website/archive/binaries/mac/src/mafft/core/genalign11.c

   1 #include "mltaln.h"
   2 #include "dp.h"
   3
   4 #define DEBUG 0
   5 #define DEBUG2 0
   6 #define XXXXXXX    0
   7 #define USE_PENALTY_EX  1
   8
   9 static TLS int localstop;
  10
  11 #if 1
  12 static void match_calc( float *match, char **s1, char **s2, int i1, int lgth2 )
  13 {
  14         char tmpc = s1[0][i1];
  15         char *seq2 = s2[0];
  16
  17         while( lgth2-- )
  18                 *match++ = amino_dis[(int)tmpc][(int)*seq2++];
  19 }
  20 #else
  21 static void match_calc( float *match, char **s1, char **s2, int i1, int lgth2 )
  22 {
  23         int j;
  24
  25         for( j=0; j<lgth2; j++ )
  26                 match[j] = amino_dis[(*s1)[i1]][(*s2)[j]];
  27 }
  28 #endif
  29
  30 #if 0
  31 static void match_calc_bk( float *match, float **cpmx1, float **cpmx2, int i1, int lgth2, float **floatwork, int **intwork, int initialize )
  32 {
  33         int j, k, l;
  34         float scarr[26];
  35         float **cpmxpd = floatwork;
  36         int **cpmxpdn = intwork;
  37         int count = 0;
  38
  39         if( initialize )
  40         {
  41                 for( j=0; j<lgth2; j++ )
  42                 {
  43                         count = 0;
  44                         for( l=0; l<26; l++ )
  45                         {
  46                                 if( cpmx2[l][j] )
  47                                 {
  48                                         cpmxpd[count][j] = cpmx2[l][j];
  49                                         cpmxpdn[count][j] = l;
  50                                         count++;
  51                                 }
  52                         }
  53                         cpmxpdn[count][j] = -1;
  54                 }
  55         }
  56
  57         for( l=0; l<26; l++ )
  58         {
  59                 scarr[l] = 0.0;
  60                 for( k=0; k<26; k++ )
  61                         scarr[l] += n_dis[k][l] * cpmx1[k][i1];
  62         }
  63 #if 0
  64         {
  65                 float *fpt, **fptpt, *fpt2;
  66                 int *ipt, **iptpt;
  67                 fpt2 = match;
  68                 iptpt = cpmxpdn;
  69                 fptpt = cpmxpd;
  70                 while( lgth2-- )
  71                 {
  72                         *fpt2 = 0.0;
  73                         ipt=*iptpt,fpt=*fptpt;
  74                         while( *ipt > -1 )
  75                                 *fpt2 += scarr[*ipt++] * *fpt++;
  76                         fpt2++,iptpt++,fptpt++;
  77                 }
  78         }
  79 #else
  80         for( j=0; j<lgth2; j++ )
  81         {
  82                 match[j] = 0.0;
  83                 for( k=0; cpmxpdn[k][j]>-1; k++ )
  84                         match[j] += scarr[cpmxpdn[k][j]] * cpmxpd[k][j];
  85         }
  86 #endif
  87 }
  88 #endif
  89
  90 static float gentracking( float *lasthorizontalw, float *lastverticalw,
  91                                                 char **seq1, char **seq2,
  92                         char **mseq1, char **mseq2,
  93                         float **cpmx1, float **cpmx2,
  94                         int **ijpi, int **ijpj, int *off1pt, int *off2pt, int endi, int endj )
  95 {
  96         int i, j, l, iin, jin, lgth1, lgth2, k, limk;
  97         int ifi=0, jfi=0; // by D.Mathog
  98 //      char gap[] = "-";
  99         char *gap;
 100         gap = newgapstr;
 101         lgth1 = strlen( seq1[0] );
 102         lgth2 = strlen( seq2[0] );
 103
 104 #if 0
 105         for( i=0; i<lgth1; i++ )
 106         {
 107                 fprintf( stderr, "lastverticalw[%d] = %f\n", i, lastverticalw[i] );
 108         }
 109 #endif
 110
 111     for( i=0; i<lgth1+1; i++ )
 112     {
 113         ijpi[i][0] = localstop;
 114         ijpj[i][0] = localstop;
 115     }
 116     for( j=0; j<lgth2+1; j++ )
 117     {
 118         ijpi[0][j] = localstop;
 119         ijpj[0][j] = localstop;
 120     }
 121
 122         mseq1[0] += lgth1+lgth2;
 123         *mseq1[0] = 0;
 124         mseq2[0] += lgth1+lgth2;
 125         *mseq2[0] = 0;
 126         iin = endi; jin = endj;
 127         limk = lgth1+lgth2;
 128         for( k=0; k<=limk; k++ )
 129         {
 130
 131                 ifi = ( ijpi[iin][jin] );
 132                 jfi = ( ijpj[iin][jin] );
 133                 l = iin - ifi;
 134 //              if( ijpi[iin][jin] < 0 || ijpj[iin][jin] < 0 )
 135 //              {
 136 //                      fprintf( stderr, "skip! %d-%d\n", ijpi[iin][jin], ijpj[iin][jin] );
 137 //                      fprintf( stderr, "1: %c-%c\n", seq1[0][iin], seq1[0][ifi] );
 138 //                      fprintf( stderr, "2: %c-%c\n", seq2[0][jin], seq2[0][jfi] );
 139 //              }
 140                 while( --l )
 141                 {
 142                         *--mseq1[0] = seq1[0][ifi+l];
 143                         *--mseq2[0] = *gap;
 144                         k++;
 145                 }
 146                 l= jin - jfi;
 147                 while( --l )
 148                 {
 149                         *--mseq1[0] = *gap;
 150                         *--mseq2[0] = seq2[0][jfi+l];
 151                         k++;
 152                 }
 153
 154                 if( iin <= 0 || jin <= 0 ) break;
 155                 *--mseq1[0] = seq1[0][ifi];
 156                 *--mseq2[0] = seq2[0][jfi];
 157
 158                 if( ijpi[ifi][jfi] == localstop ) break;
 159                 if( ijpj[ifi][jfi] == localstop ) break;
 160                 k++;
 161                 iin = ifi; jin = jfi;
 162         }
 163         if( ifi == -1 ) *off1pt = 0; else *off1pt = ifi;
 164         if( jfi == -1 ) *off2pt = 0; else *off2pt = jfi;
 165
 166 //      fprintf( stderr, "ifn = %d, jfn = %d\n", ifi, jfi );
 167
 168
 169         return( 0.0 );
 170 }
 171
 172
 173 float genL__align11( char **seq1, char **seq2, int alloclen, int *off1pt, int *off2pt )
 174 /* score no keisan no sai motokaraaru gap no atukai ni mondai ga aru */
 175 {
 176 //      int k;
 177         register int i, j;
 178         int lasti, lastj;
 179         int lgth1, lgth2;
 180         int resultlen;
 181         float wm = 0.0;   /* int ?????? */
 182         float g;
 183         float *currentw, *previousw;
 184 #if 1
 185         float *wtmp;
 186         int *ijpipt;
 187         int *ijpjpt;
 188         float *mjpt, *Mjpt, *prept, *curpt;
 189         int *mpjpt, *Mpjpt;
 190 #endif
 191         static TLS float mi, *m;
 192         static TLS float Mi, *largeM;
 193         static TLS int **ijpi;
 194         static TLS int **ijpj;
 195         static TLS int mpi, *mp;
 196         static TLS int Mpi, *Mp;
 197         static TLS float *w1, *w2;
 198         static TLS float *match;
 199         static TLS float *initverticalw;    /* kufuu sureba iranai */
 200         static TLS float *lastverticalw;    /* kufuu sureba iranai */
 201         static TLS char **mseq1;
 202         static TLS char **mseq2;
 203         static TLS char **mseq;
 204         static TLS float **cpmx1;
 205         static TLS float **cpmx2;
 206         static TLS int **intwork;
 207         static TLS float **floatwork;
 208         static TLS int orlgth1 = 0, orlgth2 = 0;
 209         float maxwm;
 210         float tbk;
 211         int tbki, tbkj;
 212         int endali, endalj;
 213 //      float localthr = 0.0;
 214 //      float localthr2 = 0.0;
 215         float fpenalty = (float)penalty;
 216         float fpenalty_OP = (float)penalty_OP;
 217         float fpenalty_ex = (float)penalty_ex;
 218 //      float fpenalty_EX = (float)penalty_EX;
 219         float foffset = (float)offset;
 220         float localthr = -foffset;
 221         float localthr2 = -foffset;
 222
 223         if( seq1 == NULL )
 224         {
 225                 if( orlgth1 > 0 && orlgth2 > 0 )
 226                 {
 227                         orlgth1 = 0;
 228                         orlgth2 = 0;
 229                         free( mseq1 );
 230                         free( mseq2 );
 231                         FreeFloatVec( w1 );
 232                         FreeFloatVec( w2 );
 233                         FreeFloatVec( match );
 234                         FreeFloatVec( initverticalw );
 235                         FreeFloatVec( lastverticalw );
 236
 237                         FreeFloatVec( m );
 238                         FreeIntVec( mp );
 239                         free( largeM );
 240                         free( Mp );
 241
 242                         FreeCharMtx( mseq );
 243
 244                         FreeFloatMtx( cpmx1 );
 245                         FreeFloatMtx( cpmx2 );
 246
 247                         FreeFloatMtx( floatwork );
 248                         FreeIntMtx( intwork );
 249
 250                 }
 251                 return( 0.0 );
 252         }
 253
 254
 255
 256 //      fprintf( stderr, "@@@@@@@@@@@@@ penalty_OP = %f, penalty_EX = %f, pelanty = %f\n", fpenalty_OP, fpenalty_EX, fpenalty );
 257
 258         if( orlgth1 == 0 )
 259         {
 260                 mseq1 = AllocateCharMtx( njob, 0 );
 261                 mseq2 = AllocateCharMtx( njob, 0 );
 262         }
 263
 264
 265         lgth1 = strlen( seq1[0] );
 266         lgth2 = strlen( seq2[0] );
 267
 268         if( lgth1 > orlgth1 || lgth2 > orlgth2 )
 269         {
 270                 int ll1, ll2;
 271
 272                 if( orlgth1 > 0 && orlgth2 > 0 )
 273                 {
 274                         FreeFloatVec( w1 );
 275                         FreeFloatVec( w2 );
 276                         FreeFloatVec( match );
 277                         FreeFloatVec( initverticalw );
 278                         FreeFloatVec( lastverticalw );
 279
 280                         FreeFloatVec( m );
 281                         FreeIntVec( mp );
 282                         FreeFloatVec( largeM );
 283                         FreeIntVec( Mp );
 284
 285                         FreeCharMtx( mseq );
 286
 287                         FreeFloatMtx( cpmx1 );
 288                         FreeFloatMtx( cpmx2 );
 289
 290                         FreeFloatMtx( floatwork );
 291                         FreeIntMtx( intwork );
 292                 }
 293
 294                 ll1 = MAX( (int)(1.3*lgth1), orlgth1 ) + 100;
 295                 ll2 = MAX( (int)(1.3*lgth2), orlgth2 ) + 100;
 296
 297 #if DEBUG
 298                 fprintf( stderr, "\ntrying to allocate (%d+%d)xn matrices ... ", ll1, ll2 );
 299 #endif
 300
 301                 w1 = AllocateFloatVec( ll2+2 );
 302                 w2 = AllocateFloatVec( ll2+2 );
 303                 match = AllocateFloatVec( ll2+2 );
 304
 305                 initverticalw = AllocateFloatVec( ll1+2 );
 306                 lastverticalw = AllocateFloatVec( ll1+2 );
 307
 308                 m = AllocateFloatVec( ll2+2 );
 309                 mp = AllocateIntVec( ll2+2 );
 310                 largeM = AllocateFloatVec( ll2+2 );
 311                 Mp = AllocateIntVec( ll2+2 );
 312
 313                 mseq = AllocateCharMtx( njob, ll1+ll2 );
 314
 315                 cpmx1 = AllocateFloatMtx( 26, ll1+2 );
 316                 cpmx2 = AllocateFloatMtx( 26, ll2+2 );
 317
 318                 floatwork = AllocateFloatMtx( 26, MAX( ll1, ll2 )+2 );
 319                 intwork = AllocateIntMtx( 26, MAX( ll1, ll2 )+2 );
 320
 321 #if DEBUG
 322                 fprintf( stderr, "succeeded\n" );
 323 #endif
 324
 325                 orlgth1 = ll1 - 100;
 326                 orlgth2 = ll2 - 100;
 327         }
 328
 329
 330         mseq1[0] = mseq[0];
 331         mseq2[0] = mseq[1];
 332
 333
 334         if( orlgth1 > commonAlloc1 || orlgth2 > commonAlloc2 )
 335         {
 336                 int ll1, ll2;
 337
 338                 if( commonAlloc1 && commonAlloc2 )
 339                 {
 340                         FreeIntMtx( commonIP );
 341                         FreeIntMtx( commonJP );
 342                 }
 343
 344                 ll1 = MAX( orlgth1, commonAlloc1 );
 345                 ll2 = MAX( orlgth2, commonAlloc2 );
 346
 347 #if DEBUG
 348                 fprintf( stderr, "\n\ntrying to allocate %dx%d matrices ... ", ll1+1, ll2+1 );
 349 #endif
 350
 351                 commonIP = AllocateIntMtx( ll1+10, ll2+10 );
 352                 commonJP = AllocateIntMtx( ll1+10, ll2+10 );
 353
 354 #if DEBUG
 355                 fprintf( stderr, "succeeded\n\n" );
 356 #endif
 357
 358                 commonAlloc1 = ll1;
 359                 commonAlloc2 = ll2;
 360         }
 361         ijpi = commonIP;
 362         ijpj = commonJP;
 363
 364
 365 #if 0
 366         for( i=0; i<lgth1; i++ )
 367                 fprintf( stderr, "ogcp1[%d]=%f\n", i, ogcp1[i] );
 368 #endif
 369
 370         currentw = w1;
 371         previousw = w2;
 372
 373         match_calc( initverticalw, seq2, seq1, 0, lgth1 );
 374
 375         match_calc( currentw, seq1, seq2, 0, lgth2 );
 376
 377
 378         lasti = lgth2+1;
 379         for( j=1; j<lasti; ++j )
 380         {
 381                 m[j] = currentw[j-1]; mp[j] = 0;
 382                 largeM[j] = currentw[j-1]; Mp[j] = 0;
 383         }
 384
 385         lastverticalw[0] = currentw[lgth2-1];
 386
 387
 388 #if 0
 389 fprintf( stderr, "currentw = \n" );
 390 for( i=0; i<lgth1+1; i++ )
 391 {
 392         fprintf( stderr, "%5.2f ", currentw[i] );
 393 }
 394 fprintf( stderr, "\n" );
 395 fprintf( stderr, "initverticalw = \n" );
 396 for( i=0; i<lgth2+1; i++ )
 397 {
 398         fprintf( stderr, "%5.2f ", initverticalw[i] );
 399 }
 400 fprintf( stderr, "\n" );
 401 #endif
 402 #if DEBUG2
 403         fprintf( stderr, "\n" );
 404         fprintf( stderr, "       " );
 405         for( j=0; j<lgth2+1; j++ )
 406                 fprintf( stderr, "%c     ", seq2[0][j] );
 407         fprintf( stderr, "\n" );
 408 #endif
 409
 410         localstop = lgth1+lgth2+1;
 411         maxwm = -999999999.9;
 412         endali = endalj = 0;
 413 #if DEBUG2
 414         fprintf( stderr, "\n" );
 415         fprintf( stderr, "%c   ", seq1[0][0] );
 416
 417         for( j=0; j<lgth2+1; j++ )
 418                 fprintf( stderr, "%5.0f ", currentw[j] );
 419         fprintf( stderr, "\n" );
 420 #endif
 421
 422         lasti = lgth1+1;
 423         for( i=1; i<lasti; i++ )
 424         {
 425                 wtmp = previousw;
 426                 previousw = currentw;
 427                 currentw = wtmp;
 428
 429                 previousw[0] = initverticalw[i-1];
 430
 431                 match_calc( currentw, seq1, seq2, i, lgth2 );
 432 #if DEBUG2
 433                 fprintf( stderr, "%c   ", seq1[0][i] );
 434                 fprintf( stderr, "%5.0f ", currentw[0] );
 435 #endif
 436
 437 #if XXXXXXX
 438 fprintf( stderr, "\n" );
 439 fprintf( stderr, "i=%d\n", i );
 440 fprintf( stderr, "currentw = \n" );
 441 for( j=0; j<lgth2; j++ )
 442 {
 443         fprintf( stderr, "%5.2f ", currentw[j] );
 444 }
 445 fprintf( stderr, "\n" );
 446 #endif
 447 #if XXXXXXX
 448 fprintf( stderr, "\n" );
 449 fprintf( stderr, "i=%d\n", i );
 450 fprintf( stderr, "currentw = \n" );
 451 for( j=0; j<lgth2; j++ )
 452 {
 453         fprintf( stderr, "%5.2f ", currentw[j] );
 454 }
 455 fprintf( stderr, "\n" );
 456 #endif
 457                 currentw[0] = initverticalw[i];
 458
 459                 mi = previousw[0]; mpi = 0;
 460                 Mi = previousw[0]; Mpi = 0;
 461
 462 #if 0
 463                 if( mi < localthr ) mi = localthr2;
 464 #endif
 465
 466                 ijpipt = ijpi[i] + 1;
 467                 ijpjpt = ijpj[i] + 1;
 468                 mjpt = m + 1;
 469                 Mjpt = largeM + 1;
 470                 prept = previousw;
 471                 curpt = currentw + 1;
 472                 mpjpt = mp + 1;
 473                 Mpjpt = Mp + 1;
 474                 tbk = -999999.9;
 475                 tbki = 0;
 476                 tbkj = 0;
 477                 lastj = lgth2+1;
 478                 for( j=1; j<lastj; j++ )
 479                 {
 480                         wm = *prept;
 481                         *ijpipt = i-1;
 482                         *ijpjpt = j-1;
 483
 484
 485 //                      fprintf( stderr, "i,j=%d,%d %c-%c\n", i, j, seq1[0][i], seq2[0][j] );
 486 //                      fprintf( stderr, "wm=%f\n", wm );
 487 #if 0
 488                         fprintf( stderr, "%5.0f->", wm );
 489 #endif
 490                         g = mi + fpenalty;
 491 #if 0
 492                         fprintf( stderr, "%5.0f?", g );
 493 #endif
 494                         if( g > wm )
 495                         {
 496                                 wm = g;
 497 //                              *ijpipt = i - 1;
 498                                 *ijpjpt = mpi;
 499                         }
 500                         g = *prept;
 501                         if( g > mi )
 502                         {
 503                                 mi = g;
 504                                 mpi = j-1;
 505                         }
 506
 507 #if USE_PENALTY_EX
 508                         mi += fpenalty_ex;
 509 #endif
 510
 511 #if 0
 512                         fprintf( stderr, "%5.0f->", wm );
 513 #endif
 514                         g = *mjpt + fpenalty;
 515 #if 0
 516                         fprintf( stderr, "m%5.0f?", g );
 517 #endif
 518                         if( g > wm )
 519                         {
 520                                 wm = g;
 521                                 *ijpipt = *mpjpt;
 522                                 *ijpjpt = j - 1; //IRU!
 523                         }
 524                         g = *prept;
 525                         if( g > *mjpt )
 526                         {
 527                                 *mjpt = g;
 528                                 *mpjpt = i-1;
 529                         }
 530 #if USE_PENALTY_EX
 531                         *mjpt += fpenalty_ex;
 532 #endif
 533
 534
 535                         g =  tbk + fpenalty_OP;
 536 //                      g =  tbk;
 537                         if( g > wm )
 538                         {
 539                                 wm = g;
 540                                 *ijpipt = tbki;
 541                                 *ijpjpt = tbkj;
 542 //                              fprintf( stderr, "hit! i%d, j%d, ijpi = %d, ijpj = %d\n", i, j, *ijpipt, *ijpjpt );
 543                         }
 544 //                      g = Mi;
 545                         if( Mi > tbk )
 546                         {
 547                                 tbk = Mi; //error desu.
 548                                 tbki = i-1;
 549                                 tbkj = Mpi;
 550                         }
 551 //                      g = *Mjpt;
 552                         if( *Mjpt > tbk )
 553                         {
 554                                 tbk = *Mjpt;
 555                                 tbki = *Mpjpt;
 556                                 tbkj = j-1;
 557                         }
 558 //                      tbk += fpenalty_EX;// + foffset;
 559
 560 //                      g = *prept;
 561                         if( *prept > *Mjpt )
 562                         {
 563                                 *Mjpt = *prept;
 564                                 *Mpjpt = i-1;
 565                         }
 566 //                      *Mjpt += fpenalty_EX;// + foffset;
 567
 568 //                      g = *prept;
 569                         if( *prept > Mi )
 570                         {
 571                                 Mi = *prept;
 572                                 Mpi = j-1;
 573                         }
 574 //                      Mi += fpenalty_EX;// + foffset;
 575
 576
 577 //                      fprintf( stderr, "wm=%f, tbk=%f(%c-%c), mi=%f, *mjpt=%f\n", wm, tbk, seq1[0][tbki], seq2[0][tbkj], mi, *mjpt );
 578 //                      fprintf( stderr, "ijp = %c,%c\n", seq1[0][abs(*ijpipt)], seq2[0][abs(*ijpjpt)] );
 579
 580
 581                         if( maxwm < wm )
 582                         {
 583                                 maxwm = wm;
 584                                 endali = i;
 585                                 endalj = j;
 586                         }
 587 #if 1
 588                         if( wm < localthr )
 589                         {
 590 //                              fprintf( stderr, "stop i=%d, j=%d, curpt=%f\n", i, j, *curpt );
 591                                 *ijpipt = localstop;
 592 //                              *ijpjpt = localstop;
 593                                 wm = localthr2;
 594                         }
 595 #endif
 596 #if 0
 597                         fprintf( stderr, "%5.0f ", *curpt );
 598 #endif
 599 #if DEBUG2
 600                         fprintf( stderr, "%5.0f ", wm );
 601 //                      fprintf( stderr, "%c-%c *ijppt = %d, localstop = %d\n", seq1[0][i], seq2[0][j], *ijppt, localstop );
 602 #endif
 603
 604                         *curpt += wm;
 605                         ijpipt++;
 606                         ijpjpt++;
 607                         mjpt++;
 608                         Mjpt++;
 609                         prept++;
 610                         mpjpt++;
 611                         Mpjpt++;
 612                         curpt++;
 613                 }
 614 #if DEBUG2
 615                 fprintf( stderr, "\n" );
 616 #endif
 617
 618                 lastverticalw[i] = currentw[lgth2-1];
 619         }
 620
 621
 622 #if DEBUG2
 623         fprintf( stderr, "maxwm = %f\n", maxwm );
 624         fprintf( stderr, "endali = %d\n", endali );
 625         fprintf( stderr, "endalj = %d\n", endalj );
 626 #endif
 627
 628         if( ijpi[endali][endalj] == localstop ) // && ijpj[endali][endalj] == localstop )
 629         {
 630                 strcpy( seq1[0], "" );
 631                 strcpy( seq2[0], "" );
 632                 *off1pt = *off2pt = 0;
 633                 return( 0.0 );
 634         }
 635
 636
 637         gentracking( currentw, lastverticalw, seq1, seq2, mseq1, mseq2, cpmx1, cpmx2, ijpi, ijpj, off1pt, off2pt, endali, endalj );
 638
 639 //      fprintf( stderr, "### impmatch = %f\n", *impmatch );
 640
 641         resultlen = strlen( mseq1[0] );
 642         if( alloclen < resultlen || resultlen > N )
 643         {
 644                 fprintf( stderr, "alloclen=%d, resultlen=%d, N=%d\n", alloclen, resultlen, N );
 645                 ErrorExit( "LENGTH OVER!\n" );
 646         }
 647
 648
 649         strcpy( seq1[0], mseq1[0] );
 650         strcpy( seq2[0], mseq2[0] );
 651
 652 #if 0
 653         fprintf( stderr, "\n" );
 654         fprintf( stderr, ">\n%s\n", mseq1[0] );
 655         fprintf( stderr, ">\n%s\n", mseq2[0] );
 656 #endif
 657
 658
 659         return( maxwm );
 660 }
 661