binaries/src/mafft/core/rnatest.c

   1 #include "mltaln.h"
   2
   3 #define DEBUG 0
   4 #define IODEBUG 0
   5 #define SCOREOUT 0
   6
   7
   8
   9 void arguments( int argc, char *argv[] )
  10 {
  11     int c;
  12
  13         inputfile = NULL;
  14         fftkeika = 0;
  15         pslocal = -1000.0;
  16         constraint = 0;
  17         nblosum = 62;
  18         fmodel = 0;
  19         calledByXced = 0;
  20         devide = 0;
  21         use_fft = 0;
  22         fftscore = 1;
  23         fftRepeatStop = 0;
  24         fftNoAnchStop = 0;
  25     weight = 3;
  26     utree = 1;
  27         tbutree = 1;
  28     refine = 0;
  29     check = 1;
  30     cut = 0.0;
  31     disp = 0;
  32     outgap = 1;
  33     alg = 'A';
  34     mix = 0;
  35         tbitr = 0;
  36         scmtd = 5;
  37         tbweight = 0;
  38         tbrweight = 3;
  39         checkC = 0;
  40         treemethod = 'x';
  41         contin = 0;
  42         scoremtx = 1;
  43         kobetsubunkatsu = 0;
  44         divpairscore = 0;
  45         dorp = NOTSPECIFIED;
  46         ppenalty = NOTSPECIFIED;
  47         ppenalty_OP = NOTSPECIFIED;
  48         ppenalty_ex = NOTSPECIFIED;
  49         ppenalty_EX = NOTSPECIFIED;
  50         poffset = NOTSPECIFIED;
  51         kimuraR = NOTSPECIFIED;
  52         pamN = NOTSPECIFIED;
  53         geta2 = GETA2;
  54         fftWinSize = NOTSPECIFIED;
  55         fftThreshold = NOTSPECIFIED;
  56         RNAppenalty = NOTSPECIFIED;
  57         RNApthr = NOTSPECIFIED;
  58
  59     while( --argc > 0 && (*++argv)[0] == '-' )
  60         {
  61         while ( ( c = *++argv[0] ) )
  62                 {
  63             switch( c )
  64             {
  65                                 case 'i':
  66                                         inputfile = *++argv;
  67                                         fprintf( stderr, "inputfile = %s\n", inputfile );
  68                                         --argc;
  69                                         goto nextoption;
  70                                 case 'o':
  71                                         RNAppenalty = (int)( atof( *++argv ) * 1000 - 0.5 );
  72                                         --argc;
  73                                         goto nextoption;
  74                                 case 'f':
  75                                         ppenalty = (int)( atof( *++argv ) * 1000 - 0.5 );
  76                                         --argc;
  77                                         goto nextoption;
  78                                 case 'g':
  79                                         ppenalty_ex = (int)( atof( *++argv ) * 1000 - 0.5 );
  80                                         --argc;
  81                                         goto nextoption;
  82                                 case 'O':
  83                                         ppenalty_OP = (int)( atof( *++argv ) * 1000 - 0.5 );
  84                                         --argc;
  85                                         goto nextoption;
  86                                 case 'E':
  87                                         ppenalty_EX = (int)( atof( *++argv ) * 1000 - 0.5 );
  88                                         --argc;
  89                                         goto nextoption;
  90                                 case 'h':
  91                                         poffset = (int)( atof( *++argv ) * 1000 - 0.5 );
  92                                         --argc;
  93                                         goto nextoption;
  94                                 case 'k':
  95                                         kimuraR = atoi( *++argv );
  96 //                                      fprintf( stderr, "kimuraR = %d\n", kimuraR );
  97                                         --argc;
  98                                         goto nextoption;
  99                                 case 'b':
 100                                         nblosum = atoi( *++argv );
 101                                         scoremtx = 1;
 102                                         fprintf( stderr, "blosum %d\n", nblosum );
 103                                         --argc;
 104                                         goto nextoption;
 105                                 case 'j':
 106                                         pamN = atoi( *++argv );
 107                                         scoremtx = 0;
 108                                         TMorJTT = JTT;
 109                                         fprintf( stderr, "jtt %d\n", pamN );
 110                                         --argc;
 111                                         goto nextoption;
 112                                 case 'm':
 113                                         pamN = atoi( *++argv );
 114                                         scoremtx = 0;
 115                                         TMorJTT = TM;
 116                                         fprintf( stderr, "TM %d\n", pamN );
 117                                         --argc;
 118                                         goto nextoption;
 119                                 case 'l':
 120                                         ppslocal = (int)( atof( *++argv ) * 1000 + 0.5 );
 121                                         pslocal = (int)( 600.0 / 1000.0 * ppslocal + 0.5);
 122 //                                      fprintf( stderr, "ppslocal = %d\n", ppslocal );
 123 //                                      fprintf( stderr, "pslocal = %d\n", pslocal );
 124                                         --argc;
 125                                         goto nextoption;
 126 #if 1
 127                                 case 'a':
 128                                         fmodel = 1;
 129                                         break;
 130 #endif
 131                                 case 'r':
 132                                         fmodel = -1;
 133                                         break;
 134                                 case 'D':
 135                                         dorp = 'd';
 136                                         break;
 137                                 case 'P':
 138                                         dorp = 'p';
 139                                         break;
 140 #if 0
 141                                 case 'e':
 142                                         fftscore = 0;
 143                                         break;
 144                                 case 'O':
 145                                         fftNoAnchStop = 1;
 146                                         break;
 147                                 case 'R':
 148                                         fftRepeatStop = 1;
 149                                         break;
 150 #endif
 151                                 case 'Q':
 152                                         calledByXced = 1;
 153                                         break;
 154                                 case 's':
 155                                         treemethod = 's';
 156                                         break;
 157                                 case 'x':
 158                                         disp = 1;
 159                                         break;
 160                                 case 'p':
 161                                         treemethod = 'p';
 162                                         break;
 163 #if 0
 164                                 case 'a':
 165                                         alg = 'a';
 166                                         break;
 167 #endif
 168                                 case 'S':
 169                                         alg = 'S';
 170                                         break;
 171                                 case 'L':
 172                                         alg = 'L';
 173                                         break;
 174                                 case 'M':
 175                                         alg = 'M';
 176                                         break;
 177                                 case 'R':
 178                                         alg = 'R';
 179                                         break;
 180                                 case 'N':
 181                                         alg = 'N';
 182                                         break;
 183                                 case 'A':
 184                                         alg = 'A';
 185                                         break;
 186                                 case 'V':
 187                                         alg = 'V';
 188                                         break;
 189                                 case 'C':
 190                                         alg = 'C';
 191                                         break;
 192                                 case 'F':
 193                                         use_fft = 1;
 194                                         break;
 195                                 case 'v':
 196                                         tbrweight = 3;
 197                                         break;
 198                                 case 'd':
 199                                         divpairscore = 1;
 200                                         break;
 201 /* Modified 01/08/27, default: user tree */
 202                                 case 'J':
 203                                         tbutree = 0;
 204                                         break;
 205 /* modification end. */
 206                                 case 'z':
 207                                         fftThreshold = atoi( *++argv );
 208                                         --argc;
 209                                         goto nextoption;
 210                                 case 'w':
 211                                         fftWinSize = atoi( *++argv );
 212                                         --argc;
 213                                         goto nextoption;
 214                                 case 'Z':
 215                                         checkC = 1;
 216                                         break;
 217                 default:
 218                     fprintf( stderr, "illegal option %c\n", c );
 219                     argc = 0;
 220                     break;
 221             }
 222                 }
 223                 nextoption:
 224                         ;
 225         }
 226     if( argc == 1 )
 227     {
 228         cut = atof( (*argv) );
 229         argc--;
 230     }
 231     if( argc != 0 )
 232     {
 233         fprintf( stderr, "options: Check source file !\n" );
 234         exit( 1 );
 235     }
 236         if( tbitr == 1 && outgap == 0 )
 237         {
 238                 fprintf( stderr, "conflicting options : o, m or u\n" );
 239                 exit( 1 );
 240         }
 241         if( alg == 'C' && outgap == 0 )
 242         {
 243                 fprintf( stderr, "conflicting options : C, o\n" );
 244                 exit( 1 );
 245         }
 246 }
 247
 248 int countamino( char *s, int end )
 249 {
 250         int val = 0;
 251         while( end-- )
 252                 if( *s++ != '-' ) val++;
 253         return( val );
 254 }
 255
 256
 257 static void WriteOptions( FILE *fp )
 258 {
 259
 260         if( dorp == 'd' ) fprintf( fp, "DNA\n" );
 261         else
 262         {
 263                 if     ( scoremtx ==  0 ) fprintf( fp, "JTT %dPAM\n", pamN );
 264                 else if( scoremtx ==  1 ) fprintf( fp, "BLOSUM %d\n", nblosum );
 265                 else if( scoremtx ==  2 ) fprintf( fp, "M-Y\n" );
 266         }
 267     fprintf( stderr, "Gap Penalty = %+5.2f, %+5.2f, %+5.2f\n", (double)ppenalty/1000, (double)ppenalty_ex/1000, (double)poffset/1000 );
 268     if( use_fft ) fprintf( fp, "FFT on\n" );
 269
 270         fprintf( fp, "tree-base method\n" );
 271         if( tbrweight == 0 ) fprintf( fp, "unweighted\n" );
 272         else if( tbrweight == 3 ) fprintf( fp, "clustalw-like weighting\n" );
 273         if( tbitr || tbweight )
 274         {
 275                 fprintf( fp, "iterate at each step\n" );
 276                 if( tbitr && tbrweight == 0 ) fprintf( fp, "  unweighted\n" );
 277                 if( tbitr && tbrweight == 3 ) fprintf( fp, "  reversely weighted\n" );
 278                 if( tbweight ) fprintf( fp, "  weighted\n" );
 279                 fprintf( fp, "\n" );
 280         }
 281
 282          fprintf( fp, "Gap Penalty = %+5.2f, %+5.2f, %+5.2f\n", (double)ppenalty/1000, (double)ppenalty_ex/1000, (double)poffset/1000 );
 283
 284         if( alg == 'a' )
 285                 fprintf( fp, "Algorithm A\n" );
 286         else if( alg == 'A' )
 287                 fprintf( fp, "Algorithm A+\n" );
 288         else if( alg == 'S' )
 289                 fprintf( fp, "Apgorithm S\n" );
 290         else if( alg == 'C' )
 291                 fprintf( fp, "Apgorithm A+/C\n" );
 292         else
 293                 fprintf( fp, "Unknown algorithm\n" );
 294
 295         if( treemethod == 'x' )
 296                 fprintf( fp, "Tree = UPGMA (3).\n" );
 297         else if( treemethod == 's' )
 298                 fprintf( fp, "Tree = UPGMA (2).\n" );
 299         else if( treemethod == 'p' )
 300                 fprintf( fp, "Tree = UPGMA (1).\n" );
 301         else
 302                 fprintf( fp, "Unknown tree.\n" );
 303
 304     if( use_fft )
 305     {
 306         fprintf( fp, "FFT on\n" );
 307         if( dorp == 'd' )
 308             fprintf( fp, "Basis : 4 nucleotides\n" );
 309         else
 310         {
 311             if( fftscore )
 312                 fprintf( fp, "Basis : Polarity and Volume\n" );
 313             else
 314                 fprintf( fp, "Basis : 20 amino acids\n" );
 315         }
 316         fprintf( fp, "Threshold   of anchors = %d%%\n", fftThreshold );
 317         fprintf( fp, "window size of anchors = %dsites\n", fftWinSize );
 318     }
 319         else
 320         fprintf( fp, "FFT off\n" );
 321         fflush( fp );
 322 }
 323
 324
 325 int main( int argc, char *argv[] )
 326 {
 327         static int  *nlen;
 328         static char **name, **seq, **useq;
 329         static char **mseq1, **mseq2;
 330         static char **aseq;
 331         static char **bseq;
 332         static double *eff;
 333         int i;
 334         FILE *infp;
 335         char c;
 336         int alloclen;
 337         RNApair **pair1;
 338         RNApair **pair2;
 339         float **map;
 340
 341         arguments( argc, argv );
 342
 343         if( inputfile )
 344         {
 345                 infp = fopen( inputfile, "r" );
 346                 if( !infp )
 347                 {
 348                         fprintf( stderr, "Cannot open %s\n", inputfile );
 349                         exit( 1 );
 350                 }
 351         }
 352         else
 353                 infp = stdin;
 354
 355         getnumlen( infp );
 356         rewind( infp );
 357
 358         if( njob > M )
 359         {
 360                 fprintf( stderr, "The number of sequences must be < %d\n", M );
 361                 fprintf( stderr, "Please try the splittbfast program for such large data.\n" );
 362                 exit( 1 );
 363         }
 364
 365     name = AllocateCharMtx( njob, B+1 );
 366     nlen = AllocateIntVec( njob );
 367
 368         seq = AllocateCharMtx( njob, nlenmax*5+1 );
 369         useq = AllocateCharMtx( njob, nlenmax*5+1 );
 370         aseq = AllocateCharMtx( njob, nlenmax*5+1 );
 371         bseq = AllocateCharMtx( njob, nlenmax*5+1 );
 372         mseq1 = AllocateCharMtx( njob, 0 );
 373         mseq2 = AllocateCharMtx( njob, 0 );
 374         alloclen = nlenmax*5;
 375
 376         pair1 = calloc( nlenmax*5+1, sizeof( RNApair *) );
 377         pair2 = calloc( nlenmax*5+1, sizeof( RNApair *) );
 378         map = AllocateFloatMtx( nlenmax+1, nlenmax );
 379
 380         eff = AllocateDoubleVec( njob );
 381
 382         readData_pointer( infp, name, nlen, seq );
 383         fclose( infp );
 384
 385         for( i=0; i<njob; i++ ) strcpy( useq[i], seq[i] );
 386
 387         constants( njob, seq );
 388
 389 #if 0
 390         fprintf( stderr, "params = %d, %d, %d\n", penalty, penalty_ex, offset );
 391 #endif
 392
 393         initSignalSM();
 394
 395         initFiles();
 396
 397         WriteOptions( trap_g );
 398
 399         c = seqcheck( seq );
 400         if( c )
 401         {
 402                 fprintf( stderr, "Illeagal character %c\n", c );
 403                 exit( 1 );
 404         }
 405
 406 //      writePre( njob, name, nlen, seq, 0 );
 407
 408         for( i=0; i<njob; i++ ) eff[i] = 1.0;
 409
 410 //      for( i=0; i<njob; i++ ) gappick0( bseq[i], seq[i] );
 411
 412
 413         fprintf( stderr, "folding group1\n" );
 414 //      rnalocal( seq, useq, eff, eff, njob, njob, alloclen, pair1 );
 415         rnaalifoldcall( seq, njob, pair1 );
 416         exit( 1 );
 417
 418 #if 0
 419         fprintf( stderr, "folding group1\n" );
 420         rnalocal( seq+1, useq+1, eff+1, eff+1, 1, 1, alloclen, pair2 );
 421         fprintf( stderr, "aligning 1 and 2, phase 1\n" );
 422         Lalignmm_hmout( seq, seq+1, eff, eff+1, 1, 1, alloclen, NULL, NULL, NULL, NULL, map );
 423
 424
 425 #if 0
 426         lgth1 = strlen( seq[0] );
 427         for( i=0; i<lgth1; i++ )
 428         {
 429                 fprintf( stderr, "\n" );
 430                 if( pair1[i].pos == -1 ) continue;
 431                 fprintf( stderr, "i=%d (%c):%d", i, seq[0][i], pair1[i].pos );
 432                 if( map12[pair1[i].pos].pos == -1 ) continue;
 433                 fprintf( stderr, "%c -> %c ", seq[0][pair1[i].pos], seq[1][map12[pair1[i].pos].pos] );
 434                 if( pair2[map12[pair1[i].pos].pos].pos == -1 ) continue;
 435                 fprintf( stderr, "%d:%d (%c)", map12[pair1[i].pos].pos, pair2[map12[pair1[i].pos].pos].pos, seq[1][pair2[map12[pair1[i].pos].pos].pos] );
 436         }
 437 #endif
 438
 439
 440         exit( 1 );
 441
 442
 443         pairalign( name, nlen, bseq, aseq, mseq1, mseq2, eff, alloclen );
 444         fprintf( trap_g, "done.\n" );
 445 #if DEBUG
 446         fprintf( stderr, "closing trap_g\n" );
 447 #endif
 448         fclose( trap_g );
 449
 450 //      writePre( njob, name, nlen, aseq, !contin );
 451 #if 0
 452         writeData( stdout, njob, name, nlen, aseq );
 453 #endif
 454 #if IODEBUG
 455         fprintf( stderr, "OSHIMAI\n" );
 456 #endif
 457         SHOWVERSION;
 458         return( 0 );
 459 #endif
 460 }