Context Navigation

source: branches/stable/NTREE/NT_concatenate.cxx

Visit:

Last change on this file was 18159, checked in by westram, 5 years ago
full update from child 'fix' into 'trunk' fix item name accessors (`GBT_get_name` + `GBT_get_name_or_description`) add `null2empty` adds: log:branches/fix@18140:18158
Property svn:eol-style set to `native` Property svn:keywords set to `Author Date Id Revision`
File size: 35.2 KB

Line
1	// =======================================================================================
2	//
3	// File : NT_concatenate.cxx
4	// Purpose : 1.Concatenatenation of sequences or alignments
5	// 2.Merging the fields of similar species and creating a new species
6	// Author : Yadhu Kumar
7	// web site : http://www.arb-home.de/
8	//
9	// Copyright Department of Microbiology (Technical University Munich)
10	//
11	// =======================================================================================
12
13	#include "NT_local.h"
14
15	#include <items.h>
16	#include <item_sel_list.h>
17	#include <awt_sel_boxes.hxx>
18	#include <AW_rename.hxx>
19	#include <aw_question.hxx>
20	#include <aw_awar.hxx>
21	#include <aw_msg.hxx>
22	#include <aw_root.hxx>
23	#include <arb_progress.h>
24	#include <arb_strbuf.h>
25	#include <arb_strarray.h>
26	#include <awt_modules.hxx>
27	#include <arb_global_defs.h>
28
29	using namespace std;
30
31	#define AWAR_CON_SEQUENCE_TYPE "tmp/concat/sequence_type"
32	#define AWAR_CON_NEW_ALIGNMENT_NAME "tmp/concat/new_alignment_name"
33	#define AWAR_CON_ALIGNMENT_SEPARATOR "tmp/concat/alignment_separator"
34	#define AWAR_CON_SELECTED_ALI "tmp/concat/database_alignments"
35	#define AWAR_CON_MERGE_FIELD "tmp/concat/merge_field"
36	#define AWAR_CON_STORE_SIM_SP_NO "tmp/concat/store_sim_sp_no"
37
38	#define AWAR_CON_ALLOW_OVERWRITE_ALI "tmp/concat/overwrite"
39	#define AWAR_CON_INSGAPS_FOR_MISS_ALIS "tmp/concat/insgaps"
40
41	#define MOVE_DOWN 0
42	#define MOVE_UP 1
43
44	struct SpeciesConcatenateList {
45	GBDATA *species;
46	char *species_name;
47
48	SpeciesConcatenateList *next;
49	};
50
51	// --------------------------creating and initializing AWARS----------------------------------------
52	void NT_createConcatenationAwars(AW_root aw_root, AW_default aw_def, GBDATA gb_main) {
53	GB_transaction ta(gb_main);
54
55	char *ali_default = GBT_get_default_alignment(gb_main);
56	char *ali_type = NULp;
57
58	if (ali_default) {
59	ali_type = GBT_get_alignment_type_string(gb_main, ali_default);
60	if (!ali_type) {
61	// Note: this message will appear during startup (i.e. stick to general statement here!)
62	aw_message(GBS_global_string("Failed to detect type of default alignment (%s)\n"
63	"(Reason: %s)", ali_default, GB_await_error()));
64	}
65	}
66	if (!ali_type) ali_type = ARB_strdup("rna");
67
68	aw_root->awar_string(AWAR_CON_SEQUENCE_TYPE, ali_type, aw_def);
69	aw_root->awar_string(AWAR_CON_NEW_ALIGNMENT_NAME, "ali_concat", aw_def)->set_srt(SRT_AUTOCORRECT_ALINAME);
70	aw_root->awar_string(AWAR_CON_ALIGNMENT_SEPARATOR, "XXX", aw_def);
71	aw_root->awar_string(AWAR_CON_SELECTED_ALI, "", aw_def);
72	aw_root->awar_string(AWAR_CON_MERGE_FIELD, "full_name", aw_def);
73	aw_root->awar_string(AWAR_CON_STORE_SIM_SP_NO, "merged_species", aw_def);
74
75	aw_root->awar_int(AWAR_CON_ALLOW_OVERWRITE_ALI, 0, aw_def);
76	aw_root->awar_int(AWAR_CON_INSGAPS_FOR_MISS_ALIS, 1, aw_def);
77
78	free(ali_type);
79	free(ali_default);
80	}
81
82	// ------------------------Selecting alignments from the database for concatenation----------------------
83
84	inline char get_alitype_eval(AW_root aw_root) {
85	return GBS_global_string_copy("%s=", aw_root->awar(AWAR_CON_SEQUENCE_TYPE)->read_char_pntr());
86	}
87
88	static void alitype_changed_cb(AW_root aw_root, AW_DB_selection db_sel) {
89	char *ali_type = get_alitype_eval(aw_root);
90	awt_reconfigure_ALI_selection_list(db_sel, ali_type);
91	free(ali_type);
92	}
93
94	static AW_DB_selection* createSelectionList(GBDATA gb_main, AW_window aws, const char *awarName) {
95
96	#ifdef DEBUG
97	static bool ran=false;
98	nt_assert(!ran);
99	ran=true; // prevents calling this function for the second time
100	#endif
101
102	AW_root *aw_root = aws->get_root();
103	char *ali_type = get_alitype_eval(aw_root);
104	AW_DB_selection *db_sel = awt_create_ALI_selection_list(gb_main, aws, awarName, ali_type);
105
106	free(ali_type);
107	return db_sel;
108	}
109
110	// ---------- Create SAI to display alignments that were concatenated --------------
111
112	static GB_ERROR create_concatInfo_SAI(GBDATA gb_main, const char new_ali_name, const char *ali_separator, const StrArray& ali_names) {
113	GB_ERROR error = NULp;
114	GBDATA *gb_extended = GBT_find_or_create_SAI(gb_main, "ConcatInfo");
115
116	if (!gb_extended) error = GB_await_error();
117	else {
118	GBDATA *gb_data = GBT_add_data(gb_extended, new_ali_name, "data", GB_STRING);
119
120	if (!gb_data) {
121	error = GB_await_error();
122	}
123	else {
124	int new_ali_length = GBT_get_alignment_len(gb_main, new_ali_name);
125	int sep_len = strlen(ali_separator);
126
127	char *info = ARB_alloc<char>(new_ali_length+1);
128	memset(info, '=', new_ali_length);
129
130	int offset = 0;
131	int last_ali_idx = ali_names.size()-1;
132
133	for (int a = 0; a <= last_ali_idx; ++a) {
134	const char *ali = ali_names[a];
135	int ali_len = GBT_get_alignment_len(gb_main, ali);
136	int ali_str_len = strlen(ali);
137
138	char *my_info = info+offset;
139
140	int half_ali_len = ali_len/2;
141	for (int i = 0; i<5; ++i) {
142	if (i<half_ali_len) {
143	my_info[i] = '<';
144	my_info[ali_len-i-1] = '>';
145	}
146	}
147
148	if (ali_str_len<ali_len) {
149	int namepos = half_ali_len - ali_str_len/2;
150	memcpy(my_info+namepos, ali, ali_str_len);
151	}
152
153	offset += ali_len;
154	if (a != last_ali_idx) {
155	memcpy(info+offset, ali_separator, sep_len);
156	offset += sep_len;
157	}
158	}
159
160	nt_assert(offset == new_ali_length); // wrong alignment length!
161	info[new_ali_length] = 0;
162
163	if (!error) error = GB_write_string(gb_data, info);
164	free(info);
165	}
166	}
167	return error;
168	}
169
170	// ---------------------------------------- Concatenation function ----------------------------------
171	static void concatenateAlignments(AW_window aws, AW_selection selected_alis) {
172	nt_assert(selected_alis);
173
174	GB_push_transaction(GLOBAL.gb_main);
175
176	long marked_species = GBT_count_marked_species(GLOBAL.gb_main);
177	AW_root *aw_root = aws->get_root();
178	char *new_ali_name = aw_root->awar(AWAR_CON_NEW_ALIGNMENT_NAME)->read_string();
179	GB_ERROR error = GBT_check_alignment_name(new_ali_name);
180
181	StrArray ali_names;
182	selected_alis->get_values(ali_names);
183
184	arb_progress progress("Concatenating alignments", marked_species);
185	size_t ali_count = ali_names.size();
186
187	if (!error && ali_count<2) {
188	error = "Not enough alignments selected for concatenation (need at least 2)";
189	}
190	if (!error) {
191	int found[ali_count], missing[ali_count], ali_length[ali_count];
192
193	for (size_t a = 0; a<ali_count; a++) {
194	found[a] = 0;
195	missing[a] = 0;
196	ali_length[a] = GBT_get_alignment_len(GLOBAL.gb_main, ali_names[a]);
197
198	if (strcmp(ali_names[a], new_ali_name) == 0) {
199	error = "Target alignment may not be one of the source alignments";
200	}
201	}
202
203	if (!error) {
204	char *ali_separator = aw_root->awar(AWAR_CON_ALIGNMENT_SEPARATOR)->read_string();
205	const int sep_len = strlen(ali_separator);
206
207	long new_alignment_len = (ali_count-1)*sep_len;
208	for (size_t a = 0; a<ali_count; ++a) { // LOOP_VECTORIZED
209	new_alignment_len += ali_length[a];
210	}
211
212	GBDATA *gb_presets = GBT_get_presets(GLOBAL.gb_main);
213	GBDATA *gb_alignment_exists = GB_find_string(gb_presets, "alignment_name", new_ali_name, GB_IGNORE_CASE, SEARCH_GRANDCHILD);
214	GBDATA *gb_new_alignment = NULp;
215	char *seq_type = aw_root->awar(AWAR_CON_SEQUENCE_TYPE)->read_string();
216
217	if (gb_alignment_exists) {
218	// target alignment exists
219	if (aw_root->awar(AWAR_CON_ALLOW_OVERWRITE_ALI)->read_int()) { // allow overwrite
220	gb_new_alignment = GBT_get_alignment(GLOBAL.gb_main, new_ali_name);
221	if (!gb_new_alignment) error = GB_await_error();
222	}
223	else {
224	error = GBS_global_string("Target alignment '%s' already exists\n(check overwrite-toggle if you really want to overwrite)", new_ali_name);
225	}
226	}
227	else {
228	// create new target alignment
229	gb_new_alignment = GBT_create_alignment(GLOBAL.gb_main, new_ali_name, new_alignment_len, 0, 0, seq_type);
230	if (!gb_new_alignment) error = GB_await_error();
231	}
232
233	if (!error) {
234	AW_repeated_question ask_about_missing_alignment;
235	bool insertGaps = aw_root->awar(AWAR_CON_INSGAPS_FOR_MISS_ALIS)->read_int();
236
237	for (GBDATA *gb_species = GBT_first_marked_species(GLOBAL.gb_main);
238	gb_species && !error;
239	gb_species = GBT_next_marked_species(gb_species))
240	{
241	GBS_strstruct concat_seq(new_alignment_len+1);
242	int data_inserted = 0;
243
244	for (size_t a = 0; a<ali_count; ++a) {
245	if (a) concat_seq.cat(ali_separator);
246
247	GBDATA *gb_seq_data = GBT_find_sequence(gb_species, ali_names[a]);
248	if (gb_seq_data) { // found data
249	const char *seq_data = GB_read_char_pntr(gb_seq_data);
250	concat_seq.cat(seq_data);
251	++found[a];
252	++data_inserted;
253	}
254	else { // missing data
255	if (insertGaps) concat_seq.nput('.', ali_length[a]);
256	++missing[a];
257	}
258	}
259
260	if (!data_inserted) {
261	error = GBS_global_string("None of the source alignments had data for species '%s'", GBT_get_name_or_description(gb_species));
262	}
263	else {
264	GBDATA *gb_data = GBT_add_data(gb_species, new_ali_name, "data", GB_STRING);
265	GB_write_string(gb_data, concat_seq.get_data());
266	}
267	progress.inc_and_check_user_abort(error);
268	}
269
270	if (!error) {
271	// ............. print missing alignments...........
272	aw_message(GBS_global_string("Concatenation of alignments was performed for %ld species.", marked_species));
273	for (size_t a = 0; a<ali_count; ++a) {
274	aw_message(GBS_global_string("%s: was found in %d species and missing in %d species.", ali_names[a], found[a], missing[a]));
275	}
276	}
277
278	if (!error) error = GBT_check_data(GLOBAL.gb_main, new_ali_name); // update alignment info (otherwise create_concatInfo_SAI fails when overwriting an alignment)
279	if (!error) error = create_concatInfo_SAI(GLOBAL.gb_main, new_ali_name, ali_separator, ali_names);
280	}
281
282	free(seq_type);
283	free(ali_separator);
284	}
285	}
286
287	if (!error) {
288	char *nfield = GBS_global_string_copy("%s/data", new_ali_name);
289	error = GBT_add_new_changekey(GLOBAL.gb_main, nfield, GB_STRING);
290	free(nfield);
291	}
292	else {
293	progress.done();
294	}
295	GB_end_transaction_show_error(GLOBAL.gb_main, error, aw_message);
296	free(new_ali_name);
297	}
298
299	static void addSpeciesToConcatenateList(SpeciesConcatenateList **sclp, GB_CSTR species_name) {
300
301	GBDATA *gb_species_data = GBT_get_species_data(GLOBAL.gb_main);
302	GBDATA *gb_species = GBT_find_species_rel_species_data(gb_species_data, species_name);
303
304	if (gb_species) {
305	SpeciesConcatenateList *scl = new SpeciesConcatenateList;
306
307	scl->species = gb_species;
308	scl->species_name = ARB_strdup(species_name);
309	scl->next = *sclp;
310	*sclp = scl;
311	}
312	}
313
314	static void freeSpeciesConcatenateList(SpeciesConcatenateList *scl) {
315	while (scl) {
316	SpeciesConcatenateList *next = scl->next;
317	free(scl->species_name);
318	delete scl;
319	scl = next;
320	}
321	}
322
323	static GB_ERROR checkAndMergeFields(GBDATA gb_new_species, GB_ERROR error, SpeciesConcatenateList scl) {
324
325	char *doneFields = ARB_strdup(";name;"); // all fields which are already merged
326	int doneLen = strlen(doneFields);
327	SpeciesConcatenateList *sl = scl;
328	int sl_length = 0; while (scl) { sl_length++; scl=scl->next; } // counting no. of similar species stored in the list
329	int *fieldStat = new int[sl_length]; // 0 = not used yet ; -1 = doesn't have field ; 1..n = field content (same number means same content)
330
331	while (sl && !error) { // with all species do..
332	char *newFields = GB_get_subfields(sl->species);
333	char *fieldStart = newFields; // points to ; before next field
334
335	while (fieldStart[1] && !error) { // with all subfields of the species do..
336	char *fieldEnd = strchr(fieldStart+1, ';');
337	nt_assert(fieldEnd);
338	char behind = fieldEnd[1]; fieldEnd[1] = 0;
339
340	if (!strstr(doneFields, fieldStart)) { // field is not merged yet
341	char *fieldName = fieldStart+1;
342	int fieldLen = int(fieldEnd-fieldName);
343
344	nt_assert(fieldEnd[0]==';');
345	fieldEnd[0] = 0;
346
347	GBDATA *gb_field = GB_search(sl->species, fieldName, GB_FIND); // field does to exist (it was found before)
348	GB_TYPES type = GB_read_type(gb_field);
349
350	if (type==GB_STRING) { // we only merge string fields
351	int i; int doneSpecies = 0; int nextStat = 1;
352
353	for (i=0; i<sl_length; i++) { fieldStat[i] = 0; } // clear field status
354
355	while (doneSpecies<sl_length) { // since all species in list were handled
356	SpeciesConcatenateList *sl2 = sl;
357	i = 0;
358
359	while (sl2) {
360	if (fieldStat[i]==0) {
361	gb_field = GB_search(sl2->species, fieldName, GB_FIND);
362	if (gb_field) {
363	char *content = GB_read_as_string(gb_field);
364	SpeciesConcatenateList *sl3 = sl2->next;
365	fieldStat[i] = nextStat;
366	int j = i+1; doneSpecies++;
367
368	while (sl3) {
369	if (fieldStat[j]==0) {
370	gb_field = GB_search(sl3->species, fieldName, GB_FIND);
371	if (gb_field) {
372	char *content2 = GB_read_as_string(gb_field);
373	if (strcmp(content, content2)==0) { // if contents are the same, they get the same status
374	fieldStat[j] = nextStat;
375	doneSpecies++;
376	}
377	free(content2);
378	}
379	else {
380	fieldStat[j] = -1;
381	doneSpecies++;
382	}
383	}
384	sl3 = sl3->next; j++;
385	}
386	free(content); nextStat++;
387	}
388	else {
389	fieldStat[i] = -1; // field does not exist here
390	doneSpecies++;
391	}
392	}
393	sl2 = sl2->next; i++;
394	}
395	if (!sl2) break;
396	}
397	nt_assert(nextStat!=1); // this would mean that none of the species contained the field
398	{
399	char *new_content = NULp;
400	int new_content_len = 0; // @@@ useless (0 where used; unused otherwise)
401
402	if (nextStat==2) { // all species contain same field content or do not have the field
403	SpeciesConcatenateList *sl2 = sl;
404	while (sl2) {
405	gb_field = GB_search(sl2->species, fieldName, GB_FIND);
406	if (gb_field) {
407	new_content = GB_read_as_string(gb_field);
408	new_content_len = strlen(new_content);
409	break;
410	}
411	sl2 = sl2->next;
412	}
413	}
414	else { // different field contents
415	int actualStat;
416	for (actualStat=1; actualStat<nextStat; actualStat++) {
417	SpeciesConcatenateList *sl2 = sl;
418
419	int names_len = 1; // open bracket
420	char *content = NULp;
421	i = 0;
422
423	while (sl2) {
424	if (fieldStat[i]==actualStat) {
425	names_len += strlen(sl2->species_name)+1;
426	if (!content) {
427	gb_field = GB_search(sl2->species, fieldName, GB_FIND);
428	nt_assert(gb_field);
429	content = GB_read_as_string(gb_field);
430	}
431	}
432	sl2 = sl2->next; i++;
433	}
434	nt_assert(content);
435	int add_len = names_len+1+strlen(content);
436	char *whole = ARB_alloc<char>(new_content_len+1+add_len+1);
437	nt_assert(whole);
438	char *add = new_content ? whole+sprintf(whole, "%s ", new_content) : whole;
439	sl2 = sl; i = 0;
440	int first = 1;
441	while (sl2) {
442	if (fieldStat[i]==actualStat) {
443	add += sprintf(add, "%c%s", first ? '{' : ';', sl2->species_name);
444	first = 0;
445	}
446	sl2 = sl2->next; i++;
447	}
448	add += sprintf(add, "} %s", content);
449
450	free(content);
451	freeset(new_content, whole);
452	new_content_len = strlen(new_content); // cppcheck-suppress deallocuse
453	}
454	}
455
456	if (new_content) {
457	error = GBT_write_string(gb_new_species, fieldName, new_content);
458	free(new_content);
459	}
460	}
461	}
462
463	// mark field as done:
464	char *new_doneFields = ARB_alloc<char>(doneLen+fieldLen+1+1);
465	sprintf(new_doneFields, "%s%s;", doneFields, fieldName);
466	doneLen += fieldLen+1;
467	freeset(doneFields, new_doneFields);
468	fieldEnd[0] = ';';
469	}
470	fieldEnd[1] = behind;
471	fieldStart = fieldEnd;
472	}
473	free(newFields);
474	sl = sl->next;
475	}
476	free(doneFields);
477	delete [] fieldStat;
478
479	return error;
480	}
481
482	static GBDATA concatenateFieldsCreateNewSpecies(AW_window , GBDATA gb_species, SpeciesConcatenateList scl) {
483	GB_push_transaction(GLOBAL.gb_main);
484
485	GB_ERROR error = NULp;
486	GBDATA *gb_species_data = GBT_get_species_data(GLOBAL.gb_main);
487
488	// data needed for name generation
489	char *full_name = NULp;
490	char *acc = NULp;
491
492	// --------------------getting the species related data --------------------
493
494	GBDATA *gb_new_species = NULp;
495
496	if (!error) {
497	// copy species to create a new species
498	gb_new_species = GB_create_container(gb_species_data, "species");
499	error = gb_new_species ? GB_copy_dropProtectMarksAndTempstate(gb_new_species, gb_species) : GB_await_error();
500
501	if (!error) { // write dummy-name (real name written below)
502	error = GBT_write_string(gb_new_species, "name", "$currcat$");
503	}
504	}
505
506	if (!error) { // copy full name
507	full_name = GBT_read_string(gb_species, "full_name");
508	if (!full_name) error = GB_await_error();
509	else error = GBT_write_string(gb_new_species, "full_name", full_name);
510	}
511
512	if (!error) {
513	ConstStrArray ali_names;
514	GBT_get_alignment_names(ali_names, GLOBAL.gb_main);
515
516	long id = 0;
517	for (SpeciesConcatenateList *speciesList = scl; speciesList; speciesList = speciesList->next) {
518	for (int no_of_alignments = 0; ali_names[no_of_alignments]; no_of_alignments++) {
519	GBDATA *gb_seq_data = GBT_find_sequence(speciesList->species, ali_names[no_of_alignments]);
520	if (gb_seq_data) {
521	const char *seq_data = GB_read_char_pntr(gb_seq_data);
522	GBDATA *gb_data = GBT_add_data(gb_new_species, ali_names[no_of_alignments], "data", GB_STRING);
523	error = GB_write_string(gb_data, seq_data);
524	if (!error) id += GBS_checksum(seq_data, 1, ".-"); // creating checksum of the each aligned sequence to generate new accession number
525	}
526	if (error) error = GB_export_errorf("Can't create alignment '%s'", ali_names[no_of_alignments]);
527	}
528	}
529
530	if (!error) {
531	acc = GBS_global_string_copy("ARB_%lX", id); // create new accession number
532	error = GBT_write_string(gb_new_species, "acc", acc);
533	}
534	}
535
536	if (!error) error = checkAndMergeFields(gb_new_species, error, scl);
537
538	// now generate new name
539	if (!error) {
540	char *new_species_name = NULp;
541
542	const char *add_field = AW_get_nameserver_addid(GLOBAL.gb_main);
543	GBDATA *gb_addid = add_field[0] ? GB_entry(gb_new_species, add_field) : NULp;
544	char *addid = NULp;
545	if (gb_addid) addid = GB_read_as_string(gb_addid);
546
547	error = AWTC_generate_one_name(GLOBAL.gb_main, full_name, acc, addid, new_species_name);
548	if (!error) { // name was created
549	if (GBT_find_species_rel_species_data(gb_species_data, new_species_name)) {
550	// if the name is not unique -> create unique name
551	UniqueNameDetector und(gb_species_data);
552	freeset(new_species_name, AWTC_makeUniqueShortName(new_species_name, und));
553	if (!new_species_name) error = GB_await_error();
554	}
555	}
556
557	if (!error) error = GBT_write_string(gb_new_species, "name", new_species_name); // insert new 'name'
558
559	free(new_species_name);
560	free(addid);
561	}
562
563	error = GB_end_transaction(GLOBAL.gb_main, error);
564	if (error) {
565	gb_new_species = NULp;
566	aw_message(error);
567	}
568
569	free(acc);
570	free(full_name);
571
572	return gb_new_species;
573	}
574
575	enum MergeSpeciesType {
576	MERGE_SPECIES_SIMPLE,
577	MERGE_SPECIES_AND_CONCAT_ALI,
578	};
579
580	static void mergeSimilarSpecies(AW_window aws, MergeSpeciesType mergeType, AW_selection selected_alis) {
581	nt_assert(correlated(selected_alis, mergeType == MERGE_SPECIES_AND_CONCAT_ALI));
582
583	GB_ERROR error = NULp;
584	arb_progress wrapper;
585	{
586	AW_root *aw_root = aws->get_root();
587	char *merge_field_name = aw_root->awar(AWAR_CON_MERGE_FIELD)->read_string();
588
589	SpeciesConcatenateList *scl = NULp; // to build list of similar species
590	SpeciesConcatenateList *newSpeciesList = NULp; // new SpeciesConcatenateList
591
592	GB_begin_transaction(GLOBAL.gb_main); // open database for transaction
593
594	const char *report_field_name = prepare_and_get_selected_itemfield(aw_root, AWAR_CON_STORE_SIM_SP_NO, GLOBAL.gb_main, SPECIES_get_selector(), FIF_NAME_SELECTED);
595	error = GB_incur_error_if(!report_field_name);
596
597	if (!error && strcmp(merge_field_name, NO_FIELD_SELECTED) == 0) {
598	error = "Please select database field for similarity detection";
599	}
600
601	if (!error) {
602	PersistentNameServerConnection stayAlive;
603	arb_progress progress("Merging similar species", GBT_count_marked_species(GLOBAL.gb_main));
604	progress.auto_subtitles("Species");
605
606	for (GBDATA * gb_species = GBT_first_marked_species(GLOBAL.gb_main);
607	gb_species && !error;
608	gb_species = GBT_next_marked_species(gb_species))
609	{
610	GBDATA *gb_species_field = GB_entry(gb_species, merge_field_name);
611
612	if (!gb_species_field) {
613	// exit if species doesn't have any data in the selected field
614	error = GBS_global_string("Species '%s' does not contain data in selected field '%s'",
615	GBT_get_name_or_description(gb_species),
616	merge_field_name);
617	}
618	else {
619	char *gb_species_field_content = GB_read_as_string(gb_species_field);
620	int similar_species = 0;
621
622	for (GBDATA * gb_species_next = GBT_next_marked_species(gb_species);
623	gb_species_next && !error;
624	gb_species_next = GBT_next_marked_species(gb_species_next))
625	{
626	GBDATA *gb_next_species_field = GB_entry(gb_species_next, merge_field_name);
627	if (!gb_next_species_field) {
628	// exit if species doesn't have any data in the selected field
629	error = GBS_global_string("Species '%s' does not contain data in selected field '%s'",
630	GBT_get_name_or_description(gb_species_next),
631	merge_field_name);
632	}
633	else {
634	char *gb_next_species_field_content = GB_read_as_string(gb_next_species_field);
635
636	if (strcmp(gb_species_field_content, gb_next_species_field_content) == 0) {
637	addSpeciesToConcatenateList(&scl, GBT_get_name_or_description(gb_species_next));
638	GB_write_flag(gb_species_next, 0);
639	++similar_species;
640	++progress;
641	}
642	free(gb_next_species_field_content);
643	}
644	}
645
646	if (similar_species > 0 && !error) {
647	++similar_species; // correct merge counter
648	addSpeciesToConcatenateList(&scl, GBT_get_name_or_description(gb_species));
649	GB_write_flag(gb_species, 0);
650
651	GBDATA *new_species_created = concatenateFieldsCreateNewSpecies(aws, gb_species, scl);
652
653	nt_assert(new_species_created);
654	if (new_species_created) { // create a list of newly created species
655	addSpeciesToConcatenateList(&newSpeciesList, GBT_get_name_or_description(new_species_created));
656	}
657
658	if (report_field_name) {
659	GBDATA *gb_report = GBT_searchOrCreate_itemfield_according_to_changekey(new_species_created, report_field_name, SPECIES_get_selector().change_key_path);
660	if (!gb_report) error = GB_await_error();
661	else error = GB_write_lossless_int(gb_report, similar_species);
662	}
663	}
664
665	freeSpeciesConcatenateList(scl); scl = NULp;
666	free(gb_species_field_content);
667	}
668
669	progress.inc_and_check_user_abort(error);
670	}
671	}
672
673	if (!error) {
674	GBT_mark_all(GLOBAL.gb_main, 0); // unmark all species in the database
675	int newSpeciesCount = 0;
676
677	for (; newSpeciesList; newSpeciesList = newSpeciesList->next) { // mark only newly created species
678	GB_write_flag(newSpeciesList->species, 1);
679	newSpeciesCount++;
680	}
681	aw_message(GBS_global_string("%i new species were created by taking \"%s\" as a criterion!", newSpeciesCount, merge_field_name));
682	freeSpeciesConcatenateList(newSpeciesList);
683	}
684
685	free(merge_field_name);
686
687	GB_end_transaction_show_error(GLOBAL.gb_main, error, aw_message);
688	}
689
690	if (mergeType == MERGE_SPECIES_AND_CONCAT_ALI && !error) {
691	// @@@ what happens if merge-process above succeeds and concatenateAlignments below fails?
692	// @@@ i think both steps should be put into ONE transaction!
693	concatenateAlignments(aws, selected_alis);
694	}
695	}
696
697	static AW_window createMergeSimilarSpeciesWindow(AW_root aw_root, MergeSpeciesType mergeType, AW_selection *selected_alis) {
698	AW_window_simple *aws = new AW_window_simple;
699
700	{
701	char *window_id = GBS_global_string_copy("MERGE_SPECIES_%i", mergeType);
702	const char *window_title = NULp;
703	switch (mergeType) {
704	case MERGE_SPECIES_SIMPLE: window_title = "Merge species"; break;
705	case MERGE_SPECIES_AND_CONCAT_ALI: window_title = "Merge and concatenate"; break;
706	}
707	aws->init(aw_root, window_id, window_title);
708	free(window_id);
709	}
710	aws->load_xfig("merge_species.fig");
711
712	aws->callback(makeHelpCallback("merge_species.hlp"));
713	aws->at("help");
714	aws->create_button("HELP", "HELP", "H");
715
716	create_itemfield_selection_button(aws, FieldSelDef(AWAR_CON_MERGE_FIELD, GLOBAL.gb_main, SPECIES_get_selector(), FIELD_FILTER_STRING_READABLE, "field to compare"), "field_select");
717	create_itemfield_selection_button(aws, FieldSelDef(AWAR_CON_STORE_SIM_SP_NO, GLOBAL.gb_main, SPECIES_get_selector(), FIELD_FILTER_INT_WRITEABLE, "report-field", SF_ALLOW_NEW), "store_sp_no");
718
719	{
720	const char *buttonText = NULp;
721	switch (mergeType) {
722	case MERGE_SPECIES_SIMPLE: buttonText = "Merge similar species"; break;
723	case MERGE_SPECIES_AND_CONCAT_ALI: buttonText = "Merge similar species and concat alignments"; break;
724	}
725
726	aws->at("merge");
727	aws->callback(makeWindowCallback(mergeSimilarSpecies, mergeType, selected_alis));
728	aws->create_autosize_button("MERGE_SIMILAR_SPECIES", buttonText, "M");
729	}
730
731	aws->at("close");
732	aws->callback(AW_POPDOWN);
733	aws->create_button("CLOSE", "CLOSE", "C");
734
735	return aws;
736	}
737
738	AW_window NT_createMergeSimilarSpeciesWindow(AW_root aw_root) {
739	static AW_window *aw = NULp;
740	if (!aw) aw = createMergeSimilarSpeciesWindow(aw_root, MERGE_SPECIES_SIMPLE, NULp);
741	return aw;
742	}
743
744	static AW_window NT_createMergeSimilarSpeciesAndConcatenateWindow(AW_root aw_root, AW_selection *selected_alis) {
745	static AW_window *aw = NULp;
746	#if defined(ASSERTION_USED)
747	static AW_selection *prev_selected_alis = NULp;
748	#endif
749
750	if (!aw) {
751	aw = createMergeSimilarSpeciesWindow(aw_root, MERGE_SPECIES_AND_CONCAT_ALI, selected_alis);
752	#if defined(ASSERTION_USED)
753	prev_selected_alis = selected_alis;
754	#endif
755	}
756	#if defined(ASSERTION_USED)
757	nt_assert(selected_alis == prev_selected_alis); // would need multiple windows in that case
758	#endif
759	return aw;
760	}
761
762	static void useSelectedAlignment(AW_window *aww) {
763	AW_root *root = aww->get_root();
764	const char *selali = root->awar(AWAR_CON_SELECTED_ALI)->read_char_pntr();
765	if (selali && strcmp(selali, NO_ALI_SELECTED) != 0) {
766	root->awar(AWAR_CON_NEW_ALIGNMENT_NAME)->write_string(selali);
767	}
768	else {
769	aw_message("Select alignment to use in the left alignment list");
770	}
771	}
772
773	// ----------------------------Creating concatenation window-----------------------------------------
774	AW_window NT_createConcatenationWindow(AW_root aw_root) {
775	AW_window_simple *aws = new AW_window_simple;
776
777	aws->init(aw_root, "CONCAT_ALIGNMENTS", "Concatenate Alignments");
778	aws->load_xfig("concatenate.fig");
779
780	aws->auto_space(5, 5);
781	aws->button_length(8);
782
783	aws->callback(makeHelpCallback("concatenate.hlp"));
784	aws->at("help");
785	aws->create_button("HELP", "HELP", "H");
786
787	aws->at("close");
788	aws->callback(AW_POPDOWN);
789	aws->create_button("CLOSE", "CLOSE", "C");
790
791	aws->at("dbAligns");
792	AW_DB_selection *all_alis = createSelectionList(GLOBAL.gb_main, aws, AWAR_CON_SELECTED_ALI);
793	AW_selection *sel_alis = awt_create_subset_selection_list(aws, all_alis->get_sellist(), "concatAligns", "collect", "sort");
794
795	aws->at("type");
796	aws->create_option_menu(AWAR_CON_SEQUENCE_TYPE, true);
797	aws->insert_option("DNA", "d", "dna");
798	aws->insert_option("RNA", "r", "rna");
799	aws->insert_default_option("PROTEIN", "p", "ami");
800	aws->update_option_menu();
801	aw_root->awar(AWAR_CON_SEQUENCE_TYPE)->add_callback(makeRootCallback(alitype_changed_cb, all_alis));
802
803	aws->at("aliSeparator");
804	aws->create_input_field(AWAR_CON_ALIGNMENT_SEPARATOR, 10);
805
806	aws->at("aliName");
807	aws->create_input_field(AWAR_CON_NEW_ALIGNMENT_NAME, 25);
808	aws->button_length(5);
809	aws->callback(useSelectedAlignment);
810	aws->create_button("USE", "Use");
811
812	aws->at("overwrite");
813	aws->label("Allow to overwrite an existing alignment?");
814	aws->create_toggle(AWAR_CON_ALLOW_OVERWRITE_ALI);
815
816	aws->at("insgaps");
817	aws->label("Insert gaps for missing alignment data?");
818	aws->create_toggle(AWAR_CON_INSGAPS_FOR_MISS_ALIS);
819
820	aws->button_length(22);
821	aws->at("go");
822
823	aws->callback(makeWindowCallback(concatenateAlignments, sel_alis));
824	aws->create_button("CONCATENATE", "CONCATENATE", "A");
825
826	aws->callback(NT_createMergeSimilarSpeciesWindow);
827	aws->create_button("MERGE_SPECIES", "MERGE SIMILAR SPECIES", "M");
828
829	aws->callback(makeCreateWindowCallback(NT_createMergeSimilarSpeciesAndConcatenateWindow, sel_alis));
830	aws->create_button("MERGE_CONCATENATE", "MERGE & CONCATENATE", "S");
831
832	return aws;
833	}
834	// -------------------------------------------------------------------------------------------------------

Note: See TracBrowser for help on using the repository browser.

Download in other formats: