1 | #include <stdio.h> |
---|
2 | #include <ctype.h> |
---|
3 | #include <string.h> |
---|
4 | #include "convert.h" |
---|
5 | #include "global.h" |
---|
6 | |
---|
7 | extern int warning_out; |
---|
8 | |
---|
9 | /* -------------------------------------------------------------- |
---|
10 | * Function init_alma(). |
---|
11 | * Init. ALMA data. |
---|
12 | */ |
---|
13 | void init_alma() { |
---|
14 | Freespace(&(data.alma.id)); |
---|
15 | Freespace(&(data.alma.filename)); |
---|
16 | Freespace(&(data.alma.sequence)); |
---|
17 | data.alma.format=UNKNOWN; |
---|
18 | data.alma.defgap='-'; |
---|
19 | data.alma.num_of_sequence=0; |
---|
20 | /* init. nbrf too */ |
---|
21 | Freespace(&(data.nbrf.id)); |
---|
22 | Freespace(&(data.nbrf.description)); |
---|
23 | } |
---|
24 | /* ---------------------------------------------------------------- |
---|
25 | * Function alma_to_macke(). |
---|
26 | * Convert from ALMA format to Macke format. |
---|
27 | */ |
---|
28 | void |
---|
29 | alma_to_macke(inf, outf) |
---|
30 | char *inf, *outf; |
---|
31 | { |
---|
32 | FILE *IFP, *ofp; |
---|
33 | FILE_BUFFER ifp; |
---|
34 | char temp[TOKENNUM]; |
---|
35 | int indi, total_num; |
---|
36 | |
---|
37 | if((IFP=fopen(inf, "r"))==NULL) { |
---|
38 | sprintf(temp, |
---|
39 | "Cannot open input file %s, EXIT.", inf); |
---|
40 | error(46, temp); |
---|
41 | } |
---|
42 | ifp = create_FILE_BUFFER(inf, IFP); |
---|
43 | if(Lenstr(outf) <= 0) ofp = stdout; |
---|
44 | else if((ofp=fopen(outf, "w"))==NULL) { |
---|
45 | sprintf(temp, |
---|
46 | "Cannot open output file %s, EXIT.", outf); |
---|
47 | error(47, temp); |
---|
48 | } |
---|
49 | |
---|
50 | init(); |
---|
51 | /* macke format seq irrelenvant header */ |
---|
52 | macke_out_header(ofp); |
---|
53 | for(indi=0; indi<3; indi++) { |
---|
54 | FILE_BUFFER_rewind(ifp); |
---|
55 | init_seq_data(); |
---|
56 | init_macke(); |
---|
57 | init_alma(); |
---|
58 | while(alma_in(ifp)!=EOF) { |
---|
59 | data.numofseq++; |
---|
60 | if(atom()) { |
---|
61 | /* convert from alma form to macke form */ |
---|
62 | switch(indi) { |
---|
63 | case 0: |
---|
64 | /* output seq display format */ |
---|
65 | macke_out0(ofp, ALMA); |
---|
66 | break; |
---|
67 | case 1: |
---|
68 | /* output seq information */ |
---|
69 | macke_out1(ofp); |
---|
70 | break; |
---|
71 | case 2: |
---|
72 | /* output seq data */ |
---|
73 | macke_out2(ofp); |
---|
74 | break; |
---|
75 | default: ; |
---|
76 | } |
---|
77 | } else error(48, |
---|
78 | "Conversion from alma to macke fails, Exit."); |
---|
79 | init_alma(); |
---|
80 | init_macke(); |
---|
81 | } |
---|
82 | total_num = data.numofseq; |
---|
83 | if(indi==0) { |
---|
84 | fprintf(ofp, "#-\n"); |
---|
85 | /* no warning messages for next loop */ |
---|
86 | warning_out = 0; |
---|
87 | } |
---|
88 | } /* for each seq; loop */ |
---|
89 | warning_out = 1; /* resume warning messages */ |
---|
90 | |
---|
91 | #ifdef log |
---|
92 | fprintf(stderr, |
---|
93 | "Total %d sequences have been processed\n", |
---|
94 | total_num); |
---|
95 | #endif |
---|
96 | } |
---|
97 | /* ---------------------------------------------------------------- |
---|
98 | * Function alma_to_genbank(). |
---|
99 | * Convert from ALMA format to GenBank format. |
---|
100 | */ |
---|
101 | void |
---|
102 | alma_to_genbank(inf, outf) |
---|
103 | char *inf, *outf; |
---|
104 | { |
---|
105 | FILE *IFP, *ofp; |
---|
106 | FILE_BUFFER ifp; |
---|
107 | char temp[TOKENNUM]; |
---|
108 | |
---|
109 | if((IFP=fopen(inf, "r"))==NULL) { |
---|
110 | sprintf(temp, |
---|
111 | "Cannot open input file %s, EXIT.", inf); |
---|
112 | error(73, temp); |
---|
113 | } |
---|
114 | ifp = create_FILE_BUFFER(inf, IFP); |
---|
115 | if(Lenstr(outf) <= 0) ofp = stdout; |
---|
116 | else if((ofp=fopen(outf, "w"))==NULL) { |
---|
117 | sprintf(temp, |
---|
118 | "Cannot open output file %s, EXIT.", outf); |
---|
119 | error(74, temp); |
---|
120 | } |
---|
121 | |
---|
122 | init(); |
---|
123 | init_seq_data(); |
---|
124 | init_macke(); |
---|
125 | init_genbank(); |
---|
126 | init_alma(); |
---|
127 | while(alma_in(ifp)!=EOF) { |
---|
128 | data.numofseq++; |
---|
129 | if(atom()&&mtog()) { |
---|
130 | genbank_out(ofp); |
---|
131 | init_alma(); |
---|
132 | init_macke(); |
---|
133 | init_genbank(); |
---|
134 | } |
---|
135 | } /* for each seq; loop */ |
---|
136 | |
---|
137 | #ifdef log |
---|
138 | fprintf(stderr, |
---|
139 | "Total %d sequences have been processed\n", |
---|
140 | data.numofseq); |
---|
141 | #endif |
---|
142 | } |
---|
143 | /* ---------------------------------------------------------------- |
---|
144 | * Function alma_in(). |
---|
145 | * Read in one ALMA sequence data. |
---|
146 | */ |
---|
147 | char |
---|
148 | alma_in(fp) |
---|
149 | FILE_BUFFER fp; |
---|
150 | { |
---|
151 | char eoen, *eof, line[LINENUM]; |
---|
152 | /* char temp[LINENUM]; */ |
---|
153 | char key[TOKENNUM], *format; |
---|
154 | int indi, len, index; |
---|
155 | /* int alma_key_word(); */ |
---|
156 | /* void alma_one_entry(), Freespace(), error(); */ |
---|
157 | |
---|
158 | /* end-of-entry; set to be 'y' after reaching end of entry */ |
---|
159 | format=NULL; |
---|
160 | eoen = ' '; |
---|
161 | for(eof=Fgetline(line, LINENUM, fp); eof!=NULL&&eoen!='y'; ) |
---|
162 | { |
---|
163 | if(Lenstr(line)<=1) { |
---|
164 | eof=Fgetline(line, LINENUM, fp); |
---|
165 | continue; /* skip empty line */ |
---|
166 | } |
---|
167 | index = alma_key_word(line, 0, key, TOKENNUM); |
---|
168 | eoen='n'; |
---|
169 | if((Cmpstr(key, "NXT ENTRY"))==EQ) { |
---|
170 | /* do nothing, just indicate beginning of entry */ |
---|
171 | } else if((Cmpstr(key, "ENTRY ID"))==EQ) { |
---|
172 | alma_one_entry(line, index, &(data.alma.id)); |
---|
173 | for(indi=0, len=Lenstr(data.alma.id); indi<len; |
---|
174 | indi++) |
---|
175 | if(data.alma.id[indi]==' ') |
---|
176 | data.alma.id[indi]='_'; |
---|
177 | } else if((Cmpstr(key, "SEQUENCE"))==EQ) { |
---|
178 | alma_one_entry(line, index, |
---|
179 | &(data.alma.filename)); |
---|
180 | } else if((Cmpstr(key, "FORMAT"))==EQ) { |
---|
181 | alma_one_entry(line, index, &format); |
---|
182 | if((Cmpstr(format, "NBRF"))==EQ) |
---|
183 | data.alma.format=NBRF; |
---|
184 | else if((Cmpstr(format, "UWGCG"))==EQ |
---|
185 | ||(Cmpstr(format, "GCG"))==EQ) |
---|
186 | data.alma.format=GCG; |
---|
187 | else if((Cmpstr(format, "EMBL"))==EQ) |
---|
188 | data.alma.format=EMBL; |
---|
189 | else if((Cmpstr(format, "STADEN"))==EQ) |
---|
190 | data.alma.format=STADEN; |
---|
191 | else { |
---|
192 | sprintf(line, |
---|
193 | "Unidentified file format %s in ALMA format, Exit.", |
---|
194 | format); |
---|
195 | error(49, line); |
---|
196 | } |
---|
197 | Freespace(&(format)); |
---|
198 | } else if((Cmpstr(key, "DEFGAP"))==EQ) { |
---|
199 | /* skip white spaces */ |
---|
200 | for(;line[index]!='['&&line[index]!='\n' |
---|
201 | &&line[index]!='\0'; index++) ; |
---|
202 | if(line[index]=='[') |
---|
203 | data.alma.defgap = line[index+1]; |
---|
204 | } else if((Cmpstr(key, "GAPS"))==EQ) { |
---|
205 | eof=alma_in_gaps(fp); |
---|
206 | eoen='y'; |
---|
207 | } else if((Cmpstr(key, "END FILE"))==EQ) { |
---|
208 | return(EOF); |
---|
209 | } |
---|
210 | if(eoen!='y') eof=Fgetline(line, LINENUM, fp); |
---|
211 | } |
---|
212 | if(eof==NULL) return(EOF); |
---|
213 | else return(EOF+1); |
---|
214 | } |
---|
215 | /* ---------------------------------------------------------------- |
---|
216 | * Function alma_key_word(). |
---|
217 | * Get the key_word from line beginning at index. |
---|
218 | */ |
---|
219 | int |
---|
220 | alma_key_word(line, index, key, length) |
---|
221 | char *line; |
---|
222 | int index; |
---|
223 | char *key; |
---|
224 | int length; |
---|
225 | { |
---|
226 | int indi, indj; |
---|
227 | |
---|
228 | if(line==NULL) { key[0]='\0'; return(index); } |
---|
229 | for(indi=index, indj=0; |
---|
230 | (index-indi)<length&&line[indi]!='>' |
---|
231 | &&line[indi]!='\n'&&line[indi]!='\0'; |
---|
232 | indi++, indj++) |
---|
233 | key[indj] = line[indi]; |
---|
234 | key[indj] = '\0'; |
---|
235 | if(line[indi]=='>') return(indi+1); |
---|
236 | else return(indi); |
---|
237 | } |
---|
238 | /* -------------------------------------------------------------- |
---|
239 | * Function alma_one_entry(). |
---|
240 | * Read in one ALMA entry lines. |
---|
241 | */ |
---|
242 | void |
---|
243 | alma_one_entry(line, index, datastring) |
---|
244 | char *line; |
---|
245 | int index; |
---|
246 | char **datastring; |
---|
247 | { |
---|
248 | int indi, length; |
---|
249 | /* void replace_entry(); */ |
---|
250 | |
---|
251 | index = Skip_white_space(line, index); |
---|
252 | |
---|
253 | for(indi=index, length=Lenstr(line+index)+index; |
---|
254 | indi<length; indi++) |
---|
255 | if(line[indi]=='\n') line[indi]='\0'; |
---|
256 | |
---|
257 | replace_entry(datastring, line+index); |
---|
258 | } |
---|
259 | /* ------------------------------------------------------------- |
---|
260 | * Function alma_in_gaps(). |
---|
261 | * Get sequence data and merge with gaps(alignment). |
---|
262 | */ |
---|
263 | char |
---|
264 | *alma_in_gaps(fp) |
---|
265 | FILE_BUFFER fp; |
---|
266 | { |
---|
267 | int return_num, gaps, residues, count=0; |
---|
268 | int indi, numofseq, bases_not_matching; |
---|
269 | char line[LINENUM], gap_chars[LINENUM]; |
---|
270 | char *eof; |
---|
271 | /* void alma_in_sequence(), warning(); */ |
---|
272 | |
---|
273 | alma_in_sequence(); |
---|
274 | |
---|
275 | numofseq = 0; |
---|
276 | |
---|
277 | bases_not_matching=0; |
---|
278 | /* alignment, merger with gaps information */ |
---|
279 | { |
---|
280 | while (1) { |
---|
281 | char *gotLine = Fgetline(line, LINENUM, fp); |
---|
282 | if (!gotLine) break; |
---|
283 | |
---|
284 | return_num = sscanf(line, "%d %d %s", &gaps, &residues, gap_chars); |
---|
285 | if (return_num == 0 || gaps == -1) { |
---|
286 | FILE_BUFFER_back(fp, line); |
---|
287 | break; |
---|
288 | } |
---|
289 | |
---|
290 | data.alma.sequence = (char*)Reallocspace(data.alma.sequence, (unsigned)(sizeof(char)*(numofseq+gaps+1))); |
---|
291 | |
---|
292 | for(indi=0; indi<gaps; indi++) data.alma.sequence[numofseq+indi] = gap_chars[1]; |
---|
293 | |
---|
294 | numofseq += gaps; |
---|
295 | |
---|
296 | if(residues>(data.seq_length-count)) bases_not_matching = 1; |
---|
297 | |
---|
298 | data.alma.sequence = (char*)Reallocspace (data.alma.sequence, (unsigned)(sizeof(char)*(numofseq+residues+1))); |
---|
299 | |
---|
300 | /* fill with gap if seq data is not enough as required */ |
---|
301 | for(indi=0; indi<residues; indi++) { |
---|
302 | if(count>=data.seq_length) data.alma.sequence[numofseq+indi] = gap_chars[1]; |
---|
303 | else data.alma.sequence[numofseq+indi] = data.sequence[count++]; |
---|
304 | } |
---|
305 | |
---|
306 | numofseq += residues; |
---|
307 | |
---|
308 | } |
---|
309 | } |
---|
310 | |
---|
311 | if(bases_not_matching) { |
---|
312 | sprintf(line, |
---|
313 | "Bases number in ALMA file is larger than that in data file %s", |
---|
314 | data.alma.filename); |
---|
315 | |
---|
316 | warning(142, line); |
---|
317 | } |
---|
318 | |
---|
319 | if(count<data.seq_length) { |
---|
320 | sprintf(line, |
---|
321 | "Bases number in ALMA file is less than that in data file %s", |
---|
322 | data.alma.filename); |
---|
323 | |
---|
324 | warning(143, line); |
---|
325 | |
---|
326 | /* Append the rest of the data at the end */ |
---|
327 | data.alma.sequence = (char*)Reallocspace (data.alma.sequence, (unsigned)(sizeof(char)*(numofseq+data.seq_length-count+1))); |
---|
328 | |
---|
329 | for(indi=0; count<data.seq_length; indi++) data.alma.sequence[numofseq++] = data.sequence[count++]; |
---|
330 | } |
---|
331 | data.alma.sequence[numofseq]='\0'; |
---|
332 | |
---|
333 | /* update sequence data */ |
---|
334 | if(numofseq>data.max) { |
---|
335 | data.max=numofseq; |
---|
336 | data.sequence=(char*)Reallocspace(data.sequence, (unsigned)(sizeof(char)*data.max)); |
---|
337 | } |
---|
338 | |
---|
339 | data.seq_length = numofseq; |
---|
340 | |
---|
341 | for(indi=0; indi<numofseq; indi++) data.sequence[indi]=data.alma.sequence[indi]; |
---|
342 | |
---|
343 | data.alma.num_of_sequence = numofseq; |
---|
344 | |
---|
345 | if(return_num!=0) { |
---|
346 | /* skip END ENTRY> line */ |
---|
347 | if(Fgetline(line, LINENUM, fp)==NULL) eof=NULL; |
---|
348 | else eof = line; |
---|
349 | } else eof = NULL; |
---|
350 | |
---|
351 | return(eof); |
---|
352 | } |
---|
353 | /* ------------------------------------------------------------- |
---|
354 | * Function alma_in_sequence(). |
---|
355 | * Read in sequence data. |
---|
356 | */ |
---|
357 | void |
---|
358 | alma_in_sequence() { |
---|
359 | |
---|
360 | char temp[LINENUM]; |
---|
361 | FILE *IFP; /* ex-infile */ |
---|
362 | FILE_BUFFER ifp; |
---|
363 | |
---|
364 | if((IFP=fopen(data.alma.filename, "r"))==NULL) { |
---|
365 | sprintf(temp, "Cannot open file %s, Exit.", data.alma.filename); |
---|
366 | error(51, temp); |
---|
367 | } |
---|
368 | |
---|
369 | ifp = create_FILE_BUFFER(data.alma.filename, IFP); |
---|
370 | |
---|
371 | if(data.alma.format==NBRF) { |
---|
372 | nbrf_in(ifp); |
---|
373 | } else if(data.alma.format==GCG) { |
---|
374 | gcg_in(ifp); |
---|
375 | } else if(data.alma.format==EMBL) { |
---|
376 | init_embl(); |
---|
377 | embl_in(ifp); |
---|
378 | } else if(data.alma.format==STADEN) { |
---|
379 | staden_in(ifp); |
---|
380 | } else { |
---|
381 | sprintf(temp, "Unidentified file format %d in ALMA file, Exit.", data.alma.format); |
---|
382 | error(50, temp); |
---|
383 | } |
---|
384 | destroy_FILE_BUFFER(ifp); |
---|
385 | } |
---|
386 | /* -------------------------------------------------------------- |
---|
387 | * Function nbrf_in(). |
---|
388 | * Read in nbrf data. |
---|
389 | */ |
---|
390 | void |
---|
391 | nbrf_in(fp) |
---|
392 | FILE_BUFFER fp; |
---|
393 | { |
---|
394 | int length, index, reach_end; |
---|
395 | char line[LINENUM], temp[TOKENNUM], *eof; |
---|
396 | /* char *Fgetline(), *Dupstr(); */ |
---|
397 | /* char *Reallocspace(); */ |
---|
398 | /* void Cpystr(), replace_entry(), error(); */ |
---|
399 | |
---|
400 | if((eof=Fgetline(line, LINENUM, fp))==NULL) |
---|
401 | error(52, |
---|
402 | "Cannot find id line in NBRF file, Exit."); |
---|
403 | Cpystr(temp, line+4); |
---|
404 | length=Lenstr(temp); |
---|
405 | if(temp[length-1]=='\n') temp[length-1]='\0'; |
---|
406 | |
---|
407 | replace_entry(&(data.nbrf.id), temp); |
---|
408 | |
---|
409 | if((eof=Fgetline(line, LINENUM, fp))==NULL) |
---|
410 | error(54, |
---|
411 | "Cannot find description line in NBRF file, Exit."); |
---|
412 | |
---|
413 | replace_entry(&(data.nbrf.description), line); |
---|
414 | |
---|
415 | /* read in sequence data */ |
---|
416 | data.seq_length = 0; |
---|
417 | for(eof=Fgetline(line, LINENUM, fp), reach_end='n'; |
---|
418 | eof!=NULL&&reach_end!='y'; |
---|
419 | eof=Fgetline(line, LINENUM, fp)) |
---|
420 | { |
---|
421 | for(index=0; line[index]!='\n'&&line[index]!='\0'; |
---|
422 | index++) |
---|
423 | { |
---|
424 | if(line[index]=='*') { |
---|
425 | reach_end='y'; |
---|
426 | continue; |
---|
427 | } |
---|
428 | if(line[index]!=' '&&data.seq_length>=data.max){ |
---|
429 | data.max += 100; |
---|
430 | |
---|
431 | data.sequence = (char*)Reallocspace |
---|
432 | (data.sequence, |
---|
433 | (unsigned)(sizeof(char)*data.max)); |
---|
434 | } |
---|
435 | if(line[index]!=' ') |
---|
436 | data.sequence[data.seq_length++] |
---|
437 | = line[index]; |
---|
438 | } /* scan through each line */ |
---|
439 | data.sequence[data.seq_length] = '\0'; |
---|
440 | } /* for each line */ |
---|
441 | } |
---|
442 | /* ---------------------------------------------------------------- |
---|
443 | * Function gcg_in(). |
---|
444 | * Read in one data entry in UWGCG format. |
---|
445 | */ |
---|
446 | void |
---|
447 | gcg_in(fp) |
---|
448 | FILE_BUFFER fp; |
---|
449 | { |
---|
450 | char line[LINENUM]; |
---|
451 | |
---|
452 | while (Fgetline(line, LINENUM, fp)) { |
---|
453 | char *two_dots = strstr(line, ".."); |
---|
454 | if (two_dots) { |
---|
455 | FILE_BUFFER_back(fp, two_dots+2); |
---|
456 | genbank_origin(line, fp); |
---|
457 | } |
---|
458 | } |
---|
459 | } |
---|
460 | /* ---------------------------------------------------------------- |
---|
461 | * Fnction staden_in(). |
---|
462 | * Read in sequence data in STADEN format. |
---|
463 | */ |
---|
464 | void |
---|
465 | staden_in(fp) |
---|
466 | FILE_BUFFER fp; |
---|
467 | { |
---|
468 | char line[LINENUM]; |
---|
469 | int len, start, indi; |
---|
470 | |
---|
471 | data.seq_length=0; |
---|
472 | while(Fgetline(line, LINENUM, fp)!=NULL) { |
---|
473 | /* skip empty line */ |
---|
474 | if((len=Lenstr(line))<=1) continue; |
---|
475 | |
---|
476 | start = data.seq_length; |
---|
477 | data.seq_length += len; |
---|
478 | line[len-1]='\0'; |
---|
479 | |
---|
480 | if(data.seq_length>data.max) { |
---|
481 | data.max += (data.seq_length + 100); |
---|
482 | data.sequence = (char*)Reallocspace |
---|
483 | (data.sequence, |
---|
484 | (unsigned)(sizeof(char)*data.max)); |
---|
485 | } |
---|
486 | |
---|
487 | for(indi=0; indi<len; indi++) |
---|
488 | data.sequence[start+indi] = line[indi]; |
---|
489 | } |
---|
490 | } |
---|
491 | /* ------------------------------------------------------------- |
---|
492 | * Function atom(). |
---|
493 | * Convert one ALMA entry to Macke entry. |
---|
494 | */ |
---|
495 | int |
---|
496 | atom() |
---|
497 | { |
---|
498 | /* int indi, len; */ |
---|
499 | /* char *Dupstr(), *Reallocspace(); */ |
---|
500 | /* void replace_entry(), error(); */ |
---|
501 | |
---|
502 | if(data.alma.format==NBRF) { |
---|
503 | data.macke.numofrem=1; |
---|
504 | data.macke.remarks=(char**)Reallocspace |
---|
505 | ((char*)data.macke.remarks, |
---|
506 | (unsigned)(sizeof(char*))); |
---|
507 | data.macke.remarks[0] |
---|
508 | =(char*)Dupstr(data.nbrf.description); |
---|
509 | } else if(data.alma.format==EMBL) { |
---|
510 | etom(); |
---|
511 | } else if(data.alma.format==GCG) { |
---|
512 | } else if(data.alma.format==STADEN) { |
---|
513 | } else { |
---|
514 | error(53, |
---|
515 | "Unidentified format type in ALMA file, Exit."); |
---|
516 | } |
---|
517 | |
---|
518 | replace_entry(&(data.macke.seqabbr), data.alma.id); |
---|
519 | |
---|
520 | return(1); |
---|
521 | } |
---|
522 | /* ------------------------------------------------------------- |
---|
523 | * Function embl_to_alma(). |
---|
524 | * Convert from EMBL to ALMA. |
---|
525 | */ |
---|
526 | void |
---|
527 | embl_to_alma(inf, outf) |
---|
528 | char *inf, *outf; |
---|
529 | { |
---|
530 | FILE *IFP, *ofp; |
---|
531 | FILE_BUFFER ifp; |
---|
532 | char temp[TOKENNUM]; |
---|
533 | |
---|
534 | if((IFP=fopen(inf, "r"))==NULL) { |
---|
535 | sprintf(temp, "Cannot open input file %s, EXIT.", inf); |
---|
536 | error(134, temp); |
---|
537 | } |
---|
538 | ifp = create_FILE_BUFFER(inf, IFP); |
---|
539 | if((ofp=fopen(outf, "w"))==NULL) { |
---|
540 | sprintf(temp, "Cannot open output file %s, EXIT.", outf); |
---|
541 | error(135, temp); |
---|
542 | } |
---|
543 | |
---|
544 | init(); |
---|
545 | init_embl(); |
---|
546 | init_alma(); |
---|
547 | alma_out_header(ofp); |
---|
548 | |
---|
549 | while(embl_in(ifp)!=EOF) { |
---|
550 | if(data.numofseq>0) fprintf(ofp, "\n"); |
---|
551 | data.numofseq++; |
---|
552 | if(etoa()) { |
---|
553 | FILE *outfile = alma_out(ofp, EMBL); |
---|
554 | embl_out(outfile); |
---|
555 | fclose(outfile); |
---|
556 | } |
---|
557 | init_embl(); |
---|
558 | init_alma(); |
---|
559 | } |
---|
560 | |
---|
561 | fprintf(ofp, "END FILE>\n"); |
---|
562 | |
---|
563 | #ifdef log |
---|
564 | fprintf(stderr, |
---|
565 | "Total %d sequences have been processed\n", |
---|
566 | data.numofseq); |
---|
567 | #endif |
---|
568 | |
---|
569 | destroy_FILE_BUFFER(ifp); fclose(ofp); |
---|
570 | } |
---|
571 | /* ------------------------------------------------------------ |
---|
572 | * Function genbank_to_alma(). |
---|
573 | * Convert from GenBank to ALMA. |
---|
574 | */ |
---|
575 | void |
---|
576 | genbank_to_alma(inf, outf) |
---|
577 | char *inf, *outf; |
---|
578 | { |
---|
579 | FILE *IFP, *ofp; |
---|
580 | FILE_BUFFER ifp; |
---|
581 | char temp[TOKENNUM]; |
---|
582 | |
---|
583 | if((IFP=fopen(inf, "r"))==NULL) { |
---|
584 | sprintf(temp, |
---|
585 | "Cannot open input file %s, EXIT.", inf); |
---|
586 | error(61, temp); |
---|
587 | } |
---|
588 | ifp = create_FILE_BUFFER(inf, IFP); |
---|
589 | if((ofp=fopen(outf, "w")) == NULL) { |
---|
590 | sprintf(temp, |
---|
591 | "Cannot open output file %s, EXIT.", outf); |
---|
592 | error(62, temp); |
---|
593 | } |
---|
594 | |
---|
595 | init(); |
---|
596 | init_genbank(); |
---|
597 | init_embl(); |
---|
598 | init_alma(); |
---|
599 | alma_out_header(ofp); |
---|
600 | |
---|
601 | while(genbank_in(ifp)!=EOF) { |
---|
602 | if(data.numofseq>0) fprintf(ofp, "\n"); |
---|
603 | data.numofseq++; |
---|
604 | if(gtoe()&&etoa()) { |
---|
605 | FILE *outfile = alma_out(ofp, EMBL); |
---|
606 | embl_out(outfile); |
---|
607 | fclose(outfile); |
---|
608 | } |
---|
609 | init_genbank(); |
---|
610 | init_embl(); |
---|
611 | init_alma(); |
---|
612 | } |
---|
613 | |
---|
614 | fprintf(ofp, "END FILE>\n"); |
---|
615 | |
---|
616 | #ifdef log |
---|
617 | fprintf(stderr, |
---|
618 | "Total %d sequences have been processed\n", |
---|
619 | data.numofseq); |
---|
620 | #endif |
---|
621 | |
---|
622 | destroy_FILE_BUFFER(ifp); fclose(ofp); |
---|
623 | } |
---|
624 | /* ------------------------------------------------------------- |
---|
625 | * Function macke_to_alma(). |
---|
626 | * Convert from MACKE to ALMA. |
---|
627 | */ |
---|
628 | void |
---|
629 | macke_to_alma(inf, outf) |
---|
630 | char *inf, *outf; |
---|
631 | { |
---|
632 | FILE *IFP1, *IFP2, *IFP3, *ofp; |
---|
633 | FILE_BUFFER ifp1, ifp2, ifp3; |
---|
634 | char temp[TOKENNUM]; |
---|
635 | |
---|
636 | if((IFP1=fopen(inf, "r"))==NULL || |
---|
637 | (IFP2=fopen(inf, "r"))==NULL || |
---|
638 | (IFP3=fopen(inf, "r"))==NULL) |
---|
639 | { |
---|
640 | sprintf(temp, |
---|
641 | "Cannot open input file %s, EXIT.", inf); |
---|
642 | error(59, temp); |
---|
643 | } |
---|
644 | |
---|
645 | ifp1 = create_FILE_BUFFER(inf, IFP1); |
---|
646 | ifp2 = create_FILE_BUFFER(inf, IFP2); |
---|
647 | ifp3 = create_FILE_BUFFER(inf, IFP3); |
---|
648 | |
---|
649 | if((ofp=fopen(outf, "w"))==NULL) { |
---|
650 | sprintf(temp, |
---|
651 | "Cannot open output file %s, EXIT.", outf); |
---|
652 | error(60, temp); |
---|
653 | } |
---|
654 | |
---|
655 | init(); |
---|
656 | init_macke(); |
---|
657 | init_genbank(); |
---|
658 | init_embl(); |
---|
659 | init_alma(); |
---|
660 | alma_out_header(ofp); |
---|
661 | while(macke_in(ifp1, ifp2, ifp3)!=EOF) { |
---|
662 | if(data.numofseq>0) fprintf(ofp, "\n"); |
---|
663 | data.numofseq++; |
---|
664 | if(mtog()&>oe()&&partial_mtoe()&&etoa()) { |
---|
665 | FILE *outfile = alma_out(ofp, EMBL); |
---|
666 | embl_out(outfile); |
---|
667 | fclose(outfile); |
---|
668 | } |
---|
669 | init_macke(); |
---|
670 | init_genbank(); |
---|
671 | init_embl(); |
---|
672 | init_alma(); |
---|
673 | } |
---|
674 | |
---|
675 | fprintf(ofp, "END FILE>\n"); |
---|
676 | |
---|
677 | #ifdef log |
---|
678 | fprintf(stderr, |
---|
679 | "Total %d sequences have been processed\n", |
---|
680 | data.numofseq); |
---|
681 | #endif |
---|
682 | |
---|
683 | destroy_FILE_BUFFER(ifp1); destroy_FILE_BUFFER(ifp2); |
---|
684 | destroy_FILE_BUFFER(ifp3); fclose(ofp); |
---|
685 | } |
---|
686 | /* ------------------------------------------------------------- |
---|
687 | * Function etoa(). |
---|
688 | * Convert from EMBL to ALMA format. |
---|
689 | */ |
---|
690 | int |
---|
691 | etoa() |
---|
692 | { |
---|
693 | char temp[TOKENNUM], t1[TOKENNUM], t2[TOKENNUM], t3[TOKENNUM]; |
---|
694 | /* char *Dupstr(), *Catstr(); */ |
---|
695 | /* void embl_key_word(), Cpystr(); */ |
---|
696 | /* void replace_entry(); */ |
---|
697 | |
---|
698 | embl_key_word(data.embl.id, 0, temp, TOKENNUM); |
---|
699 | if(Lenstr(data.embl.dr)>1) { |
---|
700 | /* get short_id from DR line if there is RDP def. */ |
---|
701 | Cpystr(t3, "dummy"); |
---|
702 | sscanf(data.embl.dr, "%s %s %s", t1, t2, t3); |
---|
703 | if(Cmpstr(t1, "RDP;")==EQ) { |
---|
704 | if(Cmpstr(t3, "dummy")!=EQ) { |
---|
705 | Cpystr(temp, t3); |
---|
706 | } else Cpystr(temp, t2); |
---|
707 | temp[Lenstr(temp)-1]='\0'; /* remove '.' */ |
---|
708 | } |
---|
709 | } |
---|
710 | replace_entry(&(data.alma.id), temp); |
---|
711 | |
---|
712 | return(1); |
---|
713 | } |
---|
714 | /* ------------------------------------------------------------- |
---|
715 | * Function alma_out_header(). |
---|
716 | * Output alma format header. |
---|
717 | */ |
---|
718 | void |
---|
719 | alma_out_header(fp) |
---|
720 | FILE *fp; |
---|
721 | { |
---|
722 | fprintf(fp, "SCREEN WIDTH> 80\n"); |
---|
723 | fprintf(fp, "ID COLUMN> 1\n"); |
---|
724 | fprintf(fp, "LINK COLUMN> 10\n"); |
---|
725 | fprintf(fp, "INFO LINE> 24\n"); |
---|
726 | fprintf(fp, "ACTIVE WINDOW> 1 1\n"); |
---|
727 | fprintf(fp, |
---|
728 | "SEQUENCE WINDOW> 1 1 1 23 12 80\n"); |
---|
729 | fprintf(fp, "SCROLL DISTANCE> 1 1 50\n"); |
---|
730 | fprintf(fp, "SEQ NUMBERS> 1 1 0 0\n"); |
---|
731 | fprintf(fp, "CURSOR SCREEN> 1 1 11 54\n"); |
---|
732 | fprintf(fp, "CURSOR ALIGNMENT> 1 1 9 143\n"); |
---|
733 | fprintf(fp, "LINK ACTIVE> 2\n"); |
---|
734 | fprintf(fp, "SCROLL> 2\n"); |
---|
735 | fprintf(fp, "MULTI WINDOWS>F\n"); |
---|
736 | fprintf(fp, "COLOURS>F\n"); |
---|
737 | fprintf(fp, "UPDATE INT#>T\n"); |
---|
738 | fprintf(fp, "HOMOLOGY>F\n"); |
---|
739 | fprintf(fp, "PRINT>\n"); |
---|
740 | fprintf(fp, "FILE_A>\n"); |
---|
741 | fprintf(fp, "FILE_B>\n"); |
---|
742 | fprintf(fp, "FILE_C>\n"); |
---|
743 | fprintf(fp, "FILE_D>\n"); |
---|
744 | fprintf(fp, "FILE_E>\n"); |
---|
745 | fprintf(fp, "ALIGNMENT TITLE>\n"); |
---|
746 | fprintf(fp, "PRINT TITLE> 0\n"); |
---|
747 | fprintf(fp, "PRINT DEVICE> 0\n"); |
---|
748 | fprintf(fp, "SPACERS TOP> 0\n"); |
---|
749 | fprintf(fp, "FONT> 0\n"); |
---|
750 | fprintf(fp, "FONT SIZE> 0\n"); |
---|
751 | fprintf(fp, "PAGE WIDTH> 0\n"); |
---|
752 | fprintf(fp, "PRINT FORMAT> 0\n"); |
---|
753 | fprintf(fp, "PAGE LENGTH> 0\n"); |
---|
754 | fprintf(fp, "SKIP LINES> 0\n"); |
---|
755 | fprintf(fp, "COLOURING> 0\n"); |
---|
756 | fprintf(fp, |
---|
757 | "COLOUR ACTIVE> 0 0 0 0 0 0 0 0 0 0\n"); |
---|
758 | fprintf(fp, "RESIDUE B(LINK)>\n"); |
---|
759 | fprintf(fp, "RESIDUE R(EVERSE)>\n"); |
---|
760 | fprintf(fp, "RESIDUE I(NCREASE)>\n"); |
---|
761 | fprintf(fp, "RESIDUE B+R>\n"); |
---|
762 | fprintf(fp, "RESIDUE B+I>\n"); |
---|
763 | fprintf(fp, "RESIDUE R+I>\n"); |
---|
764 | fprintf(fp, "RESIDUE B+R+I>\n"); |
---|
765 | fprintf(fp, "HOMOEF_A> 0\n"); |
---|
766 | fprintf(fp, "HOMOEF_B> 0\n"); |
---|
767 | fprintf(fp, "HOMOEF_C> 0\n"); |
---|
768 | fprintf(fp, "HOMOEF_D> 0\n"); |
---|
769 | fprintf(fp, "HOMOEF_E> 0\n"); |
---|
770 | fprintf(fp, "HOMOEF_F> 0\n"); |
---|
771 | fprintf(fp, "HOMOEF_G> 0\n"); |
---|
772 | fprintf(fp, "HOMOEF_H> 0\n"); |
---|
773 | fprintf(fp, "HOMOEF_I> 0\n"); |
---|
774 | fprintf(fp, "HOMOEF_J> 0\n"); |
---|
775 | fprintf(fp, "HOMOEF_K> 0\n"); |
---|
776 | fprintf(fp, "HOMOEF_L> 0\n"); |
---|
777 | fprintf(fp, "HOMOEF_M> 0\n"); |
---|
778 | fprintf(fp, "HOMOEF_N> 0\n"); |
---|
779 | fprintf(fp, "HOMOEF_O> 0\n"); |
---|
780 | fprintf(fp, "HOMOEF_P> 0\n"); |
---|
781 | fprintf(fp, "HOMOEF_Q> 0\n"); |
---|
782 | fprintf(fp, "HOMOEF_R> 0\n"); |
---|
783 | fprintf(fp, "HOMOEF_S> 0\n"); |
---|
784 | fprintf(fp, "HOMOEF_T> 0\n"); |
---|
785 | } |
---|
786 | /* ------------------------------------------------------------- |
---|
787 | * Function alma_out(). |
---|
788 | * Output one alma entry. |
---|
789 | */ |
---|
790 | FILE * |
---|
791 | alma_out(fp, format) |
---|
792 | FILE *fp; |
---|
793 | int format; |
---|
794 | { |
---|
795 | int indi, len; |
---|
796 | char filename[TOKENNUM]; |
---|
797 | FILE *outfile; |
---|
798 | |
---|
799 | Cpystr(filename, data.alma.id); |
---|
800 | for(indi=0, len=Lenstr(filename); indi<len; indi++) |
---|
801 | if(!isalnum(filename[indi])) filename[indi] = '_'; |
---|
802 | Catstr(filename, ".EMBL"); |
---|
803 | |
---|
804 | outfile = alma_out_entry_header(fp, data.alma.id, filename, format); |
---|
805 | alma_out_gaps(fp); |
---|
806 | |
---|
807 | return outfile; |
---|
808 | } |
---|
809 | /* -------------------------------------------------------------- |
---|
810 | * Function alma_out_entry_header(). |
---|
811 | * Output one ALMA entry header. |
---|
812 | */ |
---|
813 | FILE * |
---|
814 | alma_out_entry_header(fp, entry_id, filename, format_type) |
---|
815 | FILE *fp; |
---|
816 | char *entry_id, *filename; |
---|
817 | int format_type; |
---|
818 | { |
---|
819 | char temp[TOKENNUM]; |
---|
820 | FILE *outfile = 0; |
---|
821 | |
---|
822 | fprintf(fp, "NXT ENTRY>S\n"); |
---|
823 | fprintf(fp, "ENTRY ID>%s\n", entry_id); |
---|
824 | |
---|
825 | if(fopen(filename, "r")!=NULL) { |
---|
826 | sprintf(temp, |
---|
827 | "file %s is overwritten.", filename); |
---|
828 | warning(55, temp); |
---|
829 | } |
---|
830 | if((outfile=fopen(filename, "w"))==NULL) { |
---|
831 | sprintf(temp, "Cannot open file: %s, Exit.", |
---|
832 | filename); |
---|
833 | error(56, temp); |
---|
834 | } |
---|
835 | fprintf(fp, "SEQUENCE>%s\n", filename); |
---|
836 | if(format_type==EMBL) |
---|
837 | fprintf(fp, "FORMAT>EMBL\n"); |
---|
838 | else if(format_type==NBRF) |
---|
839 | fprintf(fp, "FORMAT>NBRF\n"); |
---|
840 | else if(format_type==GCG) |
---|
841 | fprintf(fp, "FORMAT>UWGCG\n"); |
---|
842 | else if(format_type==STADEN) |
---|
843 | fprintf(fp, "FORMAT>STADEN\n"); |
---|
844 | else error(57, |
---|
845 | "Unknown format type when writing ALMA format, EXIT."); |
---|
846 | |
---|
847 | fprintf(fp, "ACCEPT>ALL\n"); |
---|
848 | fprintf(fp, "DEFGAP>[-]\n"); |
---|
849 | fprintf(fp, "PARAMS>1\n"); |
---|
850 | fprintf(fp, "GAPS>\n"); |
---|
851 | |
---|
852 | return outfile; |
---|
853 | } |
---|
854 | /* -------------------------------------------------------------- |
---|
855 | * Function alma_out_gaps(). |
---|
856 | * Output gaps information of one ALMA entry. |
---|
857 | */ |
---|
858 | void |
---|
859 | alma_out_gaps(fp) |
---|
860 | FILE *fp; |
---|
861 | { |
---|
862 | int indi, index, residue, gnum, rnum, tempcount; |
---|
863 | |
---|
864 | gnum=rnum=0; |
---|
865 | |
---|
866 | if(data.sequence[0]=='.'||data.sequence[0]=='-' |
---|
867 | ||data.sequence[0]=='~') |
---|
868 | residue=0; |
---|
869 | else residue=1; |
---|
870 | |
---|
871 | tempcount=data.seq_length; |
---|
872 | for(indi=index=0; indi<tempcount; indi++) { |
---|
873 | if(data.sequence[indi]=='.' |
---|
874 | ||data.sequence[indi]=='-' |
---|
875 | ||data.sequence[indi]=='~') { |
---|
876 | if(residue) { |
---|
877 | fprintf(fp, " %4d %4d [-]\n", |
---|
878 | gnum, rnum); |
---|
879 | gnum=rnum=residue=0; |
---|
880 | } |
---|
881 | gnum++; data.seq_length--; |
---|
882 | } else { |
---|
883 | residue=1; |
---|
884 | rnum++; |
---|
885 | data.sequence[index++] |
---|
886 | =data.sequence[indi]; |
---|
887 | } |
---|
888 | } |
---|
889 | data.sequence[index]='\0'; |
---|
890 | fprintf(fp, " %4d %4d [-]\n", gnum, rnum); |
---|
891 | fprintf(fp, " -1 -1 [-]\n"); |
---|
892 | fprintf(fp, "END ENTRY>\n"); |
---|
893 | } |
---|