                 continue;
 476                                         }
 477                                         acigar = ka_global_core((uint8_t*)ref2 + tbeg - left, tend - tbeg + types[i], (uint8_t*)rs, qend - qbeg, &ap, &score[i*n+j], &n_acigar);
 478                                         x = tbeg - left; y = 0;
 479                                         for (l = 0; l < n_acigar; ++l) {
 480                                                 int op = acigar[l]&0xf;
 481                                                 int len = acigar[l]>>4;
 482                                                 if (op == BAM_CMATCH) {
 483                                                         int k;
 484                                                         for (k = 0; k < len; ++k)
 485                                                                 if (ref2[x+k] != rs[y+k]) ps += bam1_qual(p->b)[y+k];
 486                                                         x += len; y += len;
 487                                                 } else if (op == BAM_CINS || op == BAM_CSOFT_CLIP) {
 488                                                         if (op == BAM_CINS) ps += mi->q_indel * len;
 489                                                         y += len;
 490                                                 } else if (op == BAM_CDEL) {
 491                                                         ps += mi->q_indel * len;
 492                                                         x += len;
 493                                                 }
 494                                         }
 495                                         pscore[i*n+j] = ps;
 496                                         /*if (pos == 2618517) { // for debugging only
 497                                                 fprintf(stderr, "pos=%d, type=%d, j=%d, score=%d, psore=%d, %d, %d, %d, %d, ", pos+1, types[i], j, score[i*n+j], pscore[i*n+j], tbeg, tend, qbeg, qend);
 498                                                 for (l = 0; l < n_acigar; ++l) fprintf(stderr, "%d%c", acigar[l]>>4, "MIDS"[acigar[l]&0xf]); fprintf(stderr, "\n");
 499                                                 for (l = 0; l < tend - tbeg + types[i]; ++l) fputc("ACGTN"[ref2[l]], stderr); fputc('\n', stderr);
 500                                                 for (l = 0; l < qend - qbeg; ++l) fputc("ACGTN"[rs[l]], stderr); fputc('\n', stderr);
 501                                                 }*/
 502                                         free(acigar);
 503                                 }
 504                         }
 505                 }
 506                 { // get final result
 507                         int *sum, max1, max2, max1_i, max2_i;
 508                         // pick up the best two score
 509                         sum = (int*)calloc(n_types, sizeof(int));
 510                         for (i = 0; i < n_types; ++i)
 511                                 for (j = 0; j < n; ++j)
 512                                         sum[i] += -pscore[i*n+j];
 513                         max1 = max2 = -0x7fffffff; max1_i = max2_i = -1;
 514                         for (i = 0; i < n_types; ++i) {
 515                                 if (sum[i] > max1) {
 516                                         max2 = max1; max2_i = max1_i; max1 = sum[i]; max1_i = i;
 517                                 } else if (sum[i] > max2) {
 518                                         max2 = sum[i]; max2_i = i;
 519                                 }
 520                         }
 521                         free(sum);
 522                         // write ret
 523                         ret = (bam_maqindel_ret_t*)calloc(1, sizeof(bam_maqindel_ret_t));
 524                         ret->indel1 = types[max1_i]; ret->indel2 = types[max2_i];
 525                         ret->s[0] = (char*)calloc(abs(ret->indel1) + 2, 1);
 526                         ret->s[1] = (char*)calloc(abs(ret->indel2) + 2, 1);
 527                         // write indel sequence
 528                         if (ret->indel1 > 0) {
 529                                 ret->s[0][0] = '+';
 530                                 for (k = 0; k < ret->indel1; ++k)
 531                                         ret->s[0][k+1] = bam_nt16_rev_table[(int)inscns[max1_i*max_ins + k]];
 532                         } else if (ret->indel1 < 0) {
 533                                 ret->s[0][0] = '-';
 534                                 for (k = 0; k < -ret->indel1 && ref[pos + k + 1]; ++k)
 535                                         ret->s[0][k+1] = ref[pos + k + 1];
 536                         } else ret->s[0][0] = '*';
 537                         if (ret->indel2 > 0) {
 538                                 ret->s[1][0] = '+';
 539                                 for (k = 0; k < ret->indel2; ++k)
 540                                         ret->s[1][k+1] = bam_nt16_rev_table[(int)inscns[max2_i*max_ins + k]];
 541                         } else if (ret->indel2 < 0) {
 542                                 ret->s[1][0] = '-';
 543                                 for (k = 0; k < -ret->indel2 && ref[pos + k + 1]; ++k)
 544                                         ret->s[1][k+1] = ref[pos + k + 1];
 545                         } else ret->s[1][0] = '*';
 546                         // write count
 547                         for (i = 0; i < n; ++i) {
 548                                 const bam_pileup1_t *p = pl + i;
 549                                 if (p->indel == ret->indel1) ++ret->cnt1;
 550                                 else if (p->indel == ret->indel2) ++ret->cnt2;
 551                                 else ++ret->cnt_anti;
 552                         }
 553                         { // write gl[]
 554                                 int tmp, seq_err = 0;
 555                                 double x = 1.0;
 556                                 tmp = max1_i - max2_i;
 557                                 if (tmp < 0) tmp = -tmp;
 558                                 for (j = 0; j < tmp + 1; ++j) x *= INDEL_EXT_DEP;
 559                                 seq_err = mi->q_indel * (1.0 - x) / (1.0 - INDEL_EXT_DEP);
 560                                 ret->gl[0] = ret->gl[1] = 0;
 561                                 for (j = 0; j < n; ++j) {
 562                                         int s1 = pscore[max1_i*n + j], s2 = pscore[max2_i*n + j];
 563                                         //printf("%d, %d, %d, %d, %d\n", pl[j].b->core.pos+1, max1_i, max2_i, s1, s2);
 564                                         if (s1 > s2) ret->gl[0] += s1 - s2 < seq_err? s1 - s2 : seq_err;
 565                                         else ret->gl[1] += s2 - s1 < seq_err? s2 - s1 : seq_err;
 566                                 }
 567                         }
 568                         // write cnt_ref and cnt_ambi
 569                         if (max1_i != 0 && max2_i != 0) {
 570                                 for (j = 0; j < n; ++j) {
 571                                         int diff1 = score[j] - score[max1_i * n + j];
 572                                         int diff2 = score[j] - score[max2_i * n + j];
 573                                         if (diff1 > 0 && diff2 > 0) ++ret->cnt_ref;
 574                                         else if (diff1 == 0 || diff2 == 0) ++ret->cnt_ambi;
 575                                 }
 576                         }
 577                 }
 578                 free(score); free(pscore); free(ref2); free(rs); free(inscns);
 579         }
 580         { // call genotype
 581                 int q[3], qr_indel = (int)(-4.343 * log(mi->r_indel) + 0.5);
 582                 int min1, min2, min1_i;
 583                 q[0] = ret->gl[0] + (ret->s[0][0] != '*'? 0 : 0) * qr_indel;
 584                 q[1] = ret->gl[1] + (ret->s[1][0] != '*'? 0 : 0) * qr_indel;
 585                 q[2] = n * 3 + (ret->s[0][0] == '*' || ret->s[1][0] == '*'? 1 : 1) * qr_indel;
 586                 min1 = min2 = 0x7fffffff; min1_i = -1;
 587                 for (i = 0; i < 3; ++i) {
 588                         if (q[i] < min1) {
 589                                 min2 = min1; min1 = q[i]; min1_i = i;
 590                         } else if (q[i] < min2) min2 = q[i];
 591                 }
 592                 ret->gt = min1_i;
 593                 ret->q_cns = min2 - min1;
 594                 // set q_ref
 595                 if (ret->gt < 2) ret->q_ref = (ret->s[ret->gt][0] == '*')? 0 : q[1-ret->gt] - q[ret->gt] - qr_indel - 3;
 596                 else ret->q_ref = (ret->s[0][0] == '*')? q[0] - q[2] : q[1] - q[2];
 597                 if (ret->q_ref < 0) ret->q_ref = 0;
 598         }
 599         free(types);
 600         return ret;
 601 }