| 1 | #include <sys/time.h> |
|---|
| 2 | #include <stdio.h> |
|---|
| 3 | #include <stdlib.h> |
|---|
| 4 | // #include <malloc.h> |
|---|
| 5 | #include <string.h> |
|---|
| 6 | #include <time.h> |
|---|
| 7 | |
|---|
| 8 | //#include <xview/xview.h> |
|---|
| 9 | //#include <xview/panel.h> |
|---|
| 10 | |
|---|
| 11 | #include <arbdb.h> |
|---|
| 12 | #include <aw_root.hxx> |
|---|
| 13 | #include <aw_device.hxx> |
|---|
| 14 | #include <aw_window.hxx> |
|---|
| 15 | #include <awt.hxx> |
|---|
| 16 | |
|---|
| 17 | #include "gde.hxx" |
|---|
| 18 | #include "GDE_menu.h" |
|---|
| 19 | #include "GDE_def.h" |
|---|
| 20 | #include "GDE_extglob.h" |
|---|
| 21 | |
|---|
| 22 | /* |
|---|
| 23 | Copyright (c) 1990,1991,1992 Steven Smith at the Harvard Genome Laboratory. |
|---|
| 24 | All rights reserved. |
|---|
| 25 | */ |
|---|
| 26 | |
|---|
| 27 | static void AdjustGroups(NA_Alignment *aln) |
|---|
| 28 | { |
|---|
| 29 | size_t i; |
|---|
| 30 | size_t j; |
|---|
| 31 | int c,done=FALSE; |
|---|
| 32 | |
|---|
| 33 | #ifdef HGL |
|---|
| 34 | return; |
|---|
| 35 | #else |
|---|
| 36 | for(c=0;c<200 && !done;c++) |
|---|
| 37 | { |
|---|
| 38 | for(j=1;j<=aln->numgroups;j++) |
|---|
| 39 | { |
|---|
| 40 | done = FALSE; |
|---|
| 41 | for(i=0;i<aln->numelements;i++) |
|---|
| 42 | { |
|---|
| 43 | if(aln->element[i].groupid == j) |
|---|
| 44 | { |
|---|
| 45 | if(aln->element[i].groupf!=NULL || |
|---|
| 46 | aln->element[i].groupb!=NULL) |
|---|
| 47 | done = TRUE; |
|---|
| 48 | else |
|---|
| 49 | aln->element[i].groupid = 0; |
|---|
| 50 | } |
|---|
| 51 | } |
|---|
| 52 | if(done == FALSE) |
|---|
| 53 | { |
|---|
| 54 | for(i=0;i<aln->numelements;i++) |
|---|
| 55 | if(aln->element[i].groupid == aln->numgroups) |
|---|
| 56 | aln->element[i].groupid = j; |
|---|
| 57 | aln->numgroups--; |
|---|
| 58 | } |
|---|
| 59 | } |
|---|
| 60 | if(aln->numgroups == 0) |
|---|
| 61 | done = TRUE; |
|---|
| 62 | } |
|---|
| 63 | return; |
|---|
| 64 | #endif |
|---|
| 65 | } |
|---|
| 66 | |
|---|
| 67 | static void StripSpecial(char *string) |
|---|
| 68 | { |
|---|
| 69 | int i,j,len; |
|---|
| 70 | i=0; |
|---|
| 71 | |
|---|
| 72 | len = strlen(string); |
|---|
| 73 | for(j=0;j<len;j++) |
|---|
| 74 | { |
|---|
| 75 | if(string[j] == '"') |
|---|
| 76 | string[j] = '`'; |
|---|
| 77 | else if(string[j] == '{') |
|---|
| 78 | string[j] = '('; |
|---|
| 79 | else if(string[j] == '}') |
|---|
| 80 | string[j] = ')'; |
|---|
| 81 | } |
|---|
| 82 | //return; |
|---|
| 83 | } |
|---|
| 84 | |
|---|
| 85 | |
|---|
| 86 | static void RemoveQuotes(char *string) |
|---|
| 87 | { |
|---|
| 88 | int i, j, len; |
|---|
| 89 | |
|---|
| 90 | len = strlen(string); |
|---|
| 91 | for(j=0;j<len;j++) { |
|---|
| 92 | if(string[j] == '"') string[j] = ' '; |
|---|
| 93 | } |
|---|
| 94 | |
|---|
| 95 | for(j=0;string[j]==' ' && j<(int)strlen(string);j++) ; |
|---|
| 96 | |
|---|
| 97 | len = strlen(string); |
|---|
| 98 | for(i=0;i<len - j;i++) string[i] = string[i+j]; |
|---|
| 99 | |
|---|
| 100 | for(j=strlen(string)-1; j>=0 && (string[j]=='\n'||string[j]==' '); j--) { |
|---|
| 101 | string[j] = '\0'; |
|---|
| 102 | } |
|---|
| 103 | |
|---|
| 104 | //return; |
|---|
| 105 | } |
|---|
| 106 | |
|---|
| 107 | /* |
|---|
| 108 | * OverWrite(), overwrite all non-default data from a sequence entry |
|---|
| 109 | * onto any entry with the same ID or short name. |
|---|
| 110 | */ |
|---|
| 111 | static int OverWrite(NA_Sequence *thiss,NA_Alignment *aln) |
|---|
| 112 | { |
|---|
| 113 | size_t j; |
|---|
| 114 | int indx = -1; |
|---|
| 115 | NA_Sequence *that; |
|---|
| 116 | for(j=0;j<aln->numelements;j++) |
|---|
| 117 | { |
|---|
| 118 | if(Find2(thiss->id,aln->element[j].id) != -1) |
|---|
| 119 | if(Find2(aln->element[j].id,thiss->id) != -1) |
|---|
| 120 | indx = j; |
|---|
| 121 | } |
|---|
| 122 | if(indx == -1) |
|---|
| 123 | for(j=0;j<aln->numelements;j++) |
|---|
| 124 | { |
|---|
| 125 | if(Find2(thiss->short_name,aln->element[j].short_name)!= -1) |
|---|
| 126 | if(Find2(aln->element[j].short_name,thiss->short_name)!= -1) |
|---|
| 127 | indx = j; |
|---|
| 128 | } |
|---|
| 129 | if(indx != -1) |
|---|
| 130 | { |
|---|
| 131 | that = &(aln->element[indx]); |
|---|
| 132 | if(thiss->seq_name[0]) |
|---|
| 133 | strcpy(that->seq_name,thiss->seq_name); |
|---|
| 134 | if(thiss->barcode[0]) |
|---|
| 135 | strcpy(that->barcode,thiss->barcode); |
|---|
| 136 | if(thiss->contig[0]) |
|---|
| 137 | strcpy(that->contig,thiss->contig); |
|---|
| 138 | if(thiss->membrane[0]) |
|---|
| 139 | strcpy(that->membrane,thiss->membrane); |
|---|
| 140 | if(thiss->authority[0]) |
|---|
| 141 | strcpy(that->authority,thiss->authority); |
|---|
| 142 | if(thiss->short_name[0]) |
|---|
| 143 | strcpy(that->short_name,thiss->short_name); |
|---|
| 144 | if(thiss->description[0]) |
|---|
| 145 | strcpy(that->description,thiss->description); |
|---|
| 146 | if(thiss->sequence) |
|---|
| 147 | { |
|---|
| 148 | free((char*)that->sequence); |
|---|
| 149 | that->sequence = thiss->sequence; |
|---|
| 150 | that->seqlen = thiss->seqlen; |
|---|
| 151 | that->seqmaxlen = thiss->seqmaxlen; |
|---|
| 152 | } |
|---|
| 153 | if(thiss->baggage) |
|---|
| 154 | { |
|---|
| 155 | that->baggage_len += thiss->baggage_len; |
|---|
| 156 | that->baggage_maxlen += thiss->baggage_maxlen; |
|---|
| 157 | if(that->baggage) |
|---|
| 158 | that->baggage = |
|---|
| 159 | Realloc(that->baggage,that->baggage_maxlen*sizeof(char)); |
|---|
| 160 | else |
|---|
| 161 | that->baggage = Calloc(that->baggage_maxlen,sizeof(char)); |
|---|
| 162 | strncat(that->baggage,thiss->baggage,that->baggage_maxlen); |
|---|
| 163 | } |
|---|
| 164 | if(thiss->comments) |
|---|
| 165 | { |
|---|
| 166 | that->comments_len += thiss->comments_len; |
|---|
| 167 | that->comments_maxlen += thiss->comments_maxlen; |
|---|
| 168 | if(that->comments) |
|---|
| 169 | that->comments = |
|---|
| 170 | Realloc(that->comments,that->comments_maxlen*sizeof(char)); |
|---|
| 171 | else |
|---|
| 172 | that->comments = Calloc(that->comments_maxlen,sizeof(char)); |
|---|
| 173 | strncat(that->comments,thiss->comments,that->comments_maxlen); |
|---|
| 174 | } |
|---|
| 175 | if(thiss->cmask) |
|---|
| 176 | { |
|---|
| 177 | free((char*)that->cmask); |
|---|
| 178 | that->cmask = thiss->cmask; |
|---|
| 179 | } |
|---|
| 180 | if(thiss->offset != that->offset) |
|---|
| 181 | that->offset = thiss->offset; |
|---|
| 182 | if(thiss->attr != 0) |
|---|
| 183 | that->attr = thiss->attr; |
|---|
| 184 | if(thiss->groupid != 0) |
|---|
| 185 | { |
|---|
| 186 | that->groupid = thiss->groupid; |
|---|
| 187 | } |
|---|
| 188 | that->groupb = NULL; |
|---|
| 189 | that->groupf = NULL; |
|---|
| 190 | } |
|---|
| 191 | |
|---|
| 192 | return(indx); |
|---|
| 193 | } |
|---|
| 194 | |
|---|
| 195 | void ReadGDE(char *filename,NA_Alignment *dataset,int type) |
|---|
| 196 | { |
|---|
| 197 | int done = FALSE; |
|---|
| 198 | size_t len = 0, j=0; |
|---|
| 199 | int success,count,temp = 0; |
|---|
| 200 | char in_line[GBUFSIZ],c; |
|---|
| 201 | char *buffer,*line; |
|---|
| 202 | size_t buflen = GBUFSIZ; |
|---|
| 203 | int genclen = 0,curelem = 0; |
|---|
| 204 | NA_Sequence *this_elem = NULL,temp_elem; |
|---|
| 205 | FILE *file; |
|---|
| 206 | |
|---|
| 207 | type=0;count=0;c=0; |
|---|
| 208 | |
|---|
| 209 | ErrorOut5(0!=(file = fopen(filename,"r")),"No such file"); |
|---|
| 210 | |
|---|
| 211 | for(;fgets(in_line,GBUFSIZ,file) != 0;) |
|---|
| 212 | { |
|---|
| 213 | for(line = in_line;line[0]==' ' || line[0] == '\t';line++) ; |
|---|
| 214 | |
|---|
| 215 | if(Find2(line,"{")==0) |
|---|
| 216 | { |
|---|
| 217 | this_elem = &temp_elem; |
|---|
| 218 | InitNASeq(this_elem,DNA); |
|---|
| 219 | this_elem->offset = -(dataset->rel_offset); |
|---|
| 220 | } |
|---|
| 221 | else if(Find2(line,"type")==0) |
|---|
| 222 | { |
|---|
| 223 | if(Find(line,"DNA")) |
|---|
| 224 | { |
|---|
| 225 | this_elem->elementtype = DNA; |
|---|
| 226 | this_elem->tmatrix = Default_DNA_Trans; |
|---|
| 227 | this_elem->rmatrix = Default_NA_RTrans; |
|---|
| 228 | } |
|---|
| 229 | else if(Find(line,"RNA")) |
|---|
| 230 | { |
|---|
| 231 | this_elem->elementtype = RNA; |
|---|
| 232 | this_elem->tmatrix = Default_RNA_Trans; |
|---|
| 233 | this_elem->rmatrix = Default_NA_RTrans; |
|---|
| 234 | } |
|---|
| 235 | else if(Find(line,"MASK")) |
|---|
| 236 | { |
|---|
| 237 | this_elem->elementtype = MASK; |
|---|
| 238 | this_elem->rmatrix = NULL; |
|---|
| 239 | this_elem->tmatrix = NULL; |
|---|
| 240 | this_elem->col_lut = NULL; |
|---|
| 241 | } |
|---|
| 242 | else if(Find(line,"TEXT")) |
|---|
| 243 | { |
|---|
| 244 | this_elem->elementtype = TEXT; |
|---|
| 245 | this_elem->rmatrix = NULL; |
|---|
| 246 | this_elem->tmatrix = NULL; |
|---|
| 247 | this_elem->col_lut = NULL; |
|---|
| 248 | } |
|---|
| 249 | else if(Find(line,"PROT")) |
|---|
| 250 | { |
|---|
| 251 | this_elem->elementtype = PROTEIN; |
|---|
| 252 | this_elem->rmatrix = NULL; |
|---|
| 253 | this_elem->tmatrix = NULL; |
|---|
| 254 | this_elem->col_lut = Default_PROColor_LKUP; |
|---|
| 255 | } |
|---|
| 256 | /* |
|---|
| 257 | this_elem->attr = DEFAULT_X_ATTR; |
|---|
| 258 | */ |
|---|
| 259 | } |
|---|
| 260 | else if(Find2(line,"circular")==0) |
|---|
| 261 | { |
|---|
| 262 | sscanf(line,"%*s %d",&temp); |
|---|
| 263 | if(temp == 1) |
|---|
| 264 | { |
|---|
| 265 | this_elem->attr |= IS_CIRCULAR; |
|---|
| 266 | } |
|---|
| 267 | else |
|---|
| 268 | { |
|---|
| 269 | this_elem->attr &= ~IS_CIRCULAR; |
|---|
| 270 | } |
|---|
| 271 | } |
|---|
| 272 | else if(Find2(line,"orig_direction")==0) |
|---|
| 273 | { |
|---|
| 274 | sscanf(line,"%*s %d",&temp); |
|---|
| 275 | if(temp == 1) |
|---|
| 276 | { |
|---|
| 277 | this_elem->attr |= IS_ORIG_5_TO_3; |
|---|
| 278 | this_elem->attr &= ~IS_ORIG_3_TO_5; |
|---|
| 279 | } |
|---|
| 280 | else |
|---|
| 281 | { |
|---|
| 282 | this_elem->attr |= IS_ORIG_3_TO_5; |
|---|
| 283 | this_elem->attr &= ~IS_ORIG_5_TO_3; |
|---|
| 284 | } |
|---|
| 285 | } |
|---|
| 286 | else if(Find2(line,"direction")==0) |
|---|
| 287 | { |
|---|
| 288 | sscanf(line,"%*s %d",&temp); |
|---|
| 289 | if(temp == 1) |
|---|
| 290 | { |
|---|
| 291 | this_elem->attr |= IS_5_TO_3; |
|---|
| 292 | this_elem->attr &= ~IS_3_TO_5; |
|---|
| 293 | } |
|---|
| 294 | else |
|---|
| 295 | { |
|---|
| 296 | this_elem->attr |= IS_3_TO_5; |
|---|
| 297 | this_elem->attr &= ~IS_5_TO_3; |
|---|
| 298 | } |
|---|
| 299 | } |
|---|
| 300 | else if(Find2(line,"orig_strand")==0) |
|---|
| 301 | { |
|---|
| 302 | sscanf(line,"%*s %d",&temp); |
|---|
| 303 | if(temp == 1) |
|---|
| 304 | { |
|---|
| 305 | this_elem->attr |= IS_ORIG_PRIMARY; |
|---|
| 306 | this_elem->attr &= ~IS_ORIG_SECONDARY; |
|---|
| 307 | } |
|---|
| 308 | else |
|---|
| 309 | { |
|---|
| 310 | this_elem->attr |= IS_ORIG_SECONDARY; |
|---|
| 311 | this_elem->attr &= ~IS_ORIG_PRIMARY; |
|---|
| 312 | } |
|---|
| 313 | } |
|---|
| 314 | else if(Find2(line,"strandedness")==0) |
|---|
| 315 | { |
|---|
| 316 | sscanf(line,"%*s %d",&temp); |
|---|
| 317 | if(temp == 1) |
|---|
| 318 | { |
|---|
| 319 | this_elem->attr |= IS_PRIMARY; |
|---|
| 320 | this_elem->attr &= ~IS_SECONDARY; |
|---|
| 321 | } |
|---|
| 322 | else |
|---|
| 323 | { |
|---|
| 324 | this_elem->attr |= IS_SECONDARY; |
|---|
| 325 | this_elem->attr &= ~IS_PRIMARY; |
|---|
| 326 | } |
|---|
| 327 | } |
|---|
| 328 | else if(Find2(line,"creator")==0) |
|---|
| 329 | { |
|---|
| 330 | sscanf(line,"%*s %[^\n]",this_elem->authority); |
|---|
| 331 | RemoveQuotes(this_elem->authority); |
|---|
| 332 | } |
|---|
| 333 | else if(Find2(line,"longname")==0) |
|---|
| 334 | { |
|---|
| 335 | sscanf(line,"%*s %[^\n]",this_elem->seq_name); |
|---|
| 336 | RemoveQuotes(this_elem->seq_name); |
|---|
| 337 | } |
|---|
| 338 | else if(Find2(line,"descrip")==0) |
|---|
| 339 | { |
|---|
| 340 | sscanf(line,"%*s %[^\n]",this_elem->description); |
|---|
| 341 | RemoveQuotes(this_elem->description); |
|---|
| 342 | } |
|---|
| 343 | else if(Find2(line,"name")==0) |
|---|
| 344 | { |
|---|
| 345 | sscanf(line,"%*s %[^\n]",this_elem->short_name); |
|---|
| 346 | RemoveQuotes(this_elem->short_name); |
|---|
| 347 | } |
|---|
| 348 | else if(Find2(line,"group-ID")==0) |
|---|
| 349 | { |
|---|
| 350 | sscanf(line,"%*s %zu",&(this_elem->groupid)); |
|---|
| 351 | dataset->numgroups = |
|---|
| 352 | MAX(this_elem->groupid, dataset->numgroups); |
|---|
| 353 | } |
|---|
| 354 | else if(Find2(line,"sequence-ID")==0) |
|---|
| 355 | { |
|---|
| 356 | sscanf(line,"%*s %[^\n]",this_elem->id); |
|---|
| 357 | RemoveQuotes(this_elem->id); |
|---|
| 358 | } |
|---|
| 359 | else if(Find2(line,"barcode")==0) |
|---|
| 360 | { |
|---|
| 361 | sscanf(line,"%*s %[^\n]",this_elem->barcode); |
|---|
| 362 | RemoveQuotes(this_elem->barcode); |
|---|
| 363 | } |
|---|
| 364 | else if(Find2(line,"membrane")==0) |
|---|
| 365 | { |
|---|
| 366 | sscanf(line,"%*s %[^\n]",this_elem->membrane); |
|---|
| 367 | RemoveQuotes(this_elem->membrane); |
|---|
| 368 | } |
|---|
| 369 | else if(Find2(line,"contig")==0) |
|---|
| 370 | { |
|---|
| 371 | sscanf(line,"%*s %[^\n]",this_elem->contig); |
|---|
| 372 | RemoveQuotes(this_elem->contig); |
|---|
| 373 | } |
|---|
| 374 | else if(Find2(line,"creation-date")==0) |
|---|
| 375 | { |
|---|
| 376 | sscanf(line,"%*s %2d%*c%2d%*c%2d%*c%2d%*c%2d%*c%2d\n", |
|---|
| 377 | &(this_elem->t_stamp.origin.mm), |
|---|
| 378 | &(this_elem->t_stamp.origin.dd), |
|---|
| 379 | &(this_elem->t_stamp.origin.yy), |
|---|
| 380 | &(this_elem->t_stamp.origin.hr), |
|---|
| 381 | &(this_elem->t_stamp.origin.mn), |
|---|
| 382 | &(this_elem->t_stamp.origin.sc)); |
|---|
| 383 | } |
|---|
| 384 | else if(Find2(line,"offset")==0) |
|---|
| 385 | { |
|---|
| 386 | sscanf(line,"%*s %d",&(this_elem->offset)); |
|---|
| 387 | this_elem->offset -= dataset->rel_offset; |
|---|
| 388 | } |
|---|
| 389 | else if(Find2(line,"comments")==0) |
|---|
| 390 | { |
|---|
| 391 | if(this_elem->comments_maxlen ==0) |
|---|
| 392 | buflen = 2048; |
|---|
| 393 | else |
|---|
| 394 | buflen = this_elem->comments_maxlen; |
|---|
| 395 | |
|---|
| 396 | done = FALSE; |
|---|
| 397 | len = this_elem->comments_len; |
|---|
| 398 | |
|---|
| 399 | for(;line[0] != '"';line++) |
|---|
| 400 | if(line[0] == '\0') |
|---|
| 401 | ErrorOut5(0,"Error in input file"); |
|---|
| 402 | line++; |
|---|
| 403 | buffer = Calloc(buflen,sizeof(char)); |
|---|
| 404 | for(;!done;) |
|---|
| 405 | { |
|---|
| 406 | for(j=0;j<strlen(line);j++) |
|---|
| 407 | { |
|---|
| 408 | if(len+strlen(line) >= buflen) |
|---|
| 409 | { |
|---|
| 410 | buflen *=2; |
|---|
| 411 | buffer = Realloc(buffer, |
|---|
| 412 | buflen*sizeof(char)); |
|---|
| 413 | } |
|---|
| 414 | if(line[j] == '"') done = TRUE; |
|---|
| 415 | |
|---|
| 416 | else |
|---|
| 417 | buffer[len++] = line[j]; |
|---|
| 418 | } |
|---|
| 419 | /* |
|---|
| 420 | * Check pad with null |
|---|
| 421 | */ |
|---|
| 422 | buffer[len] = '\0'; |
|---|
| 423 | if(!done) |
|---|
| 424 | { |
|---|
| 425 | if(fgets(in_line,GBUFSIZ,file) == 0) |
|---|
| 426 | done = TRUE; |
|---|
| 427 | line = in_line; |
|---|
| 428 | } |
|---|
| 429 | } |
|---|
| 430 | this_elem->comments = buffer; |
|---|
| 431 | this_elem->comments_len = strlen(buffer); |
|---|
| 432 | this_elem->comments_maxlen = buflen; |
|---|
| 433 | RemoveQuotes(this_elem->comments); |
|---|
| 434 | } |
|---|
| 435 | else if(Find2(line,"sequence")==0) |
|---|
| 436 | { |
|---|
| 437 | buflen = GBUFSIZ; |
|---|
| 438 | done = FALSE; |
|---|
| 439 | len = 0; |
|---|
| 440 | |
|---|
| 441 | buffer = Calloc(buflen,sizeof(char)); |
|---|
| 442 | for(;line[0] != '"';line++) |
|---|
| 443 | if(line[0] == '\0') |
|---|
| 444 | ErrorOut5(0,"Error in input file"); |
|---|
| 445 | |
|---|
| 446 | line++; |
|---|
| 447 | for(;!done;) |
|---|
| 448 | { |
|---|
| 449 | for(j=0;j<strlen(line);j++) |
|---|
| 450 | { |
|---|
| 451 | if(len+strlen(line) >= buflen) |
|---|
| 452 | { |
|---|
| 453 | buflen *=2; |
|---|
| 454 | buffer = Realloc(buffer, |
|---|
| 455 | buflen*sizeof(char)); |
|---|
| 456 | } |
|---|
| 457 | if(line[j] == '"') done = TRUE; |
|---|
| 458 | |
|---|
| 459 | else |
|---|
| 460 | { |
|---|
| 461 | /* |
|---|
| 462 | * If not text, ignore spaces... |
|---|
| 463 | */ |
|---|
| 464 | if(this_elem->elementtype !=TEXT) |
|---|
| 465 | { |
|---|
| 466 | if(line[j]!=' ' && line[j] != |
|---|
| 467 | '\t' && line[j] != '\n') |
|---|
| 468 | buffer[len++] = line[j]; |
|---|
| 469 | } |
|---|
| 470 | else |
|---|
| 471 | if(line[j] != '\t' && line[j] != '\n') |
|---|
| 472 | buffer[len++] = line[j]; |
|---|
| 473 | } |
|---|
| 474 | } |
|---|
| 475 | if(!done) |
|---|
| 476 | { |
|---|
| 477 | if(fgets(in_line,GBUFSIZ,file) == 0) |
|---|
| 478 | done = TRUE; |
|---|
| 479 | line = in_line; |
|---|
| 480 | } |
|---|
| 481 | } |
|---|
| 482 | if(this_elem->rmatrix) { |
|---|
| 483 | for(j=0;j<len;j++) { |
|---|
| 484 | buffer[j]=this_elem->rmatrix[(unsigned char)buffer[j]]; |
|---|
| 485 | } |
|---|
| 486 | } |
|---|
| 487 | this_elem->sequence =(NA_Base*)buffer; |
|---|
| 488 | this_elem->seqlen = len; |
|---|
| 489 | this_elem->seqmaxlen = buflen; |
|---|
| 490 | } |
|---|
| 491 | |
|---|
| 492 | else if (Find2(line,"}")==0) |
|---|
| 493 | { |
|---|
| 494 | if(this_elem->id[0] == '\0') |
|---|
| 495 | strncpy(this_elem->id,uniqueID(),79); |
|---|
| 496 | if(this_elem->short_name[0] == '\0') |
|---|
| 497 | strncpy(this_elem->short_name,this_elem->id,79); |
|---|
| 498 | if(this_elem->seqlen == 0) |
|---|
| 499 | this_elem->protect= |
|---|
| 500 | PROT_BASE_CHANGES+ |
|---|
| 501 | PROT_GREY_SPACE+ |
|---|
| 502 | PROT_WHITE_SPACE+ |
|---|
| 503 | PROT_TRANSLATION; |
|---|
| 504 | genclen = 0; |
|---|
| 505 | /* |
|---|
| 506 | * Make a new sequence entry... |
|---|
| 507 | */ |
|---|
| 508 | |
|---|
| 509 | success = -1; |
|---|
| 510 | if(OVERWRITE) |
|---|
| 511 | success = OverWrite(this_elem,dataset); |
|---|
| 512 | |
|---|
| 513 | if(success == -1) |
|---|
| 514 | { |
|---|
| 515 | curelem = dataset->numelements++; |
|---|
| 516 | if(curelem == 0) |
|---|
| 517 | { |
|---|
| 518 | dataset->element=(NA_Sequence*) |
|---|
| 519 | Calloc(5,sizeof(NA_Sequence)); |
|---|
| 520 | dataset->maxnumelements = 5; |
|---|
| 521 | } |
|---|
| 522 | else if (curelem==dataset->maxnumelements) |
|---|
| 523 | { |
|---|
| 524 | (dataset->maxnumelements) *= 2; |
|---|
| 525 | dataset->element =(NA_Sequence*) |
|---|
| 526 | Realloc((char*)dataset->element, |
|---|
| 527 | dataset->maxnumelements * sizeof(NA_Sequence)); |
|---|
| 528 | } |
|---|
| 529 | dataset->element[curelem] = *this_elem; |
|---|
| 530 | } |
|---|
| 531 | } |
|---|
| 532 | else if(this_elem != NULL) |
|---|
| 533 | { |
|---|
| 534 | if (this_elem->baggage == NULL) |
|---|
| 535 | { |
|---|
| 536 | this_elem->baggage = strdup(line); |
|---|
| 537 | this_elem->baggage_maxlen = |
|---|
| 538 | this_elem->baggage_len = |
|---|
| 539 | strlen(this_elem->baggage)+1; |
|---|
| 540 | } |
|---|
| 541 | else |
|---|
| 542 | { |
|---|
| 543 | this_elem->baggage_len += strlen(line)+1; |
|---|
| 544 | this_elem->baggage = Realloc( |
|---|
| 545 | this_elem->baggage,this_elem->baggage_len * |
|---|
| 546 | sizeof(char)); |
|---|
| 547 | this_elem->baggage_maxlen = |
|---|
| 548 | this_elem->baggage_len; |
|---|
| 549 | |
|---|
| 550 | strncat(this_elem->baggage,line,GBUFSIZ); |
|---|
| 551 | } |
|---|
| 552 | } |
|---|
| 553 | } |
|---|
| 554 | |
|---|
| 555 | fclose(file); |
|---|
| 556 | NormalizeOffset(dataset); |
|---|
| 557 | Regroup(dataset); |
|---|
| 558 | AdjustGroups(dataset); |
|---|
| 559 | return; |
|---|
| 560 | } |
|---|
| 561 | |
|---|
| 562 | |
|---|
| 563 | int WriteGDE(NA_Alignment *aln,char *filename,int method,int maskable) |
|---|
| 564 | { |
|---|
| 565 | int i; |
|---|
| 566 | size_t j; |
|---|
| 567 | int k,mask = -1; |
|---|
| 568 | FILE *file; |
|---|
| 569 | NA_Sequence *this_elem; |
|---|
| 570 | //extern char month[12][6]; |
|---|
| 571 | |
|---|
| 572 | if(aln == NULL) |
|---|
| 573 | return(1); |
|---|
| 574 | //ARB |
|---|
| 575 | //if(aln->na_ddata == NULL) |
|---|
| 576 | // return(1); |
|---|
| 577 | //ARB END |
|---|
| 578 | file = fopen(filename,"w"); |
|---|
| 579 | if(file == NULL) |
|---|
| 580 | { |
|---|
| 581 | Warning("Cannot open file for output"); |
|---|
| 582 | return(1); |
|---|
| 583 | } |
|---|
| 584 | |
|---|
| 585 | if(maskable && method != SELECT_REGION) |
|---|
| 586 | for(j=0;j<aln->numelements;j++) |
|---|
| 587 | if(aln->element[j].elementtype == MASK && |
|---|
| 588 | aln->element[j].selected) |
|---|
| 589 | mask = j; |
|---|
| 590 | |
|---|
| 591 | for(j=0;j<aln->numelements;j++) |
|---|
| 592 | { |
|---|
| 593 | if((aln->element[j].selected && (int)j!=mask && method!=SELECT_REGION) |
|---|
| 594 | || (method == ALL) |
|---|
| 595 | || (aln->element[j].subselected && method == SELECT_REGION)) |
|---|
| 596 | { |
|---|
| 597 | this_elem = &(aln->element[j]); |
|---|
| 598 | /*SeqNorm(this_elem);*/ |
|---|
| 599 | fprintf(file,"{\n"); |
|---|
| 600 | if(this_elem->short_name[0]) |
|---|
| 601 | fprintf(file,"name \"%s\"\n",this_elem->short_name); |
|---|
| 602 | switch(this_elem->elementtype) |
|---|
| 603 | { |
|---|
| 604 | case DNA: |
|---|
| 605 | fprintf(file,"type \"DNA\"\n"); |
|---|
| 606 | break; |
|---|
| 607 | case RNA: |
|---|
| 608 | fprintf(file,"type \"RNA\"\n"); |
|---|
| 609 | break; |
|---|
| 610 | case PROTEIN: |
|---|
| 611 | fprintf(file,"type \"PROTEIN\"\n"); |
|---|
| 612 | break; |
|---|
| 613 | case MASK: |
|---|
| 614 | fprintf(file,"type \"MASK\"\n"); |
|---|
| 615 | break; |
|---|
| 616 | case TEXT: |
|---|
| 617 | fprintf(file,"type \"TEXT\"\n"); |
|---|
| 618 | break; |
|---|
| 619 | } |
|---|
| 620 | if(this_elem->seq_name[0]) |
|---|
| 621 | fprintf(file,"longname %s\n",this_elem->seq_name); |
|---|
| 622 | |
|---|
| 623 | if(this_elem->id[0]) |
|---|
| 624 | fprintf(file,"sequence-ID \"%s\"\n",this_elem->id); |
|---|
| 625 | RemoveQuotes(this_elem->barcode); |
|---|
| 626 | RemoveQuotes(this_elem->contig); |
|---|
| 627 | |
|---|
| 628 | if(this_elem->barcode[0]) |
|---|
| 629 | fprintf(file,"barcode \"%s\"\n",this_elem->barcode); |
|---|
| 630 | if(this_elem->membrane[0]) |
|---|
| 631 | fprintf(file,"membrane \"%s\"\n",this_elem->membrane); |
|---|
| 632 | if(this_elem->contig[0]) |
|---|
| 633 | fprintf(file,"contig \"%s\"\n",this_elem->contig); |
|---|
| 634 | if(this_elem->description[0]) |
|---|
| 635 | fprintf(file,"descrip \"%s\"\n",this_elem->description); |
|---|
| 636 | if(this_elem->authority[0]) |
|---|
| 637 | fprintf(file,"creator \"%s\"\n",this_elem->authority); |
|---|
| 638 | if(this_elem->groupid) |
|---|
| 639 | fprintf(file,"group-ID %zu\n", this_elem->groupid); |
|---|
| 640 | if(this_elem->offset+aln->rel_offset && method!=SELECT_REGION) |
|---|
| 641 | fprintf(file,"offset %d\n",this_elem->offset+aln->rel_offset); |
|---|
| 642 | if(method == SELECT_REGION) |
|---|
| 643 | { |
|---|
| 644 | /* |
|---|
| 645 | * If selecting a region, the offset should be moved to the first |
|---|
| 646 | * non-'0' space in the mask. |
|---|
| 647 | */ |
|---|
| 648 | for(k=this_elem->offset;k<aln->selection_mask_len && aln->selection_mask[k] == '0';k++) ; |
|---|
| 649 | fprintf(file,"offset %d\n", aln->rel_offset+k); |
|---|
| 650 | } |
|---|
| 651 | if(this_elem->t_stamp.origin.mm != 0) |
|---|
| 652 | fprintf(file, |
|---|
| 653 | "creation-date %2d/%2d/%2d %2d:%2d:%2d\n", |
|---|
| 654 | this_elem->t_stamp.origin.mm, |
|---|
| 655 | this_elem->t_stamp.origin.dd, |
|---|
| 656 | (this_elem->t_stamp.origin.yy)>1900? |
|---|
| 657 | (this_elem->t_stamp.origin.yy-1900): |
|---|
| 658 | (this_elem->t_stamp.origin.yy), |
|---|
| 659 | this_elem->t_stamp.origin.hr, |
|---|
| 660 | this_elem->t_stamp.origin.mn, |
|---|
| 661 | this_elem->t_stamp.origin.sc); |
|---|
| 662 | if((this_elem->attr & IS_ORIG_5_TO_3) && |
|---|
| 663 | ((this_elem->attr & IS_ORIG_3_TO_5) == 0)) |
|---|
| 664 | fprintf(file,"orig_direction 1\n"); |
|---|
| 665 | |
|---|
| 666 | if((this_elem->attr & IS_CIRCULAR) ) |
|---|
| 667 | fprintf(file,"circular 1\n"); |
|---|
| 668 | |
|---|
| 669 | if((this_elem->attr & IS_5_TO_3) && |
|---|
| 670 | ((this_elem->attr & IS_3_TO_5) == 0)) |
|---|
| 671 | fprintf(file,"direction 1\n"); |
|---|
| 672 | |
|---|
| 673 | if((this_elem->attr & IS_ORIG_3_TO_5) && |
|---|
| 674 | ((this_elem->attr & IS_ORIG_5_TO_3) == 0)) |
|---|
| 675 | fprintf(file,"orig_direction -1\n"); |
|---|
| 676 | |
|---|
| 677 | if((this_elem->attr & IS_3_TO_5) && |
|---|
| 678 | ((this_elem->attr & IS_5_TO_3) == 0)) |
|---|
| 679 | fprintf(file,"direction -1\n"); |
|---|
| 680 | |
|---|
| 681 | if((this_elem->attr & IS_ORIG_PRIMARY) && |
|---|
| 682 | ((this_elem->attr & IS_ORIG_SECONDARY) == 0)) |
|---|
| 683 | fprintf(file,"orig_strand 1\n"); |
|---|
| 684 | |
|---|
| 685 | if((this_elem->attr & IS_PRIMARY) && |
|---|
| 686 | ((this_elem->attr & IS_SECONDARY) == 0)) |
|---|
| 687 | fprintf(file,"strandedness 1\n"); |
|---|
| 688 | |
|---|
| 689 | if(((this_elem->attr & IS_ORIG_PRIMARY) == 0) && |
|---|
| 690 | (this_elem->attr & IS_ORIG_SECONDARY)) |
|---|
| 691 | fprintf(file,"orig_strand 2\n"); |
|---|
| 692 | |
|---|
| 693 | if(((this_elem->attr & IS_PRIMARY) == 0) && |
|---|
| 694 | (this_elem->attr & IS_SECONDARY)) |
|---|
| 695 | fprintf(file,"strandedness 2\n"); |
|---|
| 696 | |
|---|
| 697 | if(this_elem->comments != NULL) |
|---|
| 698 | { |
|---|
| 699 | StripSpecial(this_elem->comments); |
|---|
| 700 | fprintf(file,"comments \"%s\"\n",this_elem->comments); |
|---|
| 701 | } |
|---|
| 702 | if(this_elem->baggage != NULL) |
|---|
| 703 | { |
|---|
| 704 | if(this_elem-> |
|---|
| 705 | baggage[strlen(this_elem->baggage)-1] == '\n') |
|---|
| 706 | fprintf(file,"%s",this_elem->baggage); |
|---|
| 707 | else |
|---|
| 708 | fprintf(file,"%s\n",this_elem->baggage); |
|---|
| 709 | } |
|---|
| 710 | fprintf(file,"sequence \""); |
|---|
| 711 | if(this_elem->tmatrix) |
|---|
| 712 | { |
|---|
| 713 | if(mask == -1) |
|---|
| 714 | { |
|---|
| 715 | for(k=this_elem->offset;k<this_elem->seqlen+this_elem->offset;k++) |
|---|
| 716 | { |
|---|
| 717 | if(k%60 == 0) |
|---|
| 718 | putc('\n',file); |
|---|
| 719 | if(method == SELECT_REGION) |
|---|
| 720 | { |
|---|
| 721 | if(aln->selection_mask[k] == '1') |
|---|
| 722 | putc(this_elem->tmatrix[getelem(this_elem,k)], |
|---|
| 723 | file); |
|---|
| 724 | } |
|---|
| 725 | else |
|---|
| 726 | putc(this_elem->tmatrix[getelem(this_elem,k)], |
|---|
| 727 | file); |
|---|
| 728 | } |
|---|
| 729 | } |
|---|
| 730 | else |
|---|
| 731 | { |
|---|
| 732 | for(i=0,k=this_elem->offset;k<this_elem->seqlen+this_elem->offset;k++) |
|---|
| 733 | if(aln->element[mask].seqlen+this_elem->offset>k) |
|---|
| 734 | if((char)getelem(&(aln->element[mask]),k) != '0' |
|---|
| 735 | && ((char)getelem(&(aln->element[mask]),k) != '-')) |
|---|
| 736 | { |
|---|
| 737 | if(i%60 == 0) |
|---|
| 738 | putc('\n',file); |
|---|
| 739 | putc(this_elem->tmatrix[getelem(this_elem,k)], |
|---|
| 740 | file); |
|---|
| 741 | i++; |
|---|
| 742 | } |
|---|
| 743 | } |
|---|
| 744 | fprintf(file,"\"\n"); |
|---|
| 745 | } |
|---|
| 746 | else |
|---|
| 747 | { |
|---|
| 748 | if(mask == -1) |
|---|
| 749 | { |
|---|
| 750 | for(k=this_elem->offset;k<this_elem->seqlen+this_elem->offset;k++) |
|---|
| 751 | { |
|---|
| 752 | if(k%60 == 0) |
|---|
| 753 | putc('\n',file); |
|---|
| 754 | if(method == SELECT_REGION) |
|---|
| 755 | { |
|---|
| 756 | if(aln->selection_mask[k] == '1') |
|---|
| 757 | putc(getelem(this_elem,k),file); |
|---|
| 758 | } |
|---|
| 759 | else |
|---|
| 760 | putc(getelem(this_elem,k),file); |
|---|
| 761 | } |
|---|
| 762 | } |
|---|
| 763 | else |
|---|
| 764 | { |
|---|
| 765 | for(i=0,k=this_elem->offset;k<this_elem->seqlen+this_elem->offset;k++) |
|---|
| 766 | if(((aln->element[mask].seqlen)+(aln->element[mask]. |
|---|
| 767 | offset)) > k) |
|---|
| 768 | if((char)getelem(&(aln->element[mask]),k) == '1') |
|---|
| 769 | { |
|---|
| 770 | if(i%60 == 0) |
|---|
| 771 | putc('\n',file); |
|---|
| 772 | putc(getelem(this_elem,k),file); |
|---|
| 773 | i++; |
|---|
| 774 | } |
|---|
| 775 | } |
|---|
| 776 | fprintf(file,"\"\n"); |
|---|
| 777 | } |
|---|
| 778 | fprintf(file,"}\n"); |
|---|
| 779 | } |
|---|
| 780 | } |
|---|
| 781 | fclose(file); |
|---|
| 782 | return(0); |
|---|
| 783 | } |
|---|
| 784 | |
|---|
| 785 | |
|---|
| 786 | |
|---|
| 787 | /* |
|---|
| 788 | * Normalize seq (remove leading indels in the sequence; |
|---|
| 789 | */ |
|---|
| 790 | void SeqNorm(NA_Sequence *seq) |
|---|
| 791 | { |
|---|
| 792 | int len,j,shift_width,trailer; |
|---|
| 793 | char *sequence; |
|---|
| 794 | len = seq->seqlen; |
|---|
| 795 | |
|---|
| 796 | sequence =(char*)seq->sequence; |
|---|
| 797 | |
|---|
| 798 | if(len == 0) return; |
|---|
| 799 | |
|---|
| 800 | if(seq->tmatrix) { |
|---|
| 801 | for(shift_width=0; (shift_width<len) && ((sequence[shift_width]&15) == '\0'); shift_width++) ; |
|---|
| 802 | } |
|---|
| 803 | else { |
|---|
| 804 | for(shift_width=0; (shift_width<len) && (sequence[shift_width] == '-'); shift_width++) ; |
|---|
| 805 | } |
|---|
| 806 | |
|---|
| 807 | for(j=0;j<len-shift_width;j++) { |
|---|
| 808 | sequence[j] = sequence[j+shift_width]; |
|---|
| 809 | } |
|---|
| 810 | |
|---|
| 811 | seq->seqlen -= shift_width; |
|---|
| 812 | seq->offset += shift_width; |
|---|
| 813 | for (trailer=seq->seqlen-1; |
|---|
| 814 | (sequence[trailer] =='-' || sequence[trailer] == '\0') && trailer>=0; |
|---|
| 815 | trailer--) |
|---|
| 816 | { |
|---|
| 817 | sequence[trailer] = '\0'; |
|---|
| 818 | } |
|---|
| 819 | seq->seqlen = trailer+1; |
|---|
| 820 | return; |
|---|
| 821 | } |
|---|
| 822 | |
|---|
| 823 | /* ALWAYS COPY the result from uniqueID() to a char[32], |
|---|
| 824 | * (strlen(hostname)+1+10). Memory is lost when the function |
|---|
| 825 | * is finished. |
|---|
| 826 | */ |
|---|
| 827 | |
|---|
| 828 | char *uniqueID() |
|---|
| 829 | { |
|---|
| 830 | static char vname[32]; |
|---|
| 831 | time_t *tp; |
|---|
| 832 | static int cnt = 0; |
|---|
| 833 | |
|---|
| 834 | tp = (time_t *)Calloc(1, sizeof(time_t)); |
|---|
| 835 | |
|---|
| 836 | time(tp); |
|---|
| 837 | sprintf(vname, "host:%d:%ld", cnt, *tp); |
|---|
| 838 | cnt++; |
|---|
| 839 | Cfree((char*)tp); |
|---|
| 840 | return(vname); |
|---|
| 841 | } |
|---|
| 842 | |
|---|