binaries/src/mafft/core/f2cl.c

   1 #include "mltaln.h"
   2
   3 #define DEBUG 0
   4
   5
   6 static char *comment;
   7 static char *orderfile;
   8 static int format;
   9 static int namelen;
  10
  11 static void fillspace( char *seq, int lenmax )
  12 {
  13         int len = strlen( seq );
  14         seq += len;
  15         lenmax -= len;
  16         while( lenmax-- ) *seq++ = ' ';
  17         *seq = 0;
  18 }
  19
  20 void setmark_clustal( int nlen, int nseq, char **seq, char *mark )
  21 {
  22         int i, j, k;
  23
  24         char *strong[] = {
  25                                         "STA",
  26                                         "NEQK",
  27                                         "NHQK",
  28                                         "NDEQ",
  29                                         "QHRK",
  30                                         "MILV",
  31                                         "MILF",
  32                                         "HY",
  33                                         "FYW",
  34                                   };
  35         int nstrong = 9;
  36         char *weaker[] = {
  37                                         "CSA",
  38                                         "ATV",
  39                                         "SAG",
  40                                         "STNK",
  41                                         "STPA",
  42                                         "SGND",
  43                                         "SNDEQK",
  44                                         "NDEQHK",
  45                                         "NEQHRK",
  46                                         "FVLIM",
  47                                         "HFY",
  48                                   };
  49         int nweaker = 11;
  50
  51         for( i=0; i<nlen; i++ )
  52         {
  53                 mark[i] = ' ';
  54                 for( j=0; j<nseq; j++ )
  55                         if( '-' == seq[j][i] ) break;
  56                 if( j != nseq )
  57                 {
  58                         continue;
  59                 }
  60                 for( j=0; j<nseq; j++ )
  61                         if( toupper( seq[0][i] ) != toupper( seq[j][i] ) ) break;
  62                 if( j == nseq )
  63                 {
  64                         mark[i] = '*';
  65                         continue;
  66                 }
  67                 for( k=0; k<nstrong; k++ )
  68                 {
  69                         for( j=0; j<nseq; j++ )
  70                         {
  71                                 if( !strchr( strong[k], toupper( seq[j][i] ) ) ) break;
  72                         }
  73                         if( j == nseq ) break;
  74                 }
  75                 if( k < nstrong )
  76                 {
  77                         mark[i] = ':';
  78                         continue;
  79                 }
  80                 for( k=0; k<nweaker; k++ )
  81                 {
  82                         for( j=0; j<nseq; j++ )
  83                         {
  84                                 if( !strchr( weaker[k], toupper( seq[j][i] ) ) ) break;
  85                         }
  86                         if( j == nseq ) break;
  87                 }
  88                 if( k < nweaker )
  89                 {
  90                         mark[i] = '.';
  91                         continue;
  92                 }
  93         }
  94         mark[nlen] = 0;
  95 }
  96
  97 void setmark( int nlen, int nseq, char **seq, char *mark )
  98 {
  99         int i, j;
 100
 101         for( i=0; i<nlen; i++ )
 102         {
 103                 mark[i] = ' ';
 104                 for( j=0; j<nseq; j++ )
 105                         if( '-' == seq[j][i] ) break;
 106                 if( j != nseq )
 107                 {
 108                         continue;
 109                 }
 110                 for( j=0; j<nseq; j++ )
 111                         if( seq[0][i] != seq[j][i] ) break;
 112                 if( j == nseq )
 113                 {
 114                         mark[i] = '*';
 115                         continue;
 116                 }
 117                 for( j=0; j<nseq; j++ )
 118                         if( amino_grp[(int)seq[0][i]] != amino_grp[(int)seq[j][i]] ) break;
 119                 if( j == nseq )
 120                 {
 121                         mark[i] = '.';
 122                         continue;
 123                 }
 124         }
 125         mark[nlen] = 0;
 126 }
 127
 128 void arguments( int argc, char *argv[] )
 129 {
 130     int c;
 131         namelen = 15;
 132         scoremtx = 1;
 133         nblosum = 62;
 134         dorp = NOTSPECIFIED;
 135         kimuraR = NOTSPECIFIED;
 136         pamN = NOTSPECIFIED;
 137         inputfile = NULL;
 138         comment = NULL;
 139         orderfile = NULL;
 140         format = 'c';
 141
 142     while( --argc > 0 && (*++argv)[0] == '-' )
 143         {
 144         while ( (c = *++argv[0]) )
 145                 {
 146             switch( c )
 147             {
 148                                 case 'i':
 149                                         inputfile = *++argv;
 150                                         fprintf( stderr, "inputfile = %s\n", inputfile );
 151                                         --argc;
 152                                         goto nextoption;
 153                                 case 'c':
 154                                         comment = *++argv;
 155                                         fprintf( stderr, "comment = %s\n", comment );
 156                                         --argc;
 157                                         goto nextoption;
 158                                 case 'r':
 159                                         orderfile = *++argv;
 160                                         fprintf( stderr, "orderfile = %s\n", orderfile );
 161                                         --argc;
 162                                         goto nextoption;
 163                                 case 'n':
 164                                         namelen = atoi( *++argv );
 165                                         fprintf( stderr, "namelen = %d\n", namelen );
 166                                         --argc;
 167                                         goto nextoption;
 168                                 case 'f':
 169                                         format = 'f';
 170                                         break;
 171                                 case 'y':
 172                                         format = 'y';
 173                                         break;
 174                 default:
 175                     fprintf( stderr, "illegal option %c\n", c );
 176                     argc = 0;
 177                     break;
 178             }
 179                 }
 180                 nextoption:
 181                         ;
 182         }
 183     if( argc != 0 )
 184     {
 185         fprintf( stderr, "options: Check source file !\n" );
 186         exit( 1 );
 187     }
 188 }
 189
 190
 191 int main( int argc, char *argv[] )
 192 {
 193         static int  *nlen;
 194         static char **name, **seq, *mark;
 195         static int *order;
 196         int i;
 197         FILE *infp;
 198         FILE *orderfp;
 199         char gett[B];
 200         int nlenmin;
 201
 202         arguments( argc, argv );
 203
 204
 205         if( inputfile )
 206         {
 207                 infp = fopen( inputfile, "r" );
 208                 if( !infp )
 209                 {
 210                         fprintf( stderr, "Cannot open %s\n", inputfile );
 211                         exit( 1 );
 212                 }
 213         }
 214         else
 215                 infp = stdin;
 216
 217         getnumlen_casepreserve( infp, &nlenmin );
 218         rewind( infp );
 219
 220         seq = AllocateCharMtx( njob, nlenmax*2+1 );
 221         mark = AllocateCharVec( nlenmax*2+1 );
 222         order = AllocateIntVec( njob );
 223         name = AllocateCharMtx( njob, B+1 );
 224     nlen = AllocateIntVec( njob );
 225
 226
 227         if( orderfile )
 228         {
 229                 orderfp = fopen( orderfile, "r" );
 230                 if( !orderfile )
 231                 {
 232                         fprintf( stderr, "Cannot open %s\n", orderfile );
 233                         exit( 1 );
 234                 }
 235                 for( i=0; i<njob; i++ )
 236                 {
 237                         fgets( gett, B-1, orderfp );
 238                         order[i] = atoi( gett );
 239                 }
 240                 fclose( orderfp );
 241         }
 242         else
 243         {
 244                 for( i=0; i<njob; i++ ) order[i] = i;
 245         }
 246
 247         readData_pointer_casepreserve( infp, name, nlen, seq );
 248         fclose( infp );
 249
 250         if( format == 'c' || format == 'y' ) for( i=0; i<njob; i++ ) fillspace( seq[i], nlenmax );
 251         constants( njob, seq );
 252
 253 //      initSignalSM();
 254
 255 //      initFiles();
 256
 257
 258
 259 //      setmark( nlenmax, njob, seq, mark );
 260         setmark_clustal( nlenmax, njob, seq, mark );
 261
 262         if( format == 'f' )
 263                 writeData_reorder_pointer( stdout, njob, name, nlen, seq, order );
 264         else if( format == 'c' )
 265                 clustalout_pointer( stdout, njob, nlenmax, seq, name, mark, comment, order, namelen );
 266         else if( format == 'y' )
 267                 phylipout_pointer( stdout, njob, nlenmax, seq, name, order );
 268         else
 269                 fprintf( stderr, "Unknown format\n" );
 270
 271 //      SHOWVERSION;
 272         return( 0 );
 273 }