Context Navigation

source: tags/ms_r16q2/DIST/DI_matr.cxx

Visit:

Last change on this file was 14686, checked in by westram, 8 years ago
change datatypes used in awar-access (only changed simple cases)
Property svn:eol-style set to `native` Property svn:keywords set to `Author Date Id Revision`
File size: 65.4 KB

Line
1	// =============================================================== //
2	// //
3	// File : DI_matr.cxx //
4	// Purpose : //
5	// //
6	// Institute of Microbiology (Technical University Munich) //
7	// http://www.arb-home.de/ //
8	// //
9	// =============================================================== //
10
11	#include "di_protdist.hxx"
12	#include "di_clusters.hxx"
13	#include "dist.hxx"
14	#include "di_view_matrix.hxx"
15	#include "di_awars.hxx"
16
17	#include <neighbourjoin.hxx>
18	#include <AP_filter.hxx>
19	#include <CT_ctree.hxx>
20	#include <ColumnStat.hxx>
21
22	#include <awt.hxx>
23	#include <awt_sel_boxes.hxx>
24	#include <awt_filter.hxx>
25
26	#include <aw_preset.hxx>
27	#include <aw_awars.hxx>
28	#include <aw_file.hxx>
29	#include <aw_msg.hxx>
30	#include <aw_root.hxx>
31
32	#include <gui_aliview.hxx>
33
34	#include <climits>
35	#include <ctime>
36	#include <cmath>
37	#include <arb_sort.h>
38	#include <arb_global_defs.h>
39	#include <macros.hxx>
40	#include <ad_cb.h>
41	#include <awt_TreeAwars.hxx>
42	#include <arb_defs.h>
43
44	using std::string;
45
46	// --------------------------------------------------------------------------------
47
48	#define AWAR_DIST_BOOTSTRAP_COUNT AWAR_DIST_PREFIX "bootstrap/count"
49	#define AWAR_DIST_CANCEL_CHARS AWAR_DIST_PREFIX "cancel/chars"
50	#define AWAR_DIST_MATRIX_AUTO_RECALC AWAR_DIST_PREFIX "recalc"
51
52	#define AWAR_DIST_COLUMN_STAT_NAME AWAR_DIST_COLUMN_STAT_PREFIX "name"
53
54	#define AWAR_DIST_TREE_SORT_NAME "tmp/" AWAR_DIST_TREE_PREFIX "sort_tree_name"
55	#define AWAR_DIST_TREE_COMP_NAME "tmp/" AWAR_DIST_TREE_PREFIX "compr_tree_name"
56	#define AWAR_DIST_TREE_STD_NAME AWAR_DIST_TREE_PREFIX "tree_name"
57	#define AWAR_DIST_MATRIX_AUTO_CALC_TREE AWAR_DIST_TREE_PREFIX "autocalc"
58
59	#define AWAR_DIST_SAVE_MATRIX_TYPE AWAR_DIST_SAVE_MATRIX_BASE "/type"
60	#define AWAR_DIST_SAVE_MATRIX_FILENAME AWAR_DIST_SAVE_MATRIX_BASE "/file_name"
61
62	// --------------------------------------------------------------------------------
63
64	DI_GLOBAL_MATRIX GLOBAL_MATRIX;
65
66	static MatrixDisplay *matrixDisplay = 0;
67	static AP_userdef_matrix userdef_DNA_matrix(AP_MAX, AWAR_DIST_MATRIX_DNA_BASE);
68
69	static AP_matrix *get_user_matrix() {
70	AW_root *awr = AW_root::SINGLETON;
71	if (!awr->awar(AWAR_DIST_MATRIX_DNA_ENABLED)->read_int()) {
72	return NULL;
73	}
74	userdef_DNA_matrix.update_from_awars(awr);
75	userdef_DNA_matrix.normize();
76	return &userdef_DNA_matrix;
77	}
78
79	class BoundWindowCallback : virtual Noncopyable {
80	AW_window *aww;
81	WindowCallback cb;
82	public:
83	BoundWindowCallback(AW_window *aww_, const WindowCallback& cb_)
84	: aww(aww_),
85	cb(cb_)
86	{}
87	void operator()() { cb(aww); }
88	};
89
90	static SmartPtr<BoundWindowCallback> recalculate_matrix_cb;
91	static SmartPtr<BoundWindowCallback> recalculate_tree_cb;
92
93	static GB_ERROR last_matrix_calculation_error = NULL;
94
95	static void matrix_changed_cb() {
96	if (matrixDisplay) {
97	matrixDisplay->mark(MatrixDisplay::NEED_SETUP);
98	matrixDisplay->update_display();
99	}
100	}
101
102	struct RecalcNeeded {
103	bool matrix;
104	bool tree;
105
106	RecalcNeeded() : matrix(false), tree(false) { }
107	};
108
109	static RecalcNeeded need_recalc;
110
111	CONSTEXPR unsigned UPDATE_DELAY = 200;
112
113	static unsigned update_cb(AW_root *aw_root);
114	inline void add_update_cb() {
115	AW_root::SINGLETON->add_timed_callback(UPDATE_DELAY, makeTimedCallback(update_cb));
116	}
117
118	inline void matrix_needs_recalc_cb() {
119	need_recalc.matrix = true;
120	add_update_cb();
121	}
122	inline void tree_needs_recalc_cb() {
123	need_recalc.tree = true;
124	add_update_cb();
125	}
126
127	inline void compressed_matrix_needs_recalc_cb() {
128	if (GLOBAL_MATRIX.has_type(DI_MATRIX_COMPRESSED)) {
129	matrix_needs_recalc_cb();
130	}
131	}
132
133	static unsigned update_cb(AW_root *aw_root) {
134	if (need_recalc.matrix) {
135	GLOBAL_MATRIX.forget();
136	need_recalc.matrix = false; // because it's forgotten
137
138	int matrix_autocalc = aw_root->awar(AWAR_DIST_MATRIX_AUTO_RECALC)->read_int();
139	if (matrix_autocalc) {
140	bool recalc_now = true;
141	int tree_autocalc = aw_root->awar(AWAR_DIST_MATRIX_AUTO_CALC_TREE)->read_int();
142	if (!tree_autocalc) recalc_now = matrixDisplay ? matrixDisplay->willShow() : false;
143
144	if (recalc_now) {
145	di_assert(recalculate_matrix_cb.isSet());
146	(*recalculate_matrix_cb)();
147	di_assert(need_recalc.tree == true);
148	}
149	}
150	di_assert(need_recalc.matrix == false);
151	}
152
153	if (need_recalc.tree) {
154	int tree_autocalc = aw_root->awar(AWAR_DIST_MATRIX_AUTO_CALC_TREE)->read_int();
155	if (tree_autocalc) {
156	di_assert(recalculate_tree_cb.isSet());
157	(*recalculate_tree_cb)();
158	need_recalc.matrix = false; // otherwise endless loop, e.g. if output-tree is used for sorting
159	}
160	}
161
162	return 0; // do not call again
163	}
164
165	static void auto_calc_changed_cb(AW_root *aw_root) {
166	int matrix_autocalc = aw_root->awar(AWAR_DIST_MATRIX_AUTO_RECALC)->read_int();
167	int tree_autocalc = aw_root->awar(AWAR_DIST_MATRIX_AUTO_CALC_TREE)->read_int();
168
169	if (matrix_autocalc && !GLOBAL_MATRIX.exists()) matrix_needs_recalc_cb();
170	if (tree_autocalc && (matrix_autocalc \|\| GLOBAL_MATRIX.exists())) tree_needs_recalc_cb();
171	}
172
173	static AW_window create_dna_matrix_window(AW_root aw_root) {
174	AW_window_simple *aws = new AW_window_simple;
175	aws->init(aw_root, "SET_DNA_MATRIX", "SET MATRIX");
176	aws->auto_increment(50, 50);
177	aws->button_length(10);
178	aws->callback(AW_POPDOWN);
179	aws->create_button("CLOSE", "CLOSE");
180
181	aws->callback(makeHelpCallback("user_matrix.hlp"));
182	aws->create_button("HELP", "HELP");
183
184	aws->at_newline();
185
186	userdef_DNA_matrix.create_input_fields(aws);
187	aws->window_fit();
188	return aws;
189	}
190
191	static void selected_tree_changed_cb() {
192	if (AW_root::SINGLETON->awar(AWAR_DIST_CORR_TRANS)->read_int() == DI_TRANSFORMATION_FROM_TREE) {
193	matrix_needs_recalc_cb();
194	}
195	}
196
197	void DI_create_matrix_variables(AW_root *aw_root, AW_default def, AW_default db) {
198	GB_transaction ta(db);
199
200	userdef_DNA_matrix.set_descriptions(AP_A, "A");
201	userdef_DNA_matrix.set_descriptions(AP_C, "C");
202	userdef_DNA_matrix.set_descriptions(AP_G, "G");
203	userdef_DNA_matrix.set_descriptions(AP_T, "TU");
204	userdef_DNA_matrix.set_descriptions(AP_GAP, "GAP");
205
206	userdef_DNA_matrix.create_awars(aw_root);
207
208	RootCallback matrix_needs_recalc_callback = makeRootCallback(matrix_needs_recalc_cb);
209	aw_root->awar_int(AWAR_DIST_MATRIX_DNA_ENABLED, 0)->add_callback(matrix_needs_recalc_callback); // user matrix disabled by default
210	{
211	GBDATA *gbd = GB_search(AW_ROOT_DEFAULT, AWAR_DIST_MATRIX_DNA_BASE, GB_FIND);
212	GB_add_callback(gbd, GB_CB_CHANGED, makeDatabaseCallback(matrix_needs_recalc_cb));
213	}
214
215	aw_root->awar_string(AWAR_DIST_WHICH_SPECIES, "marked", def)->add_callback(matrix_needs_recalc_callback);
216	{
217	char *default_ali = GBT_get_default_alignment(db);
218	aw_root->awar_string(AWAR_DIST_ALIGNMENT, "", def)->add_callback(matrix_needs_recalc_callback)->write_string(default_ali);
219	free(default_ali);
220	}
221	aw_root->awar_string(AWAR_DIST_FILTER_ALIGNMENT, "none", def);
222	aw_root->awar_string(AWAR_DIST_FILTER_NAME, "none", def);
223	aw_root->awar_string(AWAR_DIST_FILTER_FILTER, "", def)->add_callback(matrix_needs_recalc_callback);
224	aw_root->awar_int (AWAR_DIST_FILTER_SIMPLIFY, 0, def)->add_callback(matrix_needs_recalc_callback);
225
226	aw_root->awar_string(AWAR_DIST_CANCEL_CHARS, ".", def)->add_callback(matrix_needs_recalc_callback);
227	aw_root->awar_int(AWAR_DIST_CORR_TRANS, (int)DI_TRANSFORMATION_SIMILARITY, def)->add_callback(matrix_needs_recalc_callback)->set_minmax(0, DI_TRANSFORMATION_COUNT-1);
228
229	aw_root->awar(AWAR_DIST_FILTER_ALIGNMENT)->map(AWAR_DIST_ALIGNMENT);
230
231	AW_create_fileselection_awars(aw_root, AWAR_DIST_SAVE_MATRIX_BASE, ".", "", "infile");
232	aw_root->awar_int(AWAR_DIST_SAVE_MATRIX_TYPE, 0, def);
233
234	enum treetype { CURR, SORT, COMPRESS, TREEAWARCOUNT };
235	AW_awar *tree_awar[TREEAWARCOUNT] = { NULL, NULL, NULL };
236
237	aw_root->awar_string(AWAR_DIST_TREE_STD_NAME, "tree_nj", def);
238	{
239	char *currentTree = aw_root->awar_string(AWAR_TREE, "", db)->read_string();
240	tree_awar[CURR] = aw_root->awar_string(AWAR_DIST_TREE_CURR_NAME, currentTree, def);
241	tree_awar[SORT] = aw_root->awar_string(AWAR_DIST_TREE_SORT_NAME, currentTree, def);
242	free(currentTree);
243	}
244	tree_awar[COMPRESS] = aw_root->awar_string(AWAR_DIST_TREE_COMP_NAME, NO_TREE_SELECTED, def);
245
246	aw_root->awar_int(AWAR_DIST_BOOTSTRAP_COUNT, 1000, def);
247	aw_root->awar_int(AWAR_DIST_MATRIX_AUTO_RECALC, 0, def)->add_callback(auto_calc_changed_cb);
248	aw_root->awar_int(AWAR_DIST_MATRIX_AUTO_CALC_TREE, 0, def)->add_callback(auto_calc_changed_cb);
249
250	aw_root->awar_float(AWAR_DIST_MIN_DIST, 0.0)->set_minmax(0.0, 4.0);
251	aw_root->awar_float(AWAR_DIST_MAX_DIST, 0.0)->set_minmax(0.0, 4.0);
252
253	aw_root->awar_string(AWAR_SPECIES_NAME, "", db);
254
255	DI_create_cluster_awars(aw_root, def, db);
256
257	#if defined(DEBUG)
258	AWT_create_db_browser_awars(aw_root, def);
259	#endif // DEBUG
260
261	{
262	GB_push_transaction(db);
263
264	GBDATA *gb_species_data = GBT_get_species_data(db);
265	GB_add_callback(gb_species_data, GB_CB_CHANGED, makeDatabaseCallback(matrix_needs_recalc_cb));
266
267	GB_pop_transaction(db);
268	}
269
270	AWT_registerTreeAwarCallback(tree_awar[CURR], makeTreeAwarCallback(selected_tree_changed_cb), true);
271	AWT_registerTreeAwarCallback(tree_awar[SORT], makeTreeAwarCallback(matrix_needs_recalc_cb), true);
272	AWT_registerTreeAwarCallback(tree_awar[COMPRESS], makeTreeAwarCallback(compressed_matrix_needs_recalc_cb), true);
273
274	auto_calc_changed_cb(aw_root);
275	}
276
277	DI_ENTRY::DI_ENTRY(GBDATA gbd, DI_MATRIX phmatrix_)
278	: phmatrix(phmatrix_),
279	full_name(NULL),
280	sequence(NULL),
281	name(NULL),
282	group_nr(0)
283	{
284	GBDATA *gb_ali = GB_entry(gbd, phmatrix->get_aliname());
285	if (gb_ali) {
286	GBDATA *gb_data = GB_entry(gb_ali, "data");
287	if (gb_data) {
288	if (phmatrix->is_AA) {
289	sequence = new AP_sequence_simple_protein(phmatrix->get_aliview());
290	}
291	else {
292	sequence = new AP_sequence_parsimony(phmatrix->get_aliview());
293	}
294	sequence->bind_to_species(gbd);
295	sequence->lazy_load_sequence(); // load sequence
296
297	name = GBT_read_string(gbd, "name");
298	full_name = GBT_read_string(gbd, "full_name");
299	}
300	}
301	}
302
303	DI_ENTRY::DI_ENTRY(const char name_, DI_MATRIX phmatrix_)
304	: phmatrix(phmatrix_),
305	full_name(NULL),
306	sequence(NULL),
307	name(strdup(name_)),
308	group_nr(0)
309	{}
310
311	DI_ENTRY::~DI_ENTRY() {
312	delete sequence;
313	free(name);
314	free(full_name);
315	}
316
317	DI_MATRIX::DI_MATRIX(const AliView& aliview_)
318	: gb_species_data(NULL),
319	seq_len(0),
320	entries_mem_size(0),
321	aliview(new AliView(aliview_)),
322	is_AA(false),
323	entries(NULL),
324	nentries(0),
325	matrix(NULL),
326	matrix_type(DI_MATRIX_FULL)
327	{
328	memset(cancel_columns, 0, sizeof(cancel_columns));
329	}
330
331	char *DI_MATRIX::unload() {
332	for (size_t i=0; i<nentries; i++) {
333	delete entries[i];
334	}
335	freenull(entries);
336	nentries = 0;
337	return 0;
338	}
339
340	DI_MATRIX::~DI_MATRIX() {
341	unload();
342	delete matrix;
343	delete aliview;
344	}
345
346	struct TreeOrderedSpecies {
347	GBDATA *gbd;
348	int order_index;
349
350	TreeOrderedSpecies(const MatrixOrder& order, GBDATA *gb_spec)
351	: gbd(gb_spec),
352	order_index(order.get_index(GBT_read_name(gbd)))
353	{}
354	};
355
356	MatrixOrder::MatrixOrder(GBDATA *gb_main, GB_CSTR sort_tree_name)
357	: name2pos(NULL),
358	leafs(0)
359	{
360	if (sort_tree_name) {
361	int size;
362	TreeNode *sort_tree = GBT_read_tree_and_size(gb_main, sort_tree_name, new SimpleRoot, &size);
363
364	if (sort_tree) {
365	leafs = size+1;
366	name2pos = GBS_create_hash(leafs, GB_IGNORE_CASE);
367
368	IF_ASSERTION_USED(int leafsLoaded = leafs);
369	leafs = 0;
370	insert_in_hash(sort_tree);
371
372	arb_assert(leafsLoaded == leafs);
373	destroy(sort_tree);
374	}
375	else {
376	GB_clear_error();
377	}
378	}
379	}
380	static int TreeOrderedSpecies_cmp(const void p1, const void p2, void *) {
381	TreeOrderedSpecies s1 = (TreeOrderedSpecies)p1;
382	TreeOrderedSpecies s2 = (TreeOrderedSpecies)p2;
383
384	return s2->order_index - s1->order_index;
385	}
386
387	void MatrixOrder::applyTo(TreeOrderedSpecies **species_array, size_t array_size) const {
388	GB_sort((void**)species_array, 0, array_size, TreeOrderedSpecies_cmp, NULL);
389	}
390
391	GB_ERROR DI_MATRIX::load(LoadWhat what, const MatrixOrder& order, bool show_warnings, GBDATA **species_list) {
392	GBDATA *gb_main = get_gb_main();
393	const char *use = get_aliname();
394
395	GB_transaction ta(gb_main);
396
397	seq_len = GBT_get_alignment_len(gb_main, use);
398	is_AA = GBT_is_alignment_protein(gb_main, use);
399	gb_species_data = GBT_get_species_data(gb_main);
400	entries_mem_size = 1000;
401
402	entries = (DI_ENTRY *)calloc(sizeof(DI_ENTRY), entries_mem_size);
403
404	nentries = 0;
405
406	size_t no_of_species = -1U;
407	switch (what) {
408	case DI_LOAD_ALL:
409	no_of_species = GBT_get_species_count(gb_main);
410	break;
411	case DI_LOAD_MARKED:
412	no_of_species = GBT_count_marked_species(gb_main);
413	break;
414	case DI_LOAD_LIST:
415	di_assert(species_list);
416	for (no_of_species = 0; species_list[no_of_species]; ++no_of_species) ;
417	break;
418	}
419
420	di_assert(no_of_species != -1U);
421	if (no_of_species<2) {
422	return GBS_global_string("Not enough input species (%zu)", no_of_species);
423	}
424
425	TreeOrderedSpecies *species_to_load[no_of_species];
426
427	{
428	size_t i = 0;
429	switch (what) {
430	case DI_LOAD_ALL: {
431	for (GBDATA *gb_species = GBT_first_species_rel_species_data(gb_species_data); gb_species; gb_species = GBT_next_species(gb_species), ++i) {
432	species_to_load[i] = new TreeOrderedSpecies(order, gb_species);
433	}
434	break;
435	}
436	case DI_LOAD_MARKED: {
437	for (GBDATA *gb_species = GBT_first_marked_species_rel_species_data(gb_species_data); gb_species; gb_species = GBT_next_marked_species(gb_species), ++i) {
438	species_to_load[i] = new TreeOrderedSpecies(order, gb_species);
439	}
440	break;
441	}
442	case DI_LOAD_LIST: {
443	for (i = 0; species_list[i]; ++i) {
444	species_to_load[i] = new TreeOrderedSpecies(order, species_list[i]);
445	}
446	break;
447	}
448	}
449	arb_assert(i == no_of_species);
450	}
451
452	if (order.defined()) {
453	order.applyTo(species_to_load, no_of_species);
454	if (show_warnings) {
455	int species_not_in_sort_tree = 0;
456	for (size_t i = 0; i<no_of_species; ++i) {
457	if (!species_to_load[i]->order_index) {
458	species_not_in_sort_tree++;
459	}
460	}
461	if (species_not_in_sort_tree) {
462	GBT_message(gb_main, GBS_global_string("Warning: %i of the affected species are not in sort-tree", species_not_in_sort_tree));
463	}
464	}
465	}
466	else {
467	if (show_warnings) {
468	static bool shown = false;
469	if (!shown) { // showing once is enough
470	GBT_message(gb_main, "Warning: No valid tree given to sort matrix (using default database order)");
471	shown = true;
472	}
473	}
474	}
475
476	if (no_of_species>entries_mem_size) {
477	entries_mem_size = no_of_species;
478	realloc_unleaked(entries, sizeof(DI_ENTRY)entries_mem_size);
479	if (!entries) return "out of memory";
480	}
481
482	GB_ERROR error = NULL;
483	arb_progress progress("Preparing sequence data", no_of_species);
484	for (size_t i = 0; i<no_of_species && !error; ++i) {
485	DI_ENTRY *phentry = new DI_ENTRY(species_to_load[i]->gbd, this);
486	if (phentry->sequence) { // a species found
487	arb_assert(nentries<entries_mem_size);
488	entries[nentries++] = phentry;
489	}
490	else {
491	delete phentry;
492	}
493	delete species_to_load[i];
494	species_to_load[i] = NULL;
495
496	progress.inc_and_check_user_abort(error);
497	}
498
499	return error;
500	}
501
502	char DI_MATRIX::calculate_overall_freqs(double rel_frequencies[AP_MAX], char cancel) {
503	di_assert(is_AA == false);
504
505	long hits2[AP_MAX];
506	long sum = 0;
507	int i;
508	int pos;
509	int b;
510	long s_len = aliview->get_length();
511
512	memset((char *) &hits2[0], 0, sizeof(hits2));
513	for (size_t row = 0; row < nentries; row++) {
514	const char *seq1 = entries[row]->get_nucl_seq()->get_sequence();
515	for (pos = 0; pos < s_len; pos++) {
516	b = *(seq1++);
517	if (cancel[b]) continue;
518	hits2[b]++;
519	}
520	}
521	for (i = 0; i < AP_MAX; i++) { // LOOP_VECTORIZED
522	sum += hits2[i];
523	}
524	for (i = 0; i < AP_MAX; i++) {
525	rel_frequencies[i] = hits2[i] / (double) sum;
526	}
527	return 0;
528	}
529
530	double DI_MATRIX::corr(double dist, double b, double & sigma) {
531	const double eps = 0.01;
532	double ar = 1.0 - dist/b;
533	sigma = 1000.0;
534	if (ar< eps) return 3.0;
535	sigma = b/ar;
536	return - b * log(1-dist/b);
537	}
538
539	GB_ERROR DI_MATRIX::calculate(const char cancel, DI_TRANSFORMATION transformation, bool aborted_flag, AP_matrix *userdef_matrix) {
540	di_assert(is_AA == false);
541
542	if (userdef_matrix) {
543	switch (transformation) {
544	case DI_TRANSFORMATION_NONE:
545	case DI_TRANSFORMATION_SIMILARITY:
546	case DI_TRANSFORMATION_JUKES_CANTOR:
547	break;
548	default:
549	aw_message("Sorry: this kind of distance correction does not support a user defined matrix - it will be ignored");
550	userdef_matrix = NULL;
551	break;
552	}
553	}
554
555	matrix = new AP_smatrix(nentries);
556
557	long s_len = aliview->get_length();
558	long hits[AP_MAX][AP_MAX];
559	size_t i;
560
561	if (nentries<=1) {
562	return "Not enough species selected to calculate matrix";
563	}
564	memset(&cancel_columns[0], 0, 256);
565
566	for (i=0; cancel[i]; i++) {
567	cancel_columns[safeCharIndex(AP_sequence_parsimony::table[safeCharIndex(cancel[i])])] = 1;
568	UNCOVERED(); // @@@ cover
569	}
570
571	long columns;
572	double b;
573	long frequencies[AP_MAX];
574	double rel_frequencies[AP_MAX];
575	double S_square = 0;
576
577	switch (transformation) {
578	case DI_TRANSFORMATION_FELSENSTEIN:
579	this->calculate_overall_freqs(rel_frequencies, cancel_columns);
580	S_square = 0.0;
581	for (i=0; i<AP_MAX; i++) S_square += rel_frequencies[i]*rel_frequencies[i];
582	break;
583	default: break;
584	};
585
586	arb_progress progress("Calculating distance matrix", matrix_halfsize(nentries, true));
587	GB_ERROR error = NULL;
588	for (size_t row = 0; row<nentries && !error; row++) {
589	for (size_t col=0; col<=row && !error; col++) {
590	columns = 0;
591
592	const unsigned char *seq1 = entries[row]->get_nucl_seq()->get_usequence();
593	const unsigned char *seq2 = entries[col]->get_nucl_seq()->get_usequence();
594
595	b = 0.0;
596	switch (transformation) {
597	case DI_TRANSFORMATION_FROM_TREE:
598	di_assert(0);
599	break;
600	case DI_TRANSFORMATION_JUKES_CANTOR:
601	b = 0.75;
602	// fall-through
603	case DI_TRANSFORMATION_NONE:
604	case DI_TRANSFORMATION_SIMILARITY: {
605	double dist = 0.0;
606	if (userdef_matrix) {
607	memset((char )hits, 0, sizeof(long) AP_MAX * AP_MAX);
608	int pos;
609	for (pos = s_len; pos >= 0; pos--) {
610	hits[(seq1++)][(seq2++)]++;
611	}
612	int x, y;
613	double diffsum = 0.0;
614	double all_sum = 0.001;
615	for (x = AP_A; x < AP_MAX; x*=2) {
616	for (y = AP_A; y < AP_MAX; y*=2) {
617	if (x==y) {
618	all_sum += hits[x][y];
619	}
620	else {
621	UNCOVERED(); // @@@ cover
622	diffsum += hits[x][y] * userdef_matrix->get(x, y);
623	all_sum += hits[x][y] * userdef_matrix->get(x, y);
624	}
625	}
626	}
627	dist = diffsum / all_sum;
628	}
629	else {
630	for (int pos = s_len; pos >= 0; pos--) {
631	int b1 = *(seq1++);
632	int b2 = *(seq2++);
633	if (cancel_columns[b1]) continue;
634	if (cancel_columns[b2]) continue;
635	columns++;
636	if (b1&b2) continue;
637	dist+=1.0;
638	}
639	if (columns == 0) columns = 1;
640	dist /= columns;
641	}
642	if (transformation==DI_TRANSFORMATION_SIMILARITY) {
643	dist = (1.0-dist);
644	}
645	else if (b) {
646	double sigma;
647	dist = this->corr(dist, b, sigma);
648	}
649	matrix->set(row, col, dist);
650	break;
651	}
652	case DI_TRANSFORMATION_KIMURA:
653	case DI_TRANSFORMATION_OLSEN:
654	case DI_TRANSFORMATION_FELSENSTEIN: {
655	int pos;
656	double dist = 0.0;
657	long N, P, Q, M;
658	double p, q;
659
660	memset((char )hits, 0, sizeof(long) AP_MAX * AP_MAX);
661	for (pos = s_len; pos >= 0; pos--) {
662	hits[(seq1++)][(seq2++)]++;
663	}
664	switch (transformation) {
665	case DI_TRANSFORMATION_KIMURA:
666	P = hits[AP_A][AP_G] +
667	hits[AP_G][AP_A] +
668	hits[AP_C][AP_T] +
669	hits[AP_T][AP_C];
670	Q = hits[AP_A][AP_C] +
671	hits[AP_A][AP_T] +
672	hits[AP_C][AP_A] +
673	hits[AP_T][AP_A] +
674	hits[AP_G][AP_C] +
675	hits[AP_G][AP_T] +
676	hits[AP_C][AP_G] +
677	hits[AP_T][AP_G];
678	M = hits[AP_A][AP_A] +
679	hits[AP_C][AP_C] +
680	hits[AP_G][AP_G] +
681	hits[AP_T][AP_T];
682	N = P+Q+M;
683	if (N==0) N=1;
684	p = (double)P/(double)N;
685	q = (double)Q/(double)N;
686	dist = - .5 * log(
687	(1.0-2.0p-q)sqrt(1.0-2.0*q)
688	);
689	break;
690
691	case DI_TRANSFORMATION_OLSEN:
692	case DI_TRANSFORMATION_FELSENSTEIN:
693
694	memset((char *)frequencies, 0,
695	sizeof(long) * AP_MAX);
696
697	N = 0;
698	M = 0;
699
700	for (i=0; i<AP_MAX; i++) {
701	if (cancel_columns[i]) continue;
702	unsigned int j;
703	for (j=0; j<i; j++) {
704	if (cancel_columns[j]) continue;
705	frequencies[i] +=
706	hits[i][j]+
707	hits[j][i];
708	}
709	frequencies[i] += hits[i][i];
710	N += frequencies[i];
711	M += hits[i][i];
712	}
713	if (N==0) N=1;
714	if (transformation == DI_TRANSFORMATION_OLSEN) { // Calc sum square freq individually for each line
715	S_square = 0.0;
716	for (i=0; i<AP_MAX; i++) S_square += frequencies[i]*frequencies[i];
717	b = 1.0 - S_square/((double)N*(double)N);
718	}
719	else {
720	b = 1.0 - S_square;
721	}
722
723	dist = ((double)(N-M)) / (double) N;
724	double sigma;
725	dist = this->corr(dist, b, sigma);
726	break;
727
728	default: return "Sorry: Transformation not implemented";
729	}
730	matrix->set(row, col, dist);
731	break;
732	}
733	default:;
734	} // switch
735	progress.inc_and_check_user_abort(error);
736	}
737	}
738	if (aborted_flag && progress.aborted()) *aborted_flag = true;
739	return error;
740	}
741
742	GB_ERROR DI_MATRIX::calculate_pro(DI_TRANSFORMATION transformation, bool *aborted_flag) {
743	di_assert(is_AA == true);
744
745	di_cattype catType;
746	switch (transformation) {
747	case DI_TRANSFORMATION_NONE: catType = NONE; break;
748	case DI_TRANSFORMATION_SIMILARITY: catType = SIMILARITY; break;
749	case DI_TRANSFORMATION_KIMURA: catType = KIMURA; break;
750	case DI_TRANSFORMATION_PAM: catType = PAM; break;
751	case DI_TRANSFORMATION_CATEGORIES_HALL: catType = HALL; break;
752	case DI_TRANSFORMATION_CATEGORIES_BARKER: catType = GEORGE; break;
753	case DI_TRANSFORMATION_CATEGORIES_CHEMICAL: catType = CHEMICAL; break;
754	default:
755	return "This correction is not available for protein data";
756	}
757	matrix = new AP_smatrix(nentries);
758
759	di_protdist prodist(UNIVERSAL, catType, nentries, entries, aliview->get_length(), matrix);
760	return prodist.makedists(aborted_flag);
761	}
762
763	typedef std::map<const char, TreeNode, charpLess> NamedNodes;
764
765	GB_ERROR link_to_tree(NamedNodes& named, TreeNode *node) {
766	GB_ERROR error = NULL;
767	if (node->is_leaf) {
768	NamedNodes::iterator found = named.find(node->name);
769	if (found != named.end()) {
770	if (found->second) {
771	error = GBS_global_string("Invalid tree (two nodes named '%s')", node->name);
772	}
773	else {
774	found->second = node;
775	}
776	}
777	// otherwise, we do not care about the node (e.g. because it is not marked)
778	}
779	else {
780	error = link_to_tree(named, node->get_leftson());
781	if (!error) error = link_to_tree(named, node->get_rightson());
782	}
783	return error;
784	}
785
786	#if defined(ASSERTION_USED)
787	static TreeNode findNode(TreeNode node, const char *name) {
788	if (node->is_leaf) {
789	return strcmp(node->name, name) == 0 ? node : NULL;
790	}
791
792	TreeNode *found = findNode(node->get_leftson(), name);
793	if (!found) found = findNode(node->get_rightson(), name);
794	return found;
795	}
796	#endif
797
798	static GB_ERROR init(NamedNodes& node, TreeNode tree, const DI_ENTRYconst*const entries, size_t nentries) {
799	GB_ERROR error = NULL;
800	for (size_t n = 0; n<nentries; ++n) {
801	node[entries[n]->name] = NULL;
802	}
803	error = link_to_tree(node, tree);
804	if (!error) { // check for missing species (needed but not in tree)
805	size_t missing = 0;
806	const char *exampleName = NULL;
807
808	for (size_t n = 0; n<nentries; ++n) {
809	NamedNodes::iterator found = node.find(entries[n]->name);
810	if (found == node.end()) {
811	++missing;
812	exampleName = entries[n]->name;
813	}
814	else {
815	di_assert(node[entries[n]->name] == findNode(tree, entries[n]->name));
816	if (!node[entries[n]->name]) {
817	++missing;
818	exampleName = entries[n]->name;
819	}
820	}
821	}
822
823	if (missing) {
824	error = GBS_global_string("Tree is missing %zu required species (e.g. '%s')", missing, exampleName);
825	}
826	}
827	return error;
828	}
829
830	GB_ERROR DI_MATRIX::extract_from_tree(const char treename, bool aborted_flag) {
831	GB_ERROR error = NULL;
832	if (nentries<=1) error = "Not enough species selected to calculate matrix";
833	else {
834	TreeNode *tree;
835	{
836	GB_transaction ta(get_gb_main());
837	tree = GBT_read_tree(get_gb_main(), treename, new SimpleRoot);
838	}
839	if (!tree) error = GB_await_error();
840	else {
841	arb_progress progress("Extracting distances from tree", matrix_halfsize(nentries, true));
842	NamedNodes node;
843
844	error = init(node, tree, entries, nentries);
845	matrix = new AP_smatrix(nentries);
846
847	for (size_t row = 0; row<nentries && !error; row++) {
848	TreeNode *rnode = node[entries[row]->name];
849	for (size_t col=0; col<=row && !error; col++) {
850	double dist;
851	if (col != row) {
852	TreeNode *cnode = node[entries[col]->name];
853	dist = rnode->intree_distance_to(cnode);
854	}
855	else {
856	dist = 0.0;
857	}
858	matrix->set(row, col, dist);
859	progress.inc_and_check_user_abort(error);
860	}
861	}
862	UNCOVERED();
863	destroy(tree);
864	if (aborted_flag && progress.aborted()) *aborted_flag = true;
865	if (error) progress.done();
866	}
867	}
868	return error;
869	}
870
871	__ATTR__USERESULT static GB_ERROR di_calculate_matrix(AW_root aw_root, const WeightedFilter weighted_filter, bool bootstrap_flag, bool show_warnings, bool *aborted_flag) {
872	// sets 'aborted_flag' to true, if it is non-NULL and the calculation has been aborted
873	GB_ERROR error = NULL;
874
875	if (GLOBAL_MATRIX.exists()) {
876	di_assert(!need_recalc.matrix);
877	}
878	else {
879	GB_transaction ta(GLOBAL_gb_main);
880
881	char *use = aw_root->awar(AWAR_DIST_ALIGNMENT)->read_string();
882	long ali_len = GBT_get_alignment_len(GLOBAL_gb_main, use);
883
884	if (ali_len<=0) {
885	error = "Please select a valid alignment";
886	GB_clear_error();
887	}
888	else {
889	arb_progress progress("Calculating matrix");
890	char *cancel = aw_root->awar(AWAR_DIST_CANCEL_CHARS)->read_string();
891	AliView *aliview = weighted_filter->create_aliview(use, error);
892
893	if (!error) {
894	if (bootstrap_flag) aliview->get_filter()->enable_bootstrap();
895
896	char *load_what = aw_root->awar(AWAR_DIST_WHICH_SPECIES)->read_string();
897	char *sort_tree_name = aw_root->awar(AWAR_DIST_TREE_SORT_NAME)->read_string();
898
899	LoadWhat all_flag = (strcmp(load_what, "all") == 0) ? DI_LOAD_ALL : DI_LOAD_MARKED;
900	{
901	DI_MATRIX phm = new DI_MATRIX(aliview);
902	phm->matrix_type = DI_MATRIX_FULL;
903
904	static SmartCharPtr last_sort_tree_name;
905	static SmartPtr<MatrixOrder> last_order;
906
907	if (last_sort_tree_name.isNull() \|\| !sort_tree_name \|\| strcmp(&*last_sort_tree_name, sort_tree_name) != 0) {
908	last_sort_tree_name = nulldup(sort_tree_name);
909	last_order = new MatrixOrder(GLOBAL_gb_main, sort_tree_name);
910	}
911	di_assert(last_order.isSet());
912	error = phm->load(all_flag, *last_order, show_warnings, NULL);
913
914	free(sort_tree_name);
915	error = ta.close(error);
916
917	bool aborted = false;
918	if (!error) {
919	if (progress.aborted()) {
920	phm->unload();
921	error = "Aborted by user";
922	aborted = true;
923	}
924	else {
925	DI_TRANSFORMATION trans = (DI_TRANSFORMATION)aw_root->awar(AWAR_DIST_CORR_TRANS)->read_int();
926
927	if (trans == DI_TRANSFORMATION_FROM_TREE) {
928	const char *treename = aw_root->awar(AWAR_DIST_TREE_CURR_NAME)->read_char_pntr();
929	error = phm->extract_from_tree(treename, &aborted);
930	}
931	else {
932	if (phm->is_AA) error = phm->calculate_pro(trans, &aborted);
933	else error = phm->calculate(cancel, trans, &aborted, get_user_matrix());
934	}
935	}
936	}
937
938	if (aborted) {
939	di_assert(error);
940	if (aborted_flag) *aborted_flag = true;
941	}
942	if (error) {
943	delete phm;
944	GLOBAL_MATRIX.forget();
945	}
946	else {
947	GLOBAL_MATRIX.replaceBy(phm);
948	tree_needs_recalc_cb();
949	need_recalc.matrix = false;
950	}
951	}
952	free(load_what);
953	}
954
955	free(cancel);
956	delete aliview;
957	}
958	free(use);
959
960	di_assert(contradicted(error, GLOBAL_MATRIX.exists()));
961	}
962	return error;
963	}
964
965	static void di_mark_by_distance(AW_window aww, WeightedFilter weighted_filter) {
966	AW_root *aw_root = aww->get_root();
967	float lowerBound = aw_root->awar(AWAR_DIST_MIN_DIST)->read_float();
968	float upperBound = aw_root->awar(AWAR_DIST_MAX_DIST)->read_float();
969
970	GB_ERROR error = 0;
971	if (lowerBound >= upperBound) {
972	error = GBS_global_string("Lower bound (%f) has to be smaller than upper bound (%f)", lowerBound, upperBound);
973	}
974	else if (lowerBound<0.0 \|\| lowerBound > 1.0) {
975	error = GBS_global_string("Lower bound (%f) is not in allowed range [0.0 .. 1.0]", lowerBound);
976	}
977	else if (upperBound<0.0 \|\| upperBound > 1.0) {
978	error = GBS_global_string("Upper bound (%f) is not in allowed range [0.0 .. 1.0]", upperBound);
979	}
980	else {
981	GB_transaction ta(GLOBAL_gb_main);
982
983	char *selected = aw_root->awar(AWAR_SPECIES_NAME)->read_string();
984	if (!selected[0]) {
985	error = "Please select a species";
986	}
987	else {
988	GBDATA *gb_selected = GBT_find_species(GLOBAL_gb_main, selected);
989	if (!gb_selected) {
990	error = GBS_global_string("Couldn't find species '%s'", selected);
991	}
992	else {
993	char *use = aw_root->awar(AWAR_DIST_ALIGNMENT)->read_string();
994	char *cancel = aw_root->awar(AWAR_DIST_CANCEL_CHARS)->read_string();
995	DI_TRANSFORMATION trans = (DI_TRANSFORMATION)aw_root->awar(AWAR_DIST_CORR_TRANS)->read_int();
996	AliView *aliview = weighted_filter->create_aliview(use, error);
997
998	if (!error) {
999	DI_MATRIX *prev_global = GLOBAL_MATRIX.swap(NULL);
1000
1001	size_t speciesCount = GBT_get_species_count(GLOBAL_gb_main);
1002	bool markedSelected = false;
1003
1004	arb_progress progress("Mark species by distance", speciesCount);
1005	MatrixOrder order(GLOBAL_gb_main, NULL);
1006
1007	for (GBDATA *gb_species = GBT_first_species(GLOBAL_gb_main);
1008	gb_species && !error;
1009	gb_species = GBT_next_species(gb_species))
1010	{
1011	DI_MATRIX phm = new DI_MATRIX(aliview);
1012	phm->matrix_type = DI_MATRIX_FULL;
1013	GBDATA *species_pair[] = { gb_selected, gb_species, NULL };
1014
1015	error = phm->load(DI_LOAD_LIST, order, false, species_pair);
1016
1017	if (phm->nentries == 2) { // if species has no alignment -> nentries<2
1018	if (!error) {
1019	if (phm->is_AA) error = phm->calculate_pro(trans, NULL);
1020	else error = phm->calculate(cancel, trans, NULL, get_user_matrix());
1021	}
1022
1023	if (!error) {
1024	double dist_value = phm->matrix->get(0, 1); // distance or conformance
1025	bool mark = (lowerBound <= dist_value && dist_value <= upperBound);
1026	GB_write_flag(gb_species, mark);
1027
1028	if (!markedSelected) {
1029	dist_value = phm->matrix->get(0, 0); // distance or conformance to self
1030	mark = (lowerBound <= dist_value && dist_value <= upperBound);
1031	GB_write_flag(gb_selected, mark);
1032
1033	markedSelected = true;
1034	}
1035	}
1036	}
1037
1038	delete phm;
1039	if (!error) progress.inc_and_check_user_abort(error);
1040	}
1041
1042	di_assert(!GLOBAL_MATRIX.exists());
1043	ASSERT_RESULT(DI_MATRIX*, NULL, GLOBAL_MATRIX.swap(prev_global));
1044
1045	if (error) progress.done();
1046	}
1047
1048	delete aliview;
1049	free(cancel);
1050	free(use);
1051	}
1052	}
1053
1054	free(selected);
1055	error = ta.close(error);
1056	}
1057
1058	if (error) {
1059	aw_message(error);
1060	}
1061	}
1062
1063	static GB_ERROR di_recalc_matrix() {
1064	// recalculate matrix
1065	last_matrix_calculation_error = NULL;
1066	if (need_recalc.matrix && GLOBAL_MATRIX.exists()) {
1067	GLOBAL_MATRIX.forget();
1068	}
1069	di_assert(recalculate_matrix_cb.isSet());
1070	(*recalculate_matrix_cb)();
1071	return last_matrix_calculation_error;
1072	}
1073
1074	static void di_view_matrix_cb(AW_window aww, save_matrix_params sparam) {
1075	GB_ERROR error = di_recalc_matrix();
1076	if (error) return;
1077
1078	if (!matrixDisplay) matrixDisplay = new MatrixDisplay;
1079
1080	static AW_window *viewer = 0;
1081	if (!viewer) viewer = DI_create_view_matrix_window(aww->get_root(), matrixDisplay, sparam);
1082
1083	matrixDisplay->mark(MatrixDisplay::NEED_SETUP);
1084	matrixDisplay->update_display();
1085
1086	GLOBAL_MATRIX.set_changed_cb(matrix_changed_cb);
1087
1088	viewer->activate();
1089	}
1090
1091	static void di_save_matrix_cb(AW_window *aww) {
1092	// save the matrix
1093	GB_ERROR error = di_recalc_matrix();
1094	if (!error) {
1095	char *filename = aww->get_root()->awar(AWAR_DIST_SAVE_MATRIX_FILENAME)->read_string();
1096	enum DI_SAVE_TYPE type = (enum DI_SAVE_TYPE)aww->get_root()->awar(AWAR_DIST_SAVE_MATRIX_TYPE)->read_int();
1097
1098	GLOBAL_MATRIX.get()->save(filename, type);
1099	free(filename);
1100	}
1101	AW_refresh_fileselection(aww->get_root(), AWAR_DIST_SAVE_MATRIX_BASE);
1102	aww->hide_or_notify(error);
1103	}
1104
1105	AW_window DI_create_save_matrix_window(AW_root aw_root, save_matrix_params *save_params) {
1106	static AW_window_simple *aws = 0;
1107	if (!aws) {
1108	aws = new AW_window_simple;
1109	aws->init(aw_root, "SAVE_MATRIX", "Save Matrix");
1110	aws->load_xfig("sel_box_user.fig");
1111
1112	aws->at("close");
1113	aws->callback(AW_POPDOWN);
1114	aws->create_button("CLOSE", "CANCEL", "C");
1115
1116
1117	aws->at("help"); aws->callback(makeHelpCallback("save_matrix.hlp"));
1118	aws->create_button("HELP", "HELP", "H");
1119
1120	aws->at("user");
1121	aws->create_option_menu(AWAR_DIST_SAVE_MATRIX_TYPE, true);
1122	aws->insert_default_option("Phylip Format (Lower Triangular Matrix)", "P", DI_SAVE_PHYLIP_COMP);
1123	aws->insert_option("Readable (using NDS)", "R", DI_SAVE_READABLE);
1124	aws->insert_option("Tabbed (using NDS)", "R", DI_SAVE_TABBED);
1125	aws->update_option_menu();
1126
1127	AW_create_standard_fileselection(aws, save_params->awar_base);
1128
1129	aws->at("save2");
1130	aws->callback(makeWindowCallback(di_save_matrix_cb));
1131	aws->create_button("SAVE", "SAVE", "S");
1132
1133	aws->at("cancel2");
1134	aws->callback(AW_POPDOWN);
1135	aws->create_button("CLOSE", "CANCEL", "C");
1136	}
1137	return aws;
1138	}
1139
1140	static AW_window awt_create_select_cancel_window(AW_root aw_root) {
1141	AW_window_simple *aws = new AW_window_simple;
1142	aws->init(aw_root, "SELECT_CHARS_TO_CANCEL_COLUMN", "CANCEL SELECT");
1143	aws->load_xfig("di_cancel.fig");
1144
1145	aws->at("close");
1146	aws->callback(AW_POPDOWN);
1147	aws->create_button("CLOSE", "CLOSE", "C");
1148
1149	aws->at("cancel");
1150	aws->create_input_field(AWAR_DIST_CANCEL_CHARS, 12);
1151
1152	return (AW_window *)aws;
1153	}
1154
1155	static const char *enum_trans_to_string[] = {
1156	"none",
1157	"similarity",
1158	"jukes_cantor",
1159	"felsenstein",
1160
1161	"pam",
1162	"hall",
1163	"barker",
1164	"chemical",
1165
1166	"kimura",
1167	"olsen",
1168	"felsenstein voigt",
1169	"olsen voigt",
1170	"max ml",
1171
1172	NULL, // treedist
1173	};
1174
1175	STATIC_ASSERT(ARRAY_ELEMS(enum_trans_to_string) == DI_TRANSFORMATION_COUNT);
1176
1177	static void di_calculate_tree_cb(AW_window aww, WeightedFilter weighted_filter, bool bootstrap_flag) {
1178	recalculate_tree_cb = new BoundWindowCallback(aww, makeWindowCallback(di_calculate_tree_cb, weighted_filter, bootstrap_flag));
1179
1180	AW_root *aw_root = aww->get_root();
1181	GB_ERROR error = 0;
1182	StrArray *all_names = 0;
1183
1184	int loop_count = 0;
1185	int bootstrap_count = aw_root->awar(AWAR_DIST_BOOTSTRAP_COUNT)->read_int();
1186
1187	{
1188	char *tree_name = aw_root->awar(AWAR_DIST_TREE_STD_NAME)->read_string();
1189	error = GBT_check_tree_name(tree_name);
1190	free(tree_name);
1191	}
1192
1193	SmartPtr<arb_progress> progress;
1194	SmartPtr<ConsensusTree> ctree;
1195
1196	if (!error) {
1197	if (bootstrap_flag) {
1198	if (bootstrap_count) {
1199	progress = new arb_progress("Calculating bootstrap trees", bootstrap_count+1);
1200	}
1201	else {
1202	progress = new arb_progress("Calculating bootstrap trees (KILL to stop)", INT_MAX);
1203	}
1204	progress->auto_subtitles("tree");
1205	}
1206	else {
1207	progress = new arb_progress("Calculating tree");
1208	}
1209
1210	if (bootstrap_flag) {
1211	GLOBAL_MATRIX.forget();
1212	GLOBAL_MATRIX.set_changed_cb(NULL); // otherwise matrix window will repeatedly pop up/down
1213
1214	error = di_calculate_matrix(aw_root, weighted_filter, bootstrap_flag, true, NULL);
1215	if (!error) {
1216	DI_MATRIX *matr = GLOBAL_MATRIX.get();
1217	if (!matr) {
1218	error = "unexpected error in di_calculate_matrix_cb (data missing)";
1219	}
1220	else {
1221	all_names = new StrArray;
1222	all_names->reserve(matr->nentries+2);
1223
1224	for (size_t i=0; i<matr->nentries; i++) {
1225	all_names->put(strdup(matr->entries[i]->name));
1226	}
1227	ctree = new ConsensusTree(*all_names);
1228	}
1229	}
1230	}
1231	}
1232
1233	TreeNode *tree = 0;
1234	do {
1235	if (error) break;
1236
1237	bool aborted = false;
1238
1239	if (bootstrap_flag) {
1240	if (loop_count>0) { // in first loop we already have a valid matrix -> no need to recalculate
1241	GLOBAL_MATRIX.forget();
1242	}
1243	}
1244	else if (need_recalc.matrix) {
1245	GLOBAL_MATRIX.forget();
1246	}
1247
1248	error = di_calculate_matrix(aw_root, weighted_filter, bootstrap_flag, !bootstrap_flag, &aborted);
1249	if (error && aborted) {
1250	error = 0; // clear error (otherwise no tree will be read below)
1251	break; // end of bootstrap
1252	}
1253
1254	if (!GLOBAL_MATRIX.exists()) {
1255	error = "unexpected error in di_calculate_matrix_cb (data missing)";
1256	break;
1257	}
1258
1259	DI_MATRIX *matr = GLOBAL_MATRIX.get();
1260	char names = (char )calloc(sizeof(char *), (size_t)matr->nentries+2);
1261
1262	for (size_t i=0; i<matr->nentries; i++) {
1263	names[i] = matr->entries[i]->name;
1264	}
1265	di_assert(matr->nentries == matr->matrix->size());
1266	tree = neighbourjoining(names, *matr->matrix, new SimpleRoot);
1267
1268	if (bootstrap_flag) {
1269	error = ctree->insert_tree_weighted(tree, matr->nentries, 1, false);
1270	UNCOVERED();
1271	destroy(tree); tree = NULL;
1272	loop_count++;
1273	progress->inc();
1274	if (!bootstrap_count) { // when waiting for kill
1275	int t = time(0);
1276	static int tlast = 0;
1277
1278	if (t>tlast) {
1279	progress->force_update();
1280	tlast = t;
1281	}
1282	}
1283	}
1284	free(names);
1285	} while (bootstrap_flag && loop_count != bootstrap_count);
1286
1287	if (!error) {
1288	if (bootstrap_flag) {
1289	tree = ctree->get_consensus_tree(error);
1290	progress->inc();
1291	if (!error) {
1292	error = GBT_is_invalid(tree);
1293	di_assert(!error);
1294	}
1295	}
1296
1297	if (!error) {
1298	char *tree_name = aw_root->awar(AWAR_DIST_TREE_STD_NAME)->read_string();
1299	GB_begin_transaction(GLOBAL_gb_main);
1300	error = GBT_write_tree(GLOBAL_gb_main, tree_name, tree);
1301
1302	if (!error) {
1303	char *filter_name = AWT_get_combined_filter_name(aw_root, "dist");
1304	int transr = aw_root->awar(AWAR_DIST_CORR_TRANS)->read_int();
1305
1306	const char *comment;
1307	if (enum_trans_to_string[transr]) {
1308	comment = GBS_global_string("PRG=dnadist CORR=%s FILTER=%s PKG=ARB", enum_trans_to_string[transr], filter_name);
1309	}
1310	else {
1311	di_assert(transr == DI_TRANSFORMATION_FROM_TREE);
1312	const char *treename = aw_root->awar(AWAR_DIST_TREE_CURR_NAME)->read_char_pntr();
1313	comment = GBS_global_string("PRG=treedist (from '%s') PKG=ARB", treename);
1314	}
1315
1316	error = GBT_write_tree_remark(GLOBAL_gb_main, tree_name, comment);
1317	free(filter_name);
1318	}
1319	error = GB_end_transaction(GLOBAL_gb_main, error);
1320	free(tree_name);
1321	}
1322	}
1323
1324	UNCOVERED();
1325	destroy(tree);
1326
1327	// aw_status(); // remove 'abort' flag (@@@ got no equiv for arb_progress yet. really needed?)
1328
1329	if (bootstrap_flag) {
1330	if (all_names) delete all_names;
1331	GLOBAL_MATRIX.forget();
1332	}
1333	#if defined(DEBUG)
1334	else {
1335	di_assert(all_names == 0);
1336	}
1337	#endif // DEBUG
1338
1339	progress->done();
1340	if (error) {
1341	aw_message(error);
1342	}
1343	else {
1344	need_recalc.tree = false;
1345	aw_root->awar(AWAR_TREE_REFRESH)->touch();
1346	}
1347	}
1348
1349
1350	static void di_autodetect_callback(AW_window *aww) {
1351	GB_push_transaction(GLOBAL_gb_main);
1352
1353	GLOBAL_MATRIX.forget();
1354
1355	AW_root *aw_root = aww->get_root();
1356	char *use = aw_root->awar(AWAR_DIST_ALIGNMENT)->read_string();
1357	long ali_len = GBT_get_alignment_len(GLOBAL_gb_main, use);
1358	GB_ERROR error = NULL;
1359
1360	if (ali_len<=0) {
1361	GB_pop_transaction(GLOBAL_gb_main);
1362	error = "Please select a valid alignment";
1363	GB_clear_error();
1364	}
1365	else {
1366	arb_progress progress("Analyzing data");
1367
1368	char *filter_str = aw_root->awar(AWAR_DIST_FILTER_FILTER)->read_string();
1369	char *cancel = aw_root->awar(AWAR_DIST_CANCEL_CHARS)->read_string();
1370
1371	AliView *aliview = NULL;
1372	{
1373	AP_filter *ap_filter = NULL;
1374	long flen = strlen(filter_str);
1375
1376	if (flen == ali_len) {
1377	ap_filter = new AP_filter(filter_str, "0", ali_len);
1378	}
1379	else {
1380	if (flen) {
1381	aw_message("Warning: your filter len is not equal to the alignment len\nfilter got truncated with zeros or cutted");
1382	ap_filter = new AP_filter(filter_str, "0", ali_len);
1383	}
1384	else {
1385	ap_filter = new AP_filter(ali_len); // unfiltered
1386	}
1387	}
1388
1389	error = ap_filter->is_invalid();
1390	if (!error) {
1391	AP_weights ap_weights(ap_filter);
1392	aliview = new AliView(GLOBAL_gb_main, *ap_filter, ap_weights, use);
1393	}
1394	delete ap_filter;
1395	}
1396
1397	if (error) {
1398	GB_pop_transaction(GLOBAL_gb_main);
1399	}
1400	else {
1401	DI_MATRIX phm(*aliview);
1402
1403	{
1404	char *load_what = aw_root->awar(AWAR_DIST_WHICH_SPECIES)->read_string();
1405	char *sort_tree_name = aw_root->awar(AWAR_DIST_TREE_SORT_NAME)->read_string();
1406
1407	LoadWhat all_flag = (strcmp(load_what, "all") == 0) ? DI_LOAD_ALL : DI_LOAD_MARKED;
1408
1409	GLOBAL_MATRIX.forget();
1410
1411	MatrixOrder order(GLOBAL_gb_main, sort_tree_name);
1412	error = phm.load(all_flag, order, true, NULL);
1413
1414	free(sort_tree_name);
1415	free(load_what);
1416	}
1417
1418	GB_pop_transaction(GLOBAL_gb_main);
1419
1420	if (!error) {
1421	progress.subtitle("Search Correction");
1422
1423	string msg;
1424	DI_TRANSFORMATION detected = phm.detect_transformation(msg);
1425	aw_root->awar(AWAR_DIST_CORR_TRANS)->write_int(detected);
1426	aw_message(msg.c_str());
1427	}
1428	}
1429
1430	free(cancel);
1431	delete aliview;
1432
1433	free(filter_str);
1434	}
1435
1436	if (error) aw_message(error);
1437
1438	free(use);
1439	}
1440
1441	__ATTR__NORETURN static void di_exit(AW_window *aww) {
1442	if (GLOBAL_gb_main) {
1443	AW_root *aw_root = aww->get_root();
1444	shutdown_macro_recording(aw_root);
1445	aw_root->unlink_awars_from_DB(GLOBAL_gb_main);
1446	GB_close(GLOBAL_gb_main);
1447	}
1448	GLOBAL_MATRIX.set_changed_cb(NULL);
1449	exit(EXIT_SUCCESS);
1450	}
1451
1452	static void di_calculate_full_matrix_cb(AW_window aww, const WeightedFilter weighted_filter) {
1453	recalculate_matrix_cb = new BoundWindowCallback(aww, makeWindowCallback(di_calculate_full_matrix_cb, weighted_filter));
1454
1455	GLOBAL_MATRIX.forget_if_not_has_type(DI_MATRIX_FULL);
1456	GB_ERROR error = di_calculate_matrix(aww->get_root(), weighted_filter, 0, true, NULL);
1457	aw_message_if(error);
1458	last_matrix_calculation_error = error;
1459	if (!error) tree_needs_recalc_cb();
1460	}
1461
1462	static void di_calculate_compressed_matrix_cb(AW_window aww, WeightedFilter weighted_filter) {
1463	recalculate_matrix_cb = new BoundWindowCallback(aww, makeWindowCallback(di_calculate_compressed_matrix_cb, weighted_filter));
1464
1465	GB_transaction ta(GLOBAL_gb_main);
1466
1467	AW_root *aw_root = aww->get_root();
1468	char *treename = aw_root->awar(AWAR_DIST_TREE_COMP_NAME)->read_string();
1469	GB_ERROR error = 0;
1470	TreeNode *tree = GBT_read_tree(GLOBAL_gb_main, treename, new SimpleRoot);
1471
1472	if (!tree) {
1473	error = GB_await_error();
1474	}
1475	else {
1476	{
1477	LocallyModify<MatrixDisplay*> skipRefresh(matrixDisplay, NULL); // skip refresh, until matrix has been compressed
1478
1479	GLOBAL_MATRIX.forget(); // always forget (as tree might have changed)
1480	error = di_calculate_matrix(aw_root, weighted_filter, 0, true, NULL);
1481	if (!error && !GLOBAL_MATRIX.exists()) {
1482	error = "Failed to calculate your matrix (bug?)";
1483	}
1484	if (!error) {
1485	error = GLOBAL_MATRIX.get()->compress(tree);
1486	}
1487	}
1488	UNCOVERED();
1489	destroy(tree);
1490
1491	// now force refresh
1492	if (matrixDisplay) {
1493	matrixDisplay->mark(MatrixDisplay::NEED_SETUP);
1494	matrixDisplay->update_display();
1495	}
1496	}
1497	free(treename);
1498	aw_message_if(error);
1499	last_matrix_calculation_error = error;
1500	if (!error) tree_needs_recalc_cb();
1501	}
1502
1503	static void di_define_sort_tree_name_cb(AW_window *aww) {
1504	AW_root *aw_root = aww->get_root();
1505	char *tree_name = aw_root->awar(AWAR_DIST_TREE_CURR_NAME)->read_string();
1506	aw_root->awar(AWAR_DIST_TREE_SORT_NAME)->write_string(tree_name);
1507	free(tree_name);
1508	}
1509	static void di_define_compression_tree_name_cb(AW_window *aww) {
1510	AW_root *aw_root = aww->get_root();
1511	char *tree_name = aw_root->awar(AWAR_DIST_TREE_CURR_NAME)->read_string();
1512	aw_root->awar(AWAR_DIST_TREE_COMP_NAME)->write_string(tree_name);
1513	free(tree_name);
1514	}
1515
1516	static void di_define_save_tree_name_cb(AW_window *aww) {
1517	AW_root *aw_root = aww->get_root();
1518	char *tree_name = aw_root->awar(AWAR_DIST_TREE_CURR_NAME)->read_string();
1519	aw_root->awar(AWAR_DIST_TREE_STD_NAME)->write_string(tree_name);
1520	free(tree_name);
1521	}
1522
1523
1524	AW_window DI_create_matrix_window(AW_root aw_root) {
1525	AW_window_simple_menu *aws = new AW_window_simple_menu;
1526	aws->init(aw_root, "NEIGHBOUR JOINING", "NEIGHBOUR JOINING [ARB_DIST]");
1527	aws->load_xfig("di_ge_ma.fig");
1528	aws->button_length(10);
1529
1530	aws->at("close");
1531	aws->callback(di_exit);
1532	aws->create_button("CLOSE", "CLOSE", "C");
1533
1534	aws->at("help");
1535	aws->callback(makeHelpCallback("dist.hlp"));
1536	aws->create_button("HELP", "HELP", "H");
1537
1538
1539	GB_push_transaction(GLOBAL_gb_main);
1540
1541	#if defined(DEBUG)
1542	AWT_create_debug_menu(aws);
1543	#endif // DEBUG
1544
1545	aws->create_menu("File", "F", AWM_ALL);
1546	insert_macro_menu_entry(aws, false);
1547	aws->insert_menu_topic("quit", "Quit", "Q", "quit.hlp", AWM_ALL, di_exit);
1548
1549	aws->create_menu("Properties", "P", AWM_ALL);
1550	#if defined(ARB_MOTIF)
1551	aws->insert_menu_topic("frame_props", "Frame settings ...", "F", "props_frame.hlp", AWM_ALL, AW_preset_window);
1552	aws->sep______________();
1553	#endif
1554	AW_insert_common_property_menu_entries(aws);
1555	aws->sep______________();
1556	aws->insert_menu_topic("save_props", "Save Properties (dist.arb)", "S", "savedef.hlp", AWM_ALL, AW_save_properties);
1557
1558	aws->insert_help_topic("ARB_DIST help", "D", "dist.hlp", AWM_ALL, makeHelpCallback("dist.hlp"));
1559
1560	// ------------------
1561	// left side
1562
1563	aws->at("which_species");
1564	aws->create_option_menu(AWAR_DIST_WHICH_SPECIES, true);
1565	aws->insert_option("all", "a", "all");
1566	aws->insert_default_option("marked", "m", "marked");
1567	aws->update_option_menu();
1568
1569	aws->at("which_alignment");
1570	awt_create_ALI_selection_list(GLOBAL_gb_main, (AW_window )aws, AWAR_DIST_ALIGNMENT, "=");
1571
1572	// filter & weights
1573
1574	AW_awar *awar_dist_alignment = aws->get_root()->awar_string(AWAR_DIST_ALIGNMENT);
1575	WeightedFilter *weighted_filter = // do NOT free (bound to callbacks)
1576	new WeightedFilter(GLOBAL_gb_main, aws->get_root(), AWAR_DIST_FILTER_NAME, AWAR_DIST_COLUMN_STAT_NAME, awar_dist_alignment);
1577
1578	aws->at("filter_select");
1579	aws->callback(makeCreateWindowCallback(awt_create_select_filter_win, weighted_filter->get_adfiltercbstruct()));
1580	aws->create_button("SELECT_FILTER", AWAR_DIST_FILTER_NAME);
1581
1582	aws->at("weights_select");
1583	aws->sens_mask(AWM_EXP);
1584	aws->callback(makeCreateWindowCallback(COLSTAT_create_selection_window, weighted_filter->get_column_stat()));
1585	aws->create_button("SELECT_COL_STAT", AWAR_DIST_COLUMN_STAT_NAME);
1586	aws->sens_mask(AWM_ALL);
1587
1588	aws->at("which_cancel");
1589	aws->create_input_field(AWAR_DIST_CANCEL_CHARS, 12);
1590
1591	aws->at("cancel_select");
1592	aws->callback(awt_create_select_cancel_window);
1593	aws->create_button("SELECT_CANCEL_CHARS", "Info", "C");
1594
1595	aws->at("change_matrix");
1596	aws->callback(create_dna_matrix_window);
1597	aws->create_button("EDIT_MATRIX", "Edit Matrix");
1598
1599	aws->at("enable");
1600	aws->create_toggle(AWAR_DIST_MATRIX_DNA_ENABLED);
1601
1602	aws->at("which_correction");
1603	aws->create_option_menu(AWAR_DIST_CORR_TRANS, true);
1604	aws->insert_option("none", "n", (int)DI_TRANSFORMATION_NONE);
1605	aws->insert_option("similarity", "n", (int)DI_TRANSFORMATION_SIMILARITY);
1606	aws->insert_option("jukes-cantor (dna)", "c", (int)DI_TRANSFORMATION_JUKES_CANTOR);
1607	aws->insert_option("felsenstein (dna)", "f", (int)DI_TRANSFORMATION_FELSENSTEIN);
1608	aws->insert_option("olsen (dna)", "o", (int)DI_TRANSFORMATION_OLSEN);
1609	aws->insert_option("felsenstein/voigt (exp)", "1", (int)DI_TRANSFORMATION_FELSENSTEIN_VOIGT);
1610	aws->insert_option("olsen/voigt (exp)", "2", (int)DI_TRANSFORMATION_OLSEN_VOIGT);
1611	aws->insert_option("kimura (pro)", "k", (int)DI_TRANSFORMATION_KIMURA);
1612	aws->insert_option("PAM (protein)", "c", (int)DI_TRANSFORMATION_PAM);
1613	aws->insert_option("Cat. Hall(exp)", "c", (int)DI_TRANSFORMATION_CATEGORIES_HALL);
1614	aws->insert_option("Cat. Barker(exp)", "c", (int)DI_TRANSFORMATION_CATEGORIES_BARKER);
1615	aws->insert_option("Cat.Chem (exp)", "c", (int)DI_TRANSFORMATION_CATEGORIES_CHEMICAL);
1616	aws->insert_option("from selected tree", "t", (int)DI_TRANSFORMATION_FROM_TREE);
1617	aws->insert_default_option("unknown", "u", (int)DI_TRANSFORMATION_NONE);
1618
1619	aws->update_option_menu();
1620
1621	aws->at("autodetect"); // auto
1622	aws->callback(di_autodetect_callback);
1623	aws->sens_mask(AWM_EXP);
1624	aws->create_button("AUTODETECT_CORRECTION", "AUTODETECT", "A");
1625	aws->sens_mask(AWM_ALL);
1626
1627	// -------------------
1628	// right side
1629
1630
1631	aws->at("mark_distance");
1632	aws->callback(makeWindowCallback(di_mark_by_distance, weighted_filter));
1633	aws->create_autosize_button("MARK_BY_DIST", "Mark all species");
1634
1635	aws->at("mark_lower");
1636	aws->create_input_field(AWAR_DIST_MIN_DIST, 5);
1637
1638	aws->at("mark_upper");
1639	aws->create_input_field(AWAR_DIST_MAX_DIST, 5);
1640
1641	// -----------------
1642
1643	// tree selection
1644
1645	aws->at("tree_list");
1646	awt_create_TREE_selection_list(GLOBAL_gb_main, aws, AWAR_DIST_TREE_CURR_NAME, true);
1647
1648	aws->at("detect_clusters");
1649	aws->callback(makeCreateWindowCallback(DI_create_cluster_detection_window, weighted_filter));
1650	aws->create_autosize_button("DETECT_CLUSTERS", "Detect homogenous clusters in tree", "D");
1651
1652	// matrix calculation
1653
1654	aws->button_length(18);
1655
1656	aws->at("calculate");
1657	aws->callback(makeWindowCallback(di_calculate_full_matrix_cb, weighted_filter));
1658	aws->create_button("CALC_FULL_MATRIX", "Calculate\nFull Matrix", "F");
1659
1660	aws->at("compress");
1661	aws->callback(makeWindowCallback(di_calculate_compressed_matrix_cb, weighted_filter));
1662	aws->create_button("CALC_COMPRESSED_MATRIX", "Calculate\nCompressed Matrix", "C");
1663
1664	recalculate_matrix_cb = new BoundWindowCallback(aws, makeWindowCallback(di_calculate_full_matrix_cb, weighted_filter));
1665
1666	aws->button_length(13);
1667
1668	{
1669	static save_matrix_params sparams;
1670
1671	sparams.awar_base = AWAR_DIST_SAVE_MATRIX_BASE;
1672	sparams.weighted_filter = weighted_filter;
1673
1674	aws->at("save_matrix");
1675	aws->callback(makeCreateWindowCallback(DI_create_save_matrix_window, &sparams));
1676	aws->create_button("SAVE_MATRIX", "Save matrix", "M");
1677
1678	aws->at("view_matrix");
1679	aws->callback(makeWindowCallback(di_view_matrix_cb, &sparams));
1680	aws->create_button("VIEW_MATRIX", "View matrix", "V");
1681	}
1682
1683	aws->button_length(22);
1684	aws->at("use_compr_tree");
1685	aws->callback(di_define_compression_tree_name_cb);
1686	aws->create_button("USE_COMPRESSION_TREE", "Use to compress", "");
1687	aws->at("use_sort_tree");
1688	aws->callback(di_define_sort_tree_name_cb);
1689	aws->create_button("USE_SORT_TREE", "Use to sort", "");
1690
1691	aws->at("compr_tree_name"); aws->create_input_field(AWAR_DIST_TREE_COMP_NAME, 12);
1692	aws->at("sort_tree_name"); aws->create_input_field(AWAR_DIST_TREE_SORT_NAME, 12);
1693
1694	// tree calculation
1695
1696	aws->button_length(18);
1697
1698	aws->at("t_calculate");
1699	aws->callback(makeWindowCallback(di_calculate_tree_cb, weighted_filter, false));
1700	aws->create_button("CALC_TREE", "Calculate \ntree", "C");
1701
1702	aws->at("bootstrap");
1703	aws->callback(makeWindowCallback(di_calculate_tree_cb, weighted_filter, true));
1704	aws->create_button("CALC_BOOTSTRAP_TREE", "Calculate \nbootstrap tree");
1705
1706	recalculate_tree_cb = new BoundWindowCallback(aws, makeWindowCallback(di_calculate_tree_cb, weighted_filter, false));
1707
1708	aws->button_length(22);
1709	aws->at("use_existing");
1710	aws->callback(di_define_save_tree_name_cb);
1711	aws->create_button("USE_NAME", "Use as new tree name", "");
1712
1713	aws->at("calc_tree_name");
1714	aws->create_input_field(AWAR_DIST_TREE_STD_NAME, 12);
1715
1716	aws->at("bcount");
1717	aws->create_input_field(AWAR_DIST_BOOTSTRAP_COUNT, 7);
1718
1719	{
1720	aws->sens_mask(AWM_EXP);
1721
1722	aws->at("auto_calc_tree");
1723	aws->label("Auto calculate tree");
1724	aws->create_toggle(AWAR_DIST_MATRIX_AUTO_CALC_TREE);
1725
1726	aws->at("auto_recalc");
1727	aws->label("Auto recalculate");
1728	aws->create_toggle(AWAR_DIST_MATRIX_AUTO_RECALC);
1729
1730	aws->sens_mask(AWM_ALL);
1731	}
1732
1733	bool disable_autocalc = !ARB_in_expert_mode(aw_root);
1734	if (disable_autocalc) {
1735	aw_root->awar(AWAR_DIST_MATRIX_AUTO_RECALC)->write_int(0);
1736	aw_root->awar(AWAR_DIST_MATRIX_AUTO_CALC_TREE)->write_int(0);
1737	}
1738
1739	GB_pop_transaction(GLOBAL_gb_main);
1740	return aws;
1741	}
1742
1743	// --------------------------------------------------------------------------------
1744
1745	#ifdef UNIT_TESTS
1746	#include <arb_diff.h>
1747	#include <arb_file.h>
1748
1749	#ifndef TEST_UNIT_H
1750	#include <test_unit.h>
1751	#endif
1752
1753	class DIST_testenv : virtual Noncopyable {
1754	GB_shell shell;
1755	GBDATA *gb_main;
1756	AliView *ali_view;
1757
1758	public:
1759	DIST_testenv(const char dbname, const char aliName)
1760	: ali_view(NULL)
1761	{
1762	gb_main = GB_open(dbname, "r");
1763	TEST_REJECT_NULL(gb_main);
1764
1765	GB_transaction ta(gb_main);
1766	size_t aliLength = GBT_get_alignment_len(gb_main, aliName);
1767
1768	AP_filter filter(aliLength);
1769	if (!filter.is_invalid()) {
1770	AP_weights weights(&filter);
1771	ali_view = new AliView(gb_main, filter, weights, aliName);
1772	}
1773	}
1774	~DIST_testenv() {
1775	delete ali_view;
1776	GB_close(gb_main);
1777	}
1778
1779	const AliView& aliview() const { return *ali_view; }
1780	GBDATA *gbmain() const { return gb_main; }
1781	};
1782
1783	void TEST_matrix() {
1784	for (int iat = GB_AT_RNA; iat<=GB_AT_AA; ++iat) {
1785	GB_alignment_type at = GB_alignment_type(iat);
1786	// ---------------
1787	// setup
1788	// GB_AT_RNA GB_AT_DNA GB_AT_AA
1789	const char *db_name[]= { NULL, "TEST_trees.arb", "TEST_realign.arb", "TEST_realign.arb", };
1790	const char *ali_name[]= { NULL, "ali_5s", "ali_dna", "ali_pro", };
1791
1792	TEST_ANNOTATE(GBS_global_string("ali_name=%s", ali_name[at]));
1793	DIST_testenv env(db_name[at], ali_name[at]);
1794
1795	DI_MATRIX matrix(env.aliview());
1796	MatrixOrder order(env.gbmain(), "tree_abc"); // no such tree!
1797	TEST_EXPECT_NO_ERROR(matrix.load(DI_LOAD_MARKED, order, true, NULL));
1798
1799	// -------------------------------
1800	// detect_transformation
1801	DI_TRANSFORMATION detected_trans;
1802	{
1803	string msg;
1804
1805	detected_trans = matrix.detect_transformation(msg);
1806	DI_TRANSFORMATION expected = DI_TRANSFORMATION_NONE_DETECTED;
1807	switch (at) {
1808	case GB_AT_RNA: expected = DI_TRANSFORMATION_NONE; break;
1809	case GB_AT_DNA: expected = DI_TRANSFORMATION_JUKES_CANTOR; break;
1810	case GB_AT_AA: expected = DI_TRANSFORMATION_PAM; break;
1811	case GB_AT_UNKNOWN: di_assert(0); break;
1812	}
1813	TEST_EXPECT_EQUAL(detected_trans, expected);
1814	}
1815
1816	// ------------------------------
1817	// calculate the matrix
1818
1819	// @@@ does not test user-defined transformation-matrix!
1820	if (at == GB_AT_AA) {
1821	matrix.calculate_pro(detected_trans, NULL);
1822	}
1823	else {
1824	if (at == GB_AT_RNA) detected_trans = DI_TRANSFORMATION_FELSENSTEIN; // force calculate_overall_freqs
1825	matrix.calculate("", detected_trans, NULL, NULL);
1826	}
1827
1828	// -----------------------------------
1829	// save in available formats
1830
1831	for (DI_SAVE_TYPE saveType = DI_SAVE_PHYLIP_COMP; saveType<=DI_SAVE_TABBED; saveType = DI_SAVE_TYPE(saveType+1)) {
1832	const char *savename = "distance/matrix.out";
1833	matrix.save(savename, saveType);
1834
1835	const char *suffixAT[] = { NULL, "rna", "dna", "pro" };
1836	const char *suffixST[] = { "phylipComp", "readable", "tabbed" };
1837	char *expected = GBS_global_string_copy("distance/matrix.%s.%s.expected", suffixAT[at], suffixST[saveType]);
1838
1839	// #define TEST_AUTO_UPDATE // uncomment to auto-update expected matrices
1840
1841	#if defined(TEST_AUTO_UPDATE)
1842	TEST_COPY_FILE(savename, expected);
1843	#else
1844	TEST_EXPECT_TEXTFILES_EQUAL(savename, expected);
1845	#endif // TEST_AUTO_UPDATE
1846	TEST_EXPECT_ZERO_OR_SHOW_ERRNO(GB_unlink(savename));
1847
1848	free(expected);
1849	}
1850	}
1851	}
1852
1853	#endif // UNIT_TESTS
1854
1855	// --------------------------------------------------------------------------------
1856

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

Download in other formats: