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