Context Navigation

mg.c

Revision 5390, 39.3 KB (checked in by westram, 4 years ago)
- TAB-Ex
Property svn:eol-style set to `native` Property svn:keywords set to `Author Date Id Revision`

Line
1	/* genabnk and Macke converting program */
2
3	#include <stdio.h>
4	#include <stdlib.h>
5	#include "convert.h"
6	#include "global.h"
7
8	extern int warning_out;
9
10	/* --------------------------------------------------------------
11	* Function init_gm_data().
12	* Initialize data structure of genbank and Macke formats.
13	*/
14	void
15	init_gm_data() {
16	/* void init_macke(), init_genbank(); */
17
18	init_macke();
19	init_genbank();
20	}
21	/* ----------------------------------------------------------
22	* Function genbank_to_macke().
23	* Convert from Genbank format to Macke format.
24	*/
25	void
26	genbank_to_macke(inf, outf)
27	char inf, outf;
28	{
29	FILE IFP, ofp;
30	FILE_BUFFER ifp;
31	char temp[TOKENNUM];
32	int indi, total_num;
33
34	if((IFP=fopen(inf, "r"))==NULL) {
35	sprintf(temp, "CANNOT open input file %s, exit.\n", inf);
36	error(0, temp);
37	}
38	ifp = create_FILE_BUFFER(inf, IFP);
39	if(Lenstr(outf) <= 0) ofp = stdout;
40	else if((ofp=fopen(outf, "w"))==NULL) {
41	sprintf(temp, "CANNOT open output file %s, exit.\n", outf);
42	error(1, temp);
43	}
44	/* seq irelenvant header */
45	init();
46	init_gm_data();
47	macke_out_header(ofp);
48
49	#ifdef log
50	fprintf(stderr, "Start converting...\n");
51	#endif
52
53	for(indi=0; indi<3; indi++) {
54	FILE_BUFFER_rewind(ifp);
55	init_seq_data();
56	while(genbank_in(ifp)!=EOF) {
57	data.numofseq++;
58	if(gtom()) {
59	/* convert from genbank form to macke form */
60	switch(indi) {
61	case 0:
62	/* output seq display format */
63	macke_out0(ofp, GENBANK);
64	break;
65	case 1:
66	/* output seq information */
67	macke_out1(ofp);
68	break;
69	case 2:
70	/* output seq data */
71	macke_out2(ofp);
72	break;
73	default: ;
74	}
75	} else error(7,
76	"Conversion from genbank to macke fails, Exit");
77	init_gm_data();
78	#ifdef log
79	if((data.numofseq % 100)==0)
80	fprintf(stderr, "%d sequences have been processed\n",
81	data.numofseq);
82	#endif
83	}
84	total_num = data.numofseq;
85	if(indi==0) {
86	fprintf(ofp, "#-\n");
87	/* no warning message for next loop */
88	warning_out = 0;
89	}
90	} /* for each seq; loop */
91
92	warning_out = 1; /* resume warning messages */
93
94	#ifdef log
95	fprintf(stderr,
96	"Total %d sequences have been processed\n", total_num);
97	#endif
98
99	}
100	/* --------------------------------------------------------------
101	* Function gtom().
102	* Convert from Genbank format to Macke format.
103	*/
104	int
105	gtom() {
106	/* void genbank_key_word(), error(), Freespace(), Cpystr(); */
107	/* void Append(), gtom_remarks(), replace_entry(); */
108	/* char Catstr(); /
109	char temp[LONGTEXT], buffer[TOKENNUM];
110	char genus[TOKENNUM], species[TOKENNUM];
111	/* char genbank_date(), today_date(); */
112	/* char genbank_get_strain(), genbank_get_subspecies(); */
113	/* char genbank_get_atcc(); /
114	/* int Lenstr(), num_of_remark(), Cmpstr(); */
115	/* int len; */
116	/* int indj, indk, remnum; */
117
118	/* copy seq abbr, assume every entry in gbk must end with \n\0 */
119	/* no '\n' at the end of the string */
120	genbank_key_word(data.gbk.locus, 0, temp, TOKENNUM);
121	replace_entry(&(data.macke.seqabbr), temp);
122
123	/* copy name and definition*/
124	if(Lenstr(data.gbk.organism)>1)
125	replace_entry(&(data.macke.name),data.gbk.organism);
126	else if(Lenstr(data.gbk.definition)>1) {
127	sscanf(data.gbk.definition, "%s %s", genus, species);
128	if(species[Lenstr(species)-1]==';')
129	species[Lenstr(species)-1] = '\0';
130	sprintf(temp, "%s %s\n", genus, species);
131	replace_entry(&(data.macke.name), temp);
132	}
133
134	/* copy cc name and number */
135	if(Lenstr(data.gbk.comments.orginf.cc)>1)
136	replace_entry(&(data.macke.atcc),
137	data.gbk.comments.orginf.cc);
138
139	/* copy rna(methods) */
140	if(Lenstr(data.gbk.comments.seqinf.methods)>1)
141	replace_entry(&(data.macke.rna),
142	data.gbk.comments.seqinf.methods);
143
144	/* copy date---DD-MMM-YYYY\n\0 */
145	if(Lenstr(data.gbk.locus)<61) {
146	data.macke.date = genbank_date(today_date());
147	Append(&(data.macke.date), "\n");
148	} else
149	replace_entry(&(data.macke.date), data.gbk.locus+50);
150
151	/* copy genbank entry (gbkentry has higher priority than gbk.accession)*/
152	if(Lenstr(data.gbk.comments.seqinf.gbkentry)>1)
153	replace_entry(&(data.macke.acs),
154	data.gbk.comments.seqinf.gbkentry);
155	else {
156	if(Lenstr(data.gbk.accession)>1
157	&&Cmpstr(data.gbk.accession, "No information\n")!=EQ) {
158	sscanf(data.gbk.accession, "%s", buffer);
159	Catstr(buffer, "\n");
160	} else Cpystr(buffer, "\n");
161	replace_entry(&(data.macke.acs), buffer);
162	}
163
164	/* copy the first reference from GenBank to Macke */
165	if(data.gbk.numofref>0) {
166	if(Lenstr(data.gbk.reference[0].author)>1)
167	replace_entry(&(data.macke.author),
168	data.gbk.reference[0].author);
169
170	if(Lenstr(data.gbk.reference[0].journal)>1)
171	replace_entry(&(data.macke.journal),
172	data.gbk.reference[0].journal);
173
174	if(Lenstr(data.gbk.reference[0].title)>1)
175	replace_entry(&(data.macke.title),
176	data.gbk.reference[0].title);
177	} /* the rest of references are put into remarks, rem:..... */
178
179	gtom_remarks();
180
181	/* adjust the strain, subspecies, and atcc information */
182	Freespace(&(data.macke.strain));
183	data.macke.strain = (char*)genbank_get_strain();
184	Freespace(&(data.macke.subspecies));
185	data.macke.subspecies = (char*)genbank_get_subspecies();
186	if(Lenstr(data.macke.atcc)<=1) {
187	Freespace(&(data.macke.atcc));
188	data.macke.atcc = (char*)genbank_get_atcc();
189	}
190
191	return(1);
192	}
193	/* --------------------------------------------------------------
194	* Function gtom_remarks().
195	* Create Macke remarks.
196	*/
197	void
198	gtom_remarks() {
199
200	int remnum, len;
201	int indi, indj;
202	char temp[LONGTEXT];
203	/* void gtom_copy_remark(); */
204
205	/* remarks in Macke format */
206	remnum = num_of_remark();
207	data.macke.remarks = (char*)calloc(1,(unsigned)(sizeof(char)*remnum));
208	remnum=0;
209
210	/* REFERENCE the first reference */
211	if(data.gbk.numofref>0)
212	gtom_copy_remark(data.gbk.reference[0].ref, "ref:", &remnum);
213
214	/* The rest of the REFERENCES */
215	for(indi=1; indi<data.gbk.numofref; indi++) {
216	gtom_copy_remark(data.gbk.reference[indi].ref, "ref:", &remnum);
217	gtom_copy_remark(data.gbk.reference[indi].author, "auth:", &remnum);
218	gtom_copy_remark(data.gbk.reference[indi].journal, "jour:", &remnum);
219	gtom_copy_remark(data.gbk.reference[indi].title, "title:", &remnum);
220	gtom_copy_remark(data.gbk.reference[indi].standard, "standard:", &remnum);
221
222	} /* loop for copying other reference */
223
224	/* copy keywords as remark */
225	gtom_copy_remark(data.gbk.keywords, "KEYWORDS:", &remnum);
226
227	/* copy accession as remark when genbank entry also exists. */
228	gtom_copy_remark(data.gbk.accession, "GenBank ACCESSION:", &remnum);
229
230	/* copy source of strain */
231	gtom_copy_remark(data.gbk.comments.orginf.source, "Source of strain:", &remnum);
232
233	/* copy former name */
234	gtom_copy_remark(data.gbk.comments.orginf.formname, "Former name:", &remnum);
235
236	/* copy alternate name */
237	gtom_copy_remark(data.gbk.comments.orginf.nickname, "Alternate name:", &remnum);
238
239	/* copy common name */
240	gtom_copy_remark(data.gbk.comments.orginf.commname, "Common name:", &remnum);
241
242	/* copy host organism */
243	gtom_copy_remark(data.gbk.comments.orginf.hostorg, "Host organism:", &remnum);
244
245	/* copy RDP ID */
246	gtom_copy_remark(data.gbk.comments.seqinf.RDPid, "RDP ID:", &remnum);
247
248	/* copy methods */
249	gtom_copy_remark(data.gbk.comments.seqinf.methods, "Sequencing methods:", &remnum);
250
251	/* copy 3' end complete */
252	if(data.gbk.comments.seqinf.comp3!=' ') {
253	if(data.gbk.comments.seqinf.comp3=='y') data.macke.remarks[remnum++] = Dupstr("3' end complete: Yes\n");
254	else data.macke.remarks[remnum++]= Dupstr("3' end complete: No\n");
255	}
256
257	/* copy 5' end complete */
258	if(data.gbk.comments.seqinf.comp5!=' ') {
259	if(data.gbk.comments.seqinf.comp5=='y') data.macke.remarks[remnum++]= Dupstr("5' end complete: Yes\n");
260	else data.macke.remarks[remnum++]= Dupstr("5' end complete: No\n");
261	}
262
263	/* other comments, not RDP DataBase specially defined */
264	if(Lenstr(data.gbk.comments.others)>0) {
265	len = Lenstr(data.gbk.comments.others);
266	for(indi=0, indj=0; indi<len; indi++)
267	{
268	temp[indj++] = data.gbk.comments.others[indi];
269	if(data.gbk.comments.others[indi]=='\n'
270	\|\| data.gbk.comments.others[indi]=='\0') {
271	temp[indj] = '\0';
272	data.macke.remarks[remnum++] =
273	(char*)Dupstr(temp);
274
275	indj=0;
276	} /* new remark line */
277	} /* for loop to find other remarks */
278	} /* other comments */
279
280	/* done with the remarks copying */
281
282	data.macke.numofrem = remnum;
283	}
284	/* --------------------------------------------------------------------
285	* Function gtom_copy_remark().
286	* If string length > 1 then copy string with key to remark.
287	*/
288	void gtom_copy_remark(string, key, remnum)
289	char *string;
290	const char *key;
291	int *remnum;
292	{
293	/* int Lenstr(); */
294	/* char Dupstr(); /
295	/* void Append(); */
296
297	/* copy host organism */
298	if(Lenstr(string)>1) {
299	data.macke.remarks[(*remnum)]
300	=(char*)Dupstr(key);
301	Append(&(data.macke.remarks[(*remnum)]), string);
302	(*remnum)++;
303	}
304	}
305	/* --------------------------------------------------------------------
306	* Function genbank_get_strain().
307	* Get strain from DEFINITION, COMMENT or SOURCE line in
308	* Genbank data file.
309	*/
310	char *genbank_get_strain() {
311
312	int indj, indk;
313	/* int find_pattern(), Lenstr(); */
314	/* int Skip_white_space(), Reach_white_space(); */
315	char strain[LONGTEXT], temp[LONGTEXT], buffer[LONGTEXT];
316	/* void get_string(), warning(), Cpystr(); */
317
318	strain[0]='\0';
319	/* get strain */
320	if(Lenstr(data.gbk.comments.others)>1) {
321	if((indj=find_pattern(data.gbk.comments.others,
322	"*source:"))>=0) {
323	if((indk=find_pattern(
324	(data.gbk.comments.others+indj),
325	"strain="))>=0) {
326	/* skip blank spaces */
327	indj=Skip_white_space(
328	data.gbk.comments.others,
329	(indj+indk+7));
330	/* get strain */
331	get_string(data.gbk.comments.others,
332	temp, indj);
333	Cpystr(strain, temp); /* copy new strain */
334	} /* get strain */
335	} /* find source: line in comment */
336	} /* look for strain on comments */
337
338	if(Lenstr(data.gbk.definition)>1) {
339	if((indj=find_pattern(data.gbk.definition, "str. "))>=0
340	\|\|(indj=find_pattern(data.gbk.definition, "strain "))>=0) {
341	/* skip the key word */
342	indj=Reach_white_space(data.gbk.definition, indj);
343	/* skip blank spaces */
344	indj=Skip_white_space(data.gbk.definition, indj);
345	/* get strain */
346	get_string(data.gbk.definition, temp, indj);
347	if(Lenstr(strain)>1) {
348	if(Cmpstr(temp, strain)!=EQ){
349	sprintf(buffer,
350	"Inconsistent strain definition in DEFINITION: %s and %s",
351	temp, strain);
352	warning(91, buffer);
353	} /* check consistency of duplicated def */
354	} else Cpystr(strain, temp); /* get strain */
355	} /* find strain in definition */
356	} /* if there is definition line */
357
358	if(Lenstr(data.gbk.source)>1) {
359	if((indj=find_pattern(data.gbk.source, "str. "))>=0
360	\|\|(indj=find_pattern(data.gbk.source, "strain "))>=0){
361	/* skip the key word */
362	indj=Reach_white_space(data.gbk.source, indj);
363	/* skip blank spaces */
364	indj=Skip_white_space(data.gbk.source, indj);
365	/* get strain */
366	get_string(data.gbk.source, temp, indj);
367	if(Lenstr(strain)>1) {
368	if(Cmpstr(temp, strain)!=EQ) {
369	sprintf(buffer,
370	"Inconsistent strain definition in SOURCE: %s and %s",
371	temp, strain);
372	warning(92, buffer);
373	}
374	} else Cpystr(strain, temp);
375	/* check consistency of duplicated def */
376	} /* find strain */
377	} /* look for strain in SOURCE line */
378
379	return(Dupstr(strain));
380	}
381	/* --------------------------------------------------------------------
382	* Function genbank_get_subspecies().
383	* Get subspecies information from SOURCE, DEFENITION, or
384	* COMMENT line of Genabnk data file.
385	*/
386	char
387	*genbank_get_subspecies() {
388
389	int indj, indk;
390	/* int find_pattern(), Lenstr(); */
391	/* int Skip_white_space(), Reach_white_space(); */
392	char subspecies[LONGTEXT], temp[LONGTEXT], buffer[LONGTEXT];
393	/* char Dupstr(); /
394	/* void get_string(), warning(), Cpystr(); */
395	/* void correct_subspecies(); */
396
397	subspecies[0]='\0';
398	/* get subspecies */
399	if(Lenstr(data.gbk.definition)>1) {
400	if((indj=find_pattern(data.gbk.definition, "subsp. "))>=0) {
401	/* skip the key word */
402	indj=Reach_white_space(data.gbk.definition, indj);
403	/* skip blank spaces */
404	indj=Skip_white_space(data.gbk.definition, indj);
405	/* get subspecies */
406	get_string(data.gbk.definition, temp, indj);
407	correct_subspecies(temp);
408	Cpystr(subspecies, temp);
409	}
410	}
411	if(Lenstr(data.gbk.comments.others)>1) {
412	if((indj=find_pattern(data.gbk.comments.others,
413	"*source:"))>=0) {
414	if((indk=find_pattern((data.gbk.comments.others+indj),
415	"sub-species="))>=0
416	\|\|(indk=find_pattern((data.gbk.comments.others+indj),
417	"subspecies="))>=0
418	\|\|(indk=find_pattern((data.gbk.comments.others+indj),
419	"subsp.="))>=0) {
420	/* skip the key word */
421	for(indj+=indk;
422	data.gbk.comments.others[indj]!='='; indj++);
423	indj++;
424	/* skip blank spaces */
425	indj=Skip_white_space(data.gbk.comments.others, indj);
426	/* get subspecies */
427	get_string(data.gbk.comments.others, temp,
428	indj);
429	if(Lenstr(subspecies)>1){
430	if(Cmpstr(temp, subspecies)!=EQ){
431	sprintf(buffer,
432	"Inconsistent subspecies definition in COMMENTS *source: %s and %s",
433	temp, subspecies);
434	warning(20, buffer);
435	}
436	} else Cpystr(subspecies, temp);
437	} /* get subspecies */
438	} /* find source: line in comment /
439	} /* look for subspecies on comments */
440
441	if(Lenstr(data.gbk.source)>1) {
442	if((indj=find_pattern(data.gbk.source, "subsp. "))>=0
443	\|\|(indj=find_pattern(data.gbk.source, "subspecies "))>=0
444	\|\|(indj=find_pattern(data.gbk.source, "sub-species "))>=0) {
445	/* skip the key word */
446	indj=Reach_white_space(data.gbk.source, indj);
447	/* skip blank spaces */
448	indj=Skip_white_space(data.gbk.source, indj);
449	/* get subspecies */
450	get_string(data.gbk.source, temp, indj);
451	correct_subspecies(temp);
452	if(Lenstr(subspecies)>1) {
453	if(Cmpstr(temp, subspecies)!=EQ){
454	sprintf(buffer,
455	"Inconsistent subspecies definition in SOURCE: %s and %s",
456	temp, subspecies);
457	warning(21, buffer);
458	}
459	} else Cpystr(subspecies, temp);
460	/* check consistency of duplicated def */
461	} /* find subspecies */
462	} /* look for subspecies in SOURCE line */
463
464	return(Dupstr(subspecies));
465	}
466	/* ---------------------------------------------------------------
467	* Function correct_subspecies().
468	* Remove the strain information in subspecies which is
469	* sometime mistakenly written into it.
470	*/
471	void
472	correct_subspecies(subspecies)
473	char *subspecies;
474	{
475	int indj;
476
477	if((indj=find_pattern(subspecies, "str\n"))>=0
478	\|\|(indj=find_pattern(subspecies, "str."))>=0
479	\|\|(indj=find_pattern(subspecies, "strain"))>=0) {
480	subspecies[indj-1]='\n';
481	subspecies[indj]='\0';
482	}
483	}
484	/* --------------------------------------------------------------------
485	* Function genbank_get_atcc().
486	* Get atcc from SOURCE line in Genbank data file.
487	*/
488	char
489	*genbank_get_atcc() {
490
491	/* int Lenstr(); */
492	char temp[LONGTEXT];
493	char *atcc;
494
495
496	atcc = NULL;
497	/* get culture collection # */
498	if(Lenstr(data.gbk.source)>1)
499	atcc = get_atcc(data.gbk.source);
500	if(Lenstr(atcc)<=1&&Lenstr(data.macke.strain)>1) {
501	/* add () to macke strain to be processed correctly */
502	sprintf(temp, "(%s)", data.macke.strain);
503	atcc = get_atcc(temp);
504	}
505	return(atcc);
506	}
507	/* ------------------------------------------------------------------- */
508	/* Function get_atcc().
509	*/
510	char
511	*get_atcc(source)
512	char *source;
513	{
514
515	static int cc_num=16;
516	static const char *CC[16] = {"ATCC", "CCM", "CDC", "CIP", "CNCTC",
517	"DSM", "EPA", "JCM", "NADC", "NCDO", "NCTC", "NRCC",
518	"NRRL", "PCC", "USDA", "VPI"};
519	/* int indk; */
520	int indi, indj, index;
521	int length;
522	/* int find_pattern(), Lenstr(); */
523	/* int paren_string(), Skip_white_space(), Reach_white_space(); */
524	char buffer[LONGTEXT], temp[LONGTEXT], pstring[LONGTEXT];
525	char atcc[LONGTEXT];
526	/* char Catstr(), Dupstr(); */
527	/* void get_atcc_string(); */
528
529	atcc[0]='\0';
530	for(indi=0; indi<cc_num; indi++) {
531	index=0;
532	while((index=paren_string(source, pstring, index))>0) {
533	if((indj=find_pattern(pstring, CC[indi]))>=0){
534	/* skip the key word */
535	indj += Lenstr(CC[indi]);
536	/* skip blank spaces */
537	indj=Skip_white_space(pstring, indj);
538	/* get strain */
539	get_atcc_string(pstring, buffer, indj);
540	sprintf(temp, "%s %s", CC[indi], buffer);
541	length=Lenstr(atcc);
542	if(length>0) {
543	atcc[length]= '\0';
544	Catstr(atcc, ", ");
545	}
546	Catstr(atcc, temp);
547	} /* find atcc */
548	} /* while loop */
549	} /* for loop */
550	/* append eoln to the atcc string */
551	length = Lenstr(atcc);
552	if (data.macke.atcc) {
553	data.macke.atcc[length] = '\0';
554	}
555	Catstr(atcc, "\n");
556	return(Dupstr(atcc));
557	}
558	/* ----------------------------------------------------------------- */
559	/* Function paren_string()
560	*/
561	int
562	paren_string(string, pstring, index)
563	char string, pstring;
564	int index;
565	{
566	int pcount=0, len, indi;
567
568	for(indi=0, len=Lenstr(string); index<len; index++) {
569	if(pcount>=1) pstring[indi++]=string[index];
570	if(string[index]=='(') pcount++;
571	if(string[index]==')') pcount--;
572	}
573	if(indi==0) return(-1);
574	pstring[--indi]='\0';
575	return(index);
576	}
577	/* ----------------------------------------------------------------
578	* Function num_of_remark().
579	* Count num of remarks needed in order to alloc spaces.
580	*/
581	int
582	num_of_remark() {
583
584	int remnum, /indi, /indj, length;
585
586	remnum = 0;
587	/* count references to be put into remarks */
588	if(data.gbk.numofref>0&&Lenstr(data.gbk.reference[0].ref)>1)
589	remnum++;
590	for(indj=1; indj<data.gbk.numofref; indj++) {
591	if(Lenstr(data.gbk.reference[indj].ref)>1)
592	remnum++;
593	if(Lenstr(data.gbk.reference[indj].journal)>1)
594	remnum++;
595	if(Lenstr(data.gbk.reference[indj].author)>1)
596	remnum++;
597	if(Lenstr(data.gbk.reference[indj].title)>1)
598	remnum++;
599	if(Lenstr(data.gbk.reference[indj].standard)>1)
600	remnum++;
601	} /* loop for copying other reference */
602	/* count the other keyword in GenBank format to be put into remarks */
603	if(Lenstr(data.gbk.keywords)>1)
604	remnum++;
605	if(Lenstr(data.gbk.accession)>1)
606	remnum++;
607	if(Lenstr(data.gbk.comments.orginf.source)>1) /* Source of strain */
608	remnum++;
609	if(Lenstr(data.gbk.comments.orginf.formname)>1)
610	remnum++;
611	if(Lenstr(data.gbk.comments.orginf.nickname)>1) /* Alternate name */
612	remnum++;
613	if(Lenstr(data.gbk.comments.orginf.commname)>1)
614	remnum++;
615	if(Lenstr(data.gbk.comments.orginf.hostorg)>1) /* host organism */
616	remnum++;
617	if(Lenstr(data.gbk.comments.seqinf.RDPid)>1)
618	remnum++;
619	if(Lenstr(data.gbk.comments.seqinf.methods)>1)
620	remnum++;
621	if(data.gbk.comments.seqinf.comp3!=' ')
622	remnum++;
623	if(data.gbk.comments.seqinf.comp5!=' ')
624	remnum++;
625	/* counting other than specific keyword comments */
626	if(Lenstr(data.gbk.comments.others)>0) {
627	length = Lenstr(data.gbk.comments.others);
628	for(indj=0; indj<length; indj++)
629	{
630	if(data.gbk.comments.others[indj]=='\n'
631	\|\| data.gbk.comments.others[indj]=='\0') {
632	remnum++;
633	} /* new remark line */
634	} /* for loop to find other remarks */
635	} /* other comments */
636	return(remnum);
637	}
638	/* -----------------------------------------------------------------
639	* Function macke_to_genbank().
640	* Convert from macke format to genbank format.
641	*/
642	void
643	macke_to_genbank(inf, outf)
644	char inf, outf;
645	{
646	FILE IFP1, IFP2, IFP3, ofp;
647	FILE_BUFFER ifp1, ifp2, ifp3;
648	char temp[TOKENNUM];
649
650	if((IFP1=fopen(inf, "r"))==NULL\|\|
651	(IFP2=fopen(inf, "r"))==NULL\|\|
652	(IFP3=fopen(inf, "r"))==NULL) {
653	sprintf(temp, "Cannot open input file %s\n", inf);
654	error(19, temp);
655	}
656
657	ifp1 = create_FILE_BUFFER(inf, IFP1);
658	ifp2 = create_FILE_BUFFER(inf, IFP2);
659	ifp3 = create_FILE_BUFFER(inf, IFP3);
660
661	if(Lenstr(outf)<=0) ofp = stdout;
662	else if((ofp=fopen(outf, "w"))==NULL) {
663	sprintf(temp, "Cannot open output file %s\n", outf);
664	error(44, temp);
665	}
666	init();
667	init_seq_data();
668	init_gm_data();
669
670	#ifdef log
671	fprintf(stderr, "Start converting...\n");
672	#endif
673
674	while(macke_in(ifp1, ifp2, ifp3)!=EOF) {
675	data.numofseq++;
676	if(mtog()) genbank_out(ofp);
677	else error(15, "Conversion from macke to genbank fails, Exit");
678	init_gm_data();
679
680	#ifdef log
681	if((data.numofseq % 50)==0)
682	fprintf(stderr,
683	"%d sequences are converted...\n",
684	data.numofseq);
685	#endif
686
687	}
688
689	#ifdef log
690	fprintf(stderr,
691	"Total %d sequences have been processed\n", data.numofseq);
692	#endif
693
694	}
695	/* ----------------------------------------------------------------
696	* Function mtog().
697	* Convert Macke format to Genbank format.
698	*/
699	int
700	mtog() {
701	int indi;
702	/* int len, Lenstr(), Cmpstr(); */
703	char temp[LONGTEXT];
704	/* char today_date(); /
705	/* char Dupstr(), Reallocspace(), macke_copyrem(); /
706	/* char genbank_date(); /
707	/* void Freespace(), Cpystr(), Append(), Append_char(); */
708	/* void mtog_genbank_def_and_source(); */
709	/* void mtog_decode_ref_and_remarks(); */
710	/* void init_reference(), replace_entry(), warning(); */
711
712	Cpystr(temp, data.macke.seqabbr);
713
714	for(indi=Lenstr(temp); indi<13; temp[indi++] = ' ') ;
715
716	if(Lenstr(data.macke.date)>1)
717
718	sprintf((temp+10),
719	"%7d bp RNA RNA %s\n",
720	data.seq_length, genbank_date(data.macke.date));
721
722	else sprintf((temp+10),
723	"%7d bp RNA RNA %s\n",
724	data.seq_length, genbank_date(today_date()));
725
726	replace_entry(&(data.gbk.locus), temp);
727
728	/* GenBank ORGANISM */
729	if(Lenstr(data.macke.name)>1) {
730	replace_entry(&(data.gbk.organism), data.macke.name);
731
732	/* append a '.' at the end */
733	Append_char(&(data.gbk.organism), '.');
734	}
735
736	if(Lenstr(data.macke.rna)>1) {
737	data.gbk.comments.seqinf.exist = 1;
738	replace_entry(&(data.gbk.comments.seqinf.methods),
739	data.macke.rna);
740	}
741	if(Lenstr(data.macke.acs)>1) {
742	/* #### not converted to accession but to comment gbkentry only, temporarily
743	Freespace(&(data.gbk.accession));
744	data.gbk.accession = Dupstr(data.macke.acs);
745	*/
746	data.gbk.comments.seqinf.exist = 1;
747	replace_entry(&(data.gbk.comments.seqinf.gbkentry),
748	data.macke.acs);
749	} else if(Lenstr(data.macke.nbk)>1) {
750	/* #### not converted to accession but to comment gbkentry only, temp
751	Freespace(&(data.gbk.accession));
752	data.gbk.accession = Dupstr(data.macke.nbk);
753	*/
754	data.gbk.comments.seqinf.exist = 1;
755	replace_entry(&(data.gbk.comments.seqinf.gbkentry),
756	data.macke.nbk);
757	}
758	if(Lenstr(data.macke.atcc)>1) {
759	data.gbk.comments.orginf.exist = 1;
760	replace_entry(&(data.gbk.comments.orginf.cc),
761	data.macke.atcc);
762	}
763	mtog_decode_ref_and_remarks();
764	/* final conversion of cc */
765	if(Lenstr(data.gbk.comments.orginf.cc)<=1&&Lenstr(data.macke.atcc)>1){
766	replace_entry(&(data.gbk.comments.orginf.cc),
767	data.macke.atcc);
768	}
769
770	/* define GenBank DEFINITION, after GenBank KEYWORD is defined. */
771	mtog_genbank_def_and_source();
772
773	return(1);
774	}
775	/* ---------------------------------------------------------------
776	* Function mtog_decode_remarks().
777	* Decode remarks of Macke to GenBank format.
778	*/
779	void
780	mtog_decode_ref_and_remarks() {
781
782	int indi, indj;
783	int acount, tcount, jcount, rcount, scount;
784	char key[TOKENNUM], temp[LONGTEXT];
785	/* char macke_copyrem(), Reallocspace(), Dupstr(); /
786	/* void Append_char(), Append(), Cpystr(); */
787	/* void mtog_copy_remark(), init_reference(); */
788
789	data.gbk.numofref=acount=tcount=jcount=rcount=scount=0;
790
791	if(Lenstr(data.macke.author)>1) {
792	if((acount+1)>data.gbk.numofref) {
793	/* new reference */
794	data.gbk.reference = (Reference)Reallocspace(data.gbk.reference, sizeof(Reference)(acount+1));
795	data.gbk.numofref = acount+1;
796	init_reference(&(data.gbk.reference[acount]), AUTHOR);
797	} else acount = data.gbk.numofref - 1;
798	data.gbk.reference[acount++].author
799	= (char*)Dupstr(data.macke.author);
800	}
801	if(Lenstr(data.macke.journal)>1) {
802	if((jcount+1)>data.gbk.numofref) {
803	data.gbk.reference =
804	(Reference*)Reallocspace(data.gbk.reference,
805	sizeof(Reference)*(jcount+1));
806	data.gbk.numofref = jcount+1;
807	init_reference(&(data.gbk.reference[jcount]), JOURNAL);
808	} else jcount = data.gbk.numofref - 1;
809	data.gbk.reference[jcount++].journal
810	= (char*)Dupstr(data.macke.journal);
811	}
812	if(Lenstr(data.macke.title)>1) {
813	if((tcount+1)>data.gbk.numofref) {
814	data.gbk.reference =
815	(Reference*)Reallocspace(data.gbk.reference,
816	sizeof(Reference)*(tcount+1));
817	data.gbk.numofref = tcount+1;
818	init_reference(&(data.gbk.reference[tcount]), TITLE);
819	} else tcount = data.gbk.numofref - 1;
820	data.gbk.reference[tcount++].title
821	= (char*)Dupstr(data.macke.title);
822	}
823	for(indi=0; indi<data.macke.numofrem; indi++) {
824	indj = macke_key_word(data.macke.remarks[indi],
825	0, key, TOKENNUM);
826	if(Cmpstr(key, "KEYWORDS")==EQ) {
827	mtog_copy_remark(&(data.gbk.keywords),
828	&indi, indj);
829
830	/* append a '.' at the end */
831	Append_char(&(data.gbk.keywords), '.');
832
833	} else if(Cmpstr(key, "GenBank ACCESSION")==EQ) {
834	mtog_copy_remark(&(data.gbk.accession),
835	&indi, indj);
836
837	} else if(Cmpstr(key, "ref")==EQ) {
838	if((rcount+1)>data.gbk.numofref) {
839	/* new reference */
840	data.gbk.reference = (Reference*)
841	Reallocspace(data.gbk.reference,
842	sizeof(Reference)*(rcount+1));
843	data.gbk.numofref = rcount+1;
844	init_reference(&(data.gbk.reference[rcount]),
845	REF);
846	} else rcount = data.gbk.numofref - 1;
847	data.gbk.reference[rcount++].ref
848	= macke_copyrem(data.macke.remarks, &indi,
849	data.macke.numofrem, indj);
850	} else if(Cmpstr(key, "auth")==EQ) {
851	if((acount+1)>data.gbk.numofref) {
852	/* new reference */
853	data.gbk.reference = (Reference*)
854	Reallocspace(data.gbk.reference,
855	sizeof(Reference)*(acount+1));
856	data.gbk.numofref = acount+1;
857	init_reference(&(data.gbk.reference[acount]),
858	AUTHOR);
859	} else acount = data.gbk.numofref - 1;
860	data.gbk.reference[acount++].author
861	= macke_copyrem(data.macke.remarks, &indi,
862	data.macke.numofrem, indj);
863	} else if(Cmpstr(key, "title")==EQ) {
864	if((tcount+1)>data.gbk.numofref) {
865	data.gbk.reference = (Reference*)
866	Reallocspace(data.gbk.reference,
867	sizeof(Reference)*(tcount+1));
868	data.gbk.numofref = tcount+1;
869	init_reference(&(data.gbk.reference[tcount]),
870	TITLE);
871	} else tcount = data.gbk.numofref - 1;
872	data.gbk.reference[tcount++].title
873	= macke_copyrem(data.macke.remarks, &indi,
874	data.macke.numofrem, indj);
875	} else if(Cmpstr(key, "jour")==EQ) {
876	if((jcount+1)>data.gbk.numofref) {
877	data.gbk.reference = (Reference*)
878	Reallocspace(data.gbk.reference,
879	sizeof(Reference)*(jcount+1));
880	data.gbk.numofref = jcount+1;
881	init_reference(&(data.gbk.reference[jcount]),
882	JOURNAL);
883	} else jcount = data.gbk.numofref - 1;
884	data.gbk.reference[jcount++].journal
885	= macke_copyrem(data.macke.remarks, &indi,
886	data.macke.numofrem, indj);
887	} else if(Cmpstr(key, "standard")==EQ) {
888	if((scount+1)>data.gbk.numofref) {
889	data.gbk.reference = (Reference*)
890	Reallocspace(data.gbk.reference,
891	sizeof(Reference)*(scount+1));
892	data.gbk.numofref = scount+1;
893	init_reference(&(data.gbk.reference[scount]),
894	STANDARD);
895	} else scount = data.gbk.numofref - 1;
896	data.gbk.reference[scount++].standard
897	= macke_copyrem(data.macke.remarks, &indi,
898	data.macke.numofrem, indj);
899
900	} else if(Cmpstr(key, "Source of strain")==EQ) {
901
902	data.gbk.comments.orginf.exist = 1;
903	mtog_copy_remark(
904	&(data.gbk.comments.orginf.source),
905	&indi, indj);
906
907	} else if(Cmpstr(key, "Former name")==EQ) {
908
909	data.gbk.comments.orginf.exist = 1;
910	mtog_copy_remark(
911	&(data.gbk.comments.orginf.formname),
912	&indi, indj);
913
914	} else if(Cmpstr(key, "Alternate name")==EQ) {
915
916	data.gbk.comments.orginf.exist = 1;
917	mtog_copy_remark(
918	&(data.gbk.comments.orginf.nickname),
919	&indi, indj);
920
921	} else if(Cmpstr(key, "Common name")==EQ) {
922
923	data.gbk.comments.orginf.exist = 1;
924	mtog_copy_remark(
925	&(data.gbk.comments.orginf.commname),
926	&indi, indj);
927
928	} else if(Cmpstr(key, "Host organism")==EQ) {
929
930	data.gbk.comments.orginf.exist = 1;
931	mtog_copy_remark(
932	&(data.gbk.comments.orginf.hostorg),
933	&indi, indj);
934
935	} else if(Cmpstr(key, "RDP ID")==EQ) {
936
937	data.gbk.comments.seqinf.exist = 1;
938	mtog_copy_remark(
939	&(data.gbk.comments.seqinf.RDPid),
940	&indi, indj);
941
942	} else if(Cmpstr(key, "Sequencing methods")==EQ) {
943
944	data.gbk.comments.seqinf.exist = 1;
945	mtog_copy_remark(
946	&(data.gbk.comments.seqinf.methods),
947	&indi, indj);
948
949	} else if(Cmpstr(key, "3' end complete")==EQ) {
950
951	data.gbk.comments.seqinf.exist = 1;
952	sscanf(data.macke.remarks[indi]+indj, "%s", key);
953	if(Cmpstr(key, "Yes")==EQ)
954	data.gbk.comments.seqinf.comp3 = 'y';
955	else data.gbk.comments.seqinf.comp3 = 'n';
956
957	} else if(Cmpstr(key, "5' end complete")==EQ) {
958
959	data.gbk.comments.seqinf.exist = 1;
960	sscanf(data.macke.remarks[indi]+indj, "%s", key);
961
962	if(Cmpstr(key, "Yes")==EQ)
963	data.gbk.comments.seqinf.comp5
964	= 'y';
965	else data.gbk.comments.seqinf.comp5
966	= 'n';
967
968	} else { /* other comments */
969	Cpystr(temp, data.macke.remarks[indi]);
970	if(data.gbk.comments.others==NULL)
971	data.gbk.comments.others
972	= (char*)Dupstr(temp);
973	else Append(&(data.gbk.comments.others), temp);
974	}
975	} /* for each rem */
976	}
977	/* ---------------------------------------------------------------
978	* Function mtog_copy_remark().
979	* Convert one remark back to GenBank format.
980	*/
981	void
982	mtog_copy_remark(string, indi, indj)
983	char **string;
984	int *indi, indj;
985	{
986	/* void Freespace(); */
987	/* char macke_copyrem(); /
988
989	Freespace(string);
990	(string) = (char)macke_copyrem(data.macke.remarks, indi,
991	data.macke.numofrem, indj);
992	}
993	/* ------------------------------------------------------------------
994	* Function macke_copyrem().
995	* uncode rem lines.
996	*/
997	char *macke_copyrem(strings, index, maxline, pointer)
998	char **strings;
999	int *index, maxline, pointer;
1000	{
1001	char *string;
1002	int indi;
1003	/* int len, length, macke_rem_continue_line(); */
1004	/* char Catstr(), Dupstr(); */
1005	/* void Append_rp_eoln(); */
1006
1007	string = (char)Dupstr(strings[(index)]+pointer);
1008	for(indi=(*index)+1; indi<maxline
1009	&&macke_rem_continue_line(strings, indi); indi++)
1010	Append_rp_eoln(&string, strings[indi]+3);
1011	(*index) = indi-1;
1012	return(string);
1013	}
1014	/* ------------------------------------------------------------------
1015	* Function mtog_genbank_def_and_source().
1016	* Define GenBank DEFINITION and SOURCE lines the way RDP
1017	* group likes.
1018	*/
1019	void
1020	mtog_genbank_def_and_source() {
1021
1022	/* int Lenstr(); */
1023	/* void Freespace(), Append(), Append_rm_eoln(), Append_char(); */
1024	/* void replace_entry(), warning(); */
1025
1026	if(Lenstr(data.macke.name)>1)
1027	replace_entry(&(data.gbk.definition),
1028	data.macke.name);
1029
1030	if(Lenstr(data.macke.subspecies)>1) {
1031
1032	if(Lenstr(data.gbk.definition)<=1) {
1033	warning(22, "Genus and Species not defined");
1034	Append_rm_eoln(&(data.gbk.definition), "subsp. ");
1035	} else Append_rm_eoln(&(data.gbk.definition), " subsp. ");
1036
1037	Append(&(data.gbk.definition), data.macke.subspecies);
1038	}
1039
1040	if(Lenstr(data.macke.strain)>1) {
1041
1042	if(Lenstr(data.gbk.definition)<=1) {
1043	warning(23,
1044	"Genus and Species and Subspecies not defined");
1045	Append_rm_eoln(&(data.gbk.definition), "str. ");
1046	} else Append_rm_eoln(&(data.gbk.definition), " str. ");
1047
1048	Append(&(data.gbk.definition), data.macke.strain);
1049	}
1050
1051	/* create SOURCE line, temp. */
1052	if(Lenstr(data.gbk.definition)>1) {
1053	replace_entry(&(data.gbk.source), data.gbk.definition);
1054	Append_char(&(data.gbk.source), '.');
1055	}
1056
1057	/* append keyword to definition, if there is keyword. */
1058	if(Lenstr(data.gbk.keywords)>1) {
1059
1060	if(Lenstr(data.gbk.definition)>1)
1061	Append_rm_eoln(&(data.gbk.definition),
1062	"; \n");
1063
1064	/* Here keywords must be ended by a '.' already */
1065	Append_rm_eoln(&(data.gbk.definition),
1066	data.gbk.keywords);
1067
1068	} else Append_rm_eoln((&data.gbk.definition), ".\n");
1069
1070	}
1071	/* -------------------------------------------------------------------
1072	* Function get_string().
1073	* Get the rest of the string untill reaching certain
1074	* terminators, such as ';', ',', '.',...
1075	* Always append "\n" at the end of the returned string.
1076	*/
1077	void
1078	get_string(line, temp, index)
1079	char line, temp;
1080	int index;
1081	{
1082	int indk, len, paren_num;
1083	/* int not_ending_mark(); */
1084
1085	len = Lenstr(line);
1086	paren_num=0;
1087	for(indk=0; index<len; index++, indk++) {
1088	temp[indk]=line[index];
1089	if(temp[indk]=='(') paren_num++;
1090	if(temp[indk]==')')
1091	if(paren_num==0) break;
1092	else paren_num--;
1093	else if(temp[indk]=='\n'\|\|(paren_num==0
1094	&&temp[indk]==';')) break;
1095	}
1096	if(indk>1 && !(not_ending_mark(temp[indk-1]))) indk--;
1097	temp[indk++]='\n';
1098	temp[indk]='\0';
1099	}
1100	/* -------------------------------------------------------------------
1101	* Function get_atcc_string().
1102	* Get the rest of the string untill reaching certain
1103	* terminators, such as ';', ',', '.',...
1104	*/
1105	void
1106	get_atcc_string(line, temp, index)
1107	char line, temp;
1108	int index;
1109	{
1110	int indk, len, paren_num;
1111
1112	len = Lenstr(line);
1113	paren_num=0;
1114	for(indk=0; index<len; index++, indk++){
1115	temp[indk]=line[index];
1116	if(temp[indk]=='(') paren_num++;
1117	if(temp[indk]==')')
1118	if(paren_num==0) break;
1119	else paren_num--;
1120	else if(paren_num==0&&(temp[indk]==';'\|\|temp[indk]=='.'
1121	\|\|temp[indk]==','\|\|temp[indk]=='/'\|\|temp[indk]=='\n'))
1122	break;
1123	}
1124	temp[indk]='\0';
1125	}

Note: See TracBrowser for help on using the browser.

Context Navigation

root/branches/stable_5.0/CONVERTALN/mg.c

Download in other formats: