binaries/src/mafft/core/Lalign11.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; // 060910
  10
  11 #if 1
  12 static void match_calc( float *match, char **s1, char **s2, int i1, int lgth2 )
  13 {
  14         char *seq2 = s2[0];
  15         int *intptr;
  16
  17         intptr = amino_dis[(int)s1[0][i1]];
  18         while( lgth2-- )
  19                 *match++ = intptr[(int)*seq2++];
  20 }
  21 #else
  22 static void match_calc( float *match, char **s1, char **s2, int i1, int lgth2 )
  23 {
  24         int j;
  25
  26         for( j=0; j<lgth2; j++ )
  27                 match[j] = amino_dis[(*s1)[i1]][(*s2)[j]];
  28 }
  29 #endif
  30
  31 #if 0
  32 static void match_calc_bk( float *match, float **cpmx1, float **cpmx2, int i1, int lgth2, float **floatwork, int **intwork, int initialize )
  33 {
  34         int j, k, l;
  35         float scarr[26];
  36         float **cpmxpd = floatwork;
  37         int **cpmxpdn = intwork;
  38         int count = 0;
  39
  40         if( initialize )
  41         {
  42                 for( j=0; j<lgth2; j++ )
  43                 {
  44                         count = 0;
  45                         for( l=0; l<26; l++ )
  46                         {
  47                                 if( cpmx2[l][j] )
  48                                 {
  49                                         cpmxpd[count][j] = cpmx2[l][j];
  50                                         cpmxpdn[count][j] = l;
  51                                         count++;
  52                                 }
  53                         }
  54                         cpmxpdn[count][j] = -1;
  55                 }
  56         }
  57
  58         for( l=0; l<26; l++ )
  59         {
  60                 scarr[l] = 0.0;
  61                 for( k=0; k<26; k++ )
  62                         scarr[l] += n_dis[k][l] * cpmx1[k][i1];
  63         }
  64 #if 0 /* ¤³¤ì¤ò»È¤¦¤È¤\ad¤Ïfloatwork¤Î¥¢¥í¥±¡¼¥È¤òµÕ¤Ë¤¹¤ë */
  65         {
  66                 float *fpt, **fptpt, *fpt2;
  67                 int *ipt, **iptpt;
  68                 fpt2 = match;
  69                 iptpt = cpmxpdn;
  70                 fptpt = cpmxpd;
  71                 while( lgth2-- )
  72                 {
  73                         *fpt2 = 0.0;
  74                         ipt=*iptpt,fpt=*fptpt;
  75                         while( *ipt > -1 )
  76                                 *fpt2 += scarr[*ipt++] * *fpt++;
  77                         fpt2++,iptpt++,fptpt++;
  78                 }
  79         }
  80 #else
  81         for( j=0; j<lgth2; j++ )
  82         {
  83                 match[j] = 0.0;
  84                 for( k=0; cpmxpdn[k][j]>-1; k++ )
  85                         match[j] += scarr[cpmxpdn[k][j]] * cpmxpd[k][j];
  86         }
  87 #endif
  88 }
  89 #endif
  90
  91 static float Ltracking( float *lasthorizontalw, float *lastverticalw,
  92                                                 char **seq1, char **seq2,
  93                         char **mseq1, char **mseq2,
  94                         int **ijp, 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, a guess
  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         ijp[i][0] = localstop;
 112     }
 113     for( j=0; j<lgth2+1; j++ )
 114     {
 115         ijp[0][j] = localstop;
 116     }
 117
 118         mseq1[0] += lgth1+lgth2;
 119         *mseq1[0] = 0;
 120         mseq2[0] += lgth1+lgth2;
 121         *mseq2[0] = 0;
 122         iin = endi; jin = endj;
 123         limk = lgth1+lgth2;
 124         for( k=0; k<=limk; k++ )
 125         {
 126                 if( ijp[iin][jin] < 0 )
 127                 {
 128                         ifi = iin-1; jfi = jin+ijp[iin][jin];
 129                 }
 130                 else if( ijp[iin][jin] > 0 )
 131                 {
 132                         ifi = iin-ijp[iin][jin]; jfi = jin-1;
 133                 }
 134                 else
 135                 {
 136                         ifi = iin-1; jfi = jin-1;
 137                 }
 138                 l = iin - ifi;
 139                 while( --l )
 140                 {
 141                         *--mseq1[0] = seq1[0][ifi+l];
 142                         *--mseq2[0] = *gap;
 143                         k++;
 144                 }
 145                 l= jin - jfi;
 146                 while( --l )
 147                 {
 148                         *--mseq1[0] = *gap;
 149                         *--mseq2[0] = seq2[0][jfi+l];
 150                         k++;
 151                 }
 152
 153                 if( iin <= 0 || jin <= 0 ) break;
 154                 *--mseq1[0] = seq1[0][ifi];
 155                 *--mseq2[0] = seq2[0][jfi];
 156                 if( ijp[ifi][jfi] == localstop ) break;
 157                 k++;
 158                 iin = ifi; jin = jfi;
 159         }
 160         if( ifi == -1 ) *off1pt = 0; else *off1pt = ifi;
 161         if( jfi == -1 ) *off2pt = 0; else *off2pt = jfi;
 162
 163 //      fprintf( stderr, "ifn = %d, jfn = %d\n", ifi, jfi );
 164
 165
 166         return( 0.0 );
 167 }
 168
 169
 170 float L__align11( char **seq1, char **seq2, int alloclen, int *off1pt, int *off2pt )
 171 /* score no keisan no sai motokaraaru gap no atukai ni mondai ga aru */
 172 {
 173 //      int k;
 174         register int i, j;
 175         int lasti, lastj;                      /* outgap == 0 -> lgth1, outgap == 1 -> lgth1+1 */
 176         int lgth1, lgth2;
 177         int resultlen;
 178         float wm = 0.0;   /* int ?????? */
 179         float g;
 180         float *currentw, *previousw;
 181 #if 1
 182         float *wtmp;
 183         int *ijppt;
 184         float *mjpt, *prept, *curpt;
 185         int *mpjpt;
 186 #endif
 187         static float mi, *m;
 188         static int **ijp;
 189         static int mpi, *mp;
 190         static float *w1, *w2;
 191         static float *match;
 192         static float *initverticalw;    /* kufuu sureba iranai */
 193         static float *lastverticalw;    /* kufuu sureba iranai */
 194         static char **mseq1;
 195         static char **mseq2;
 196         static char **mseq;
 197 //      static int **intwork;
 198 //      static float **floatwork;
 199         static int orlgth1 = 0, orlgth2 = 0;
 200         float maxwm;
 201         int endali = 0, endalj = 0; // by D.Mathog, a guess
 202 //      int endali, endalj;
 203         float localthr = -offset;
 204         float localthr2 = -offset;
 205 //      float localthr = 100;
 206 //      float localthr2 = 100;
 207         float fpenalty = (float)penalty;
 208         float fpenalty_ex = (float)penalty_ex;
 209
 210
 211         if( orlgth1 == 0 )
 212         {
 213                 mseq1 = AllocateCharMtx( njob, 0 );
 214                 mseq2 = AllocateCharMtx( njob, 0 );
 215         }
 216
 217
 218         lgth1 = strlen( seq1[0] );
 219         lgth2 = strlen( seq2[0] );
 220
 221         if( lgth1 > orlgth1 || lgth2 > orlgth2 )
 222         {
 223                 int ll1, ll2;
 224
 225                 if( orlgth1 > 0 && orlgth2 > 0 )
 226                 {
 227                         FreeFloatVec( w1 );
 228                         FreeFloatVec( w2 );
 229                         FreeFloatVec( match );
 230                         FreeFloatVec( initverticalw );
 231                         FreeFloatVec( lastverticalw );
 232
 233                         FreeFloatVec( m );
 234                         FreeIntVec( mp );
 235
 236                         FreeCharMtx( mseq );
 237
 238
 239
 240 //                      FreeFloatMtx( floatwork );
 241 //                      FreeIntMtx( intwork );
 242                 }
 243
 244                 ll1 = MAX( (int)(1.3*lgth1), orlgth1 ) + 100;
 245                 ll2 = MAX( (int)(1.3*lgth2), orlgth2 ) + 100;
 246
 247 #if DEBUG
 248                 fprintf( stderr, "\ntrying to allocate (%d+%d)xn matrices ... ", ll1, ll2 );
 249 #endif
 250
 251                 w1 = AllocateFloatVec( ll2+2 );
 252                 w2 = AllocateFloatVec( ll2+2 );
 253                 match = AllocateFloatVec( ll2+2 );
 254
 255                 initverticalw = AllocateFloatVec( ll1+2 );
 256                 lastverticalw = AllocateFloatVec( ll1+2 );
 257
 258                 m = AllocateFloatVec( ll2+2 );
 259                 mp = AllocateIntVec( ll2+2 );
 260
 261                 mseq = AllocateCharMtx( njob, ll1+ll2 );
 262
 263
 264 //              floatwork = AllocateFloatMtx( 26, MAX( ll1, ll2 )+2 );
 265 //              intwork = AllocateIntMtx( 26, MAX( ll1, ll2 )+2 );
 266
 267 #if DEBUG
 268                 fprintf( stderr, "succeeded\n" );
 269 #endif
 270
 271                 orlgth1 = ll1 - 100;
 272                 orlgth2 = ll2 - 100;
 273         }
 274
 275
 276         mseq1[0] = mseq[0];
 277         mseq2[0] = mseq[1];
 278
 279
 280         if( orlgth1 > commonAlloc1 || orlgth2 > commonAlloc2 )
 281         {
 282                 int ll1, ll2;
 283
 284                 if( commonAlloc1 && commonAlloc2 )
 285                 {
 286                         FreeIntMtx( commonIP );
 287                 }
 288
 289                 ll1 = MAX( orlgth1, commonAlloc1 );
 290                 ll2 = MAX( orlgth2, commonAlloc2 );
 291
 292 #if DEBUG
 293                 fprintf( stderr, "\n\ntrying to allocate %dx%d matrices ... ", ll1+1, ll2+1 );
 294 #endif
 295
 296                 commonIP = AllocateIntMtx( ll1+10, ll2+10 );
 297
 298 #if DEBUG
 299                 fprintf( stderr, "succeeded\n\n" );
 300 #endif
 301
 302                 commonAlloc1 = ll1;
 303                 commonAlloc2 = ll2;
 304         }
 305         ijp = commonIP;
 306
 307
 308 #if 0
 309         for( i=0; i<lgth1; i++ )
 310                 fprintf( stderr, "ogcp1[%d]=%f\n", i, ogcp1[i] );
 311 #endif
 312
 313         currentw = w1;
 314         previousw = w2;
 315
 316         match_calc( initverticalw, seq2, seq1, 0, lgth1 );
 317
 318         match_calc( currentw, seq1, seq2, 0, lgth2 );
 319
 320
 321         lasti = lgth2+1;
 322         for( j=1; j<lasti; ++j )
 323         {
 324                 m[j] = currentw[j-1]; mp[j] = 0;
 325 #if 0
 326                 if( m[j] < localthr ) m[j] = localthr2;
 327 #endif
 328         }
 329
 330         lastverticalw[0] = currentw[lgth2-1];
 331
 332         lasti = lgth1+1;
 333
 334 #if 0
 335 fprintf( stderr, "currentw = \n" );
 336 for( i=0; i<lgth1+1; i++ )
 337 {
 338         fprintf( stderr, "%5.2f ", currentw[i] );
 339 }
 340 fprintf( stderr, "\n" );
 341 fprintf( stderr, "initverticalw = \n" );
 342 for( i=0; i<lgth2+1; i++ )
 343 {
 344         fprintf( stderr, "%5.2f ", initverticalw[i] );
 345 }
 346 fprintf( stderr, "\n" );
 347 #endif
 348 #if DEBUG2
 349         fprintf( stderr, "\n" );
 350         fprintf( stderr, "       " );
 351         for( j=0; j<lgth2; j++ )
 352                 fprintf( stderr, "%c     ", seq2[0][j] );
 353         fprintf( stderr, "\n" );
 354 #endif
 355
 356         localstop = lgth1+lgth2+1;
 357         maxwm = -999999999.9;
 358 #if DEBUG2
 359         fprintf( stderr, "\n" );
 360         fprintf( stderr, "%c   ", seq1[0][0] );
 361
 362         for( j=0; j<lgth2+1; j++ )
 363                 fprintf( stderr, "%5.0f ", currentw[j] );
 364         fprintf( stderr, "\n" );
 365 #endif
 366
 367         for( i=1; i<lasti; i++ )
 368         {
 369                 wtmp = previousw;
 370                 previousw = currentw;
 371                 currentw = wtmp;
 372
 373                 previousw[0] = initverticalw[i-1];
 374
 375                 match_calc( currentw, seq1, seq2, i, lgth2 );
 376 #if DEBUG2
 377                 fprintf( stderr, "%c   ", seq1[0][i] );
 378                 fprintf( stderr, "%5.0f ", currentw[0] );
 379 #endif
 380
 381 #if XXXXXXX
 382 fprintf( stderr, "\n" );
 383 fprintf( stderr, "i=%d\n", i );
 384 fprintf( stderr, "currentw = \n" );
 385 for( j=0; j<lgth2; j++ )
 386 {
 387         fprintf( stderr, "%5.2f ", currentw[j] );
 388 }
 389 fprintf( stderr, "\n" );
 390 #endif
 391 #if XXXXXXX
 392 fprintf( stderr, "\n" );
 393 fprintf( stderr, "i=%d\n", i );
 394 fprintf( stderr, "currentw = \n" );
 395 for( j=0; j<lgth2; j++ )
 396 {
 397         fprintf( stderr, "%5.2f ", currentw[j] );
 398 }
 399 fprintf( stderr, "\n" );
 400 #endif
 401                 currentw[0] = initverticalw[i];
 402
 403                 mi = previousw[0]; mpi = 0;
 404
 405 #if 0
 406                 if( mi < localthr ) mi = localthr2;
 407 #endif
 408
 409                 ijppt = ijp[i] + 1;
 410                 mjpt = m + 1;
 411                 prept = previousw;
 412                 curpt = currentw + 1;
 413                 mpjpt = mp + 1;
 414                 lastj = lgth2+1;
 415                 for( j=1; j<lastj; j++ )
 416                 {
 417                         wm = *prept;
 418                         *ijppt = 0;
 419
 420 #if 0
 421                         fprintf( stderr, "%5.0f->", wm );
 422 #endif
 423 #if 0
 424                         fprintf( stderr, "%5.0f?", g );
 425 #endif
 426                         if( (g=mi+fpenalty) > wm )
 427                         {
 428                                 wm = g;
 429                                 *ijppt = -( j - mpi );
 430                         }
 431                         if( *prept > mi )
 432                         {
 433                                 mi = *prept;
 434                                 mpi = j-1;
 435                         }
 436
 437 #if USE_PENALTY_EX
 438                         mi += fpenalty_ex;
 439 #endif
 440
 441 #if 0
 442                         fprintf( stderr, "%5.0f?", g );
 443 #endif
 444                         if( (g=*mjpt+fpenalty) > wm )
 445                         {
 446                                 wm = g;
 447                                 *ijppt = +( i - *mpjpt );
 448                         }
 449                         if( *prept > *mjpt )
 450                         {
 451                                 *mjpt = *prept;
 452                                 *mpjpt = i-1;
 453                         }
 454 #if USE_PENALTY_EX
 455                         *mjpt += fpenalty_ex;
 456 #endif
 457
 458                         if( maxwm < wm )
 459                         {
 460                                 maxwm = wm;
 461                                 endali = i;
 462                                 endalj = j;
 463                         }
 464 #if 1
 465                         if( wm < localthr )
 466                         {
 467 //                              fprintf( stderr, "stop i=%d, j=%d, curpt=%f\n", i, j, *curpt );
 468                                 *ijppt = localstop;
 469                                 wm = localthr2;
 470                         }
 471 #endif
 472 #if 0
 473                         fprintf( stderr, "%5.0f ", *curpt );
 474 #endif
 475 #if DEBUG2
 476                         fprintf( stderr, "%5.0f ", wm );
 477 //                      fprintf( stderr, "%c-%c *ijppt = %d, localstop = %d\n", seq1[0][i], seq2[0][j], *ijppt, localstop );
 478 #endif
 479
 480                         *curpt++ += wm;
 481                         ijppt++;
 482                         mjpt++;
 483                         prept++;
 484                         mpjpt++;
 485                 }
 486 #if DEBUG2
 487                 fprintf( stderr, "\n" );
 488 #endif
 489
 490                 lastverticalw[i] = currentw[lgth2-1];
 491         }
 492
 493
 494 #if 0
 495         fprintf( stderr, "maxwm = %f\n", maxwm );
 496         fprintf( stderr, "endali = %d\n", endali );
 497         fprintf( stderr, "endalj = %d\n", endalj );
 498 #endif
 499
 500         if( ijp[endali][endalj] == localstop )
 501         {
 502                 strcpy( seq1[0], "" );
 503                 strcpy( seq2[0], "" );
 504                 *off1pt = *off2pt = 0;
 505                 return( 0.0 );
 506         }
 507
 508         Ltracking( currentw, lastverticalw, seq1, seq2, mseq1, mseq2, ijp, off1pt, off2pt, endali, endalj );
 509
 510
 511         resultlen = strlen( mseq1[0] );
 512         if( alloclen < resultlen || resultlen > N )
 513         {
 514                 fprintf( stderr, "alloclen=%d, resultlen=%d, N=%d\n", alloclen, resultlen, N );
 515                 ErrorExit( "LENGTH OVER!\n" );
 516         }
 517
 518
 519         strcpy( seq1[0], mseq1[0] );
 520         strcpy( seq2[0], mseq2[0] );
 521
 522 #if 0
 523         fprintf( stderr, "wm=%f\n", wm );
 524         fprintf( stderr, ">\n%s\n", mseq1[0] );
 525         fprintf( stderr, ">\n%s\n", mseq2[0] );
 526
 527         fprintf( stderr, "maxwm = %f\n", maxwm );
 528         fprintf( stderr, "   wm = %f\n",    wm );
 529 #endif
 530
 531         return( maxwm );
 532 }
 533