1 | // ================================================================ // |
---|
2 | // // |
---|
3 | // File : ED4_dots.cxx // |
---|
4 | // Purpose : Insert dots where bases may be missing // |
---|
5 | // // |
---|
6 | // Coded by Ralf Westram (coder@reallysoft.de) in December 2008 // |
---|
7 | // Institute of Microbiology (Technical University Munich) // |
---|
8 | // http://www.arb-home.de/ // |
---|
9 | // // |
---|
10 | // ================================================================ // |
---|
11 | |
---|
12 | #include "ed4_dots.hxx" |
---|
13 | #include "ed4_class.hxx" |
---|
14 | |
---|
15 | #include <awt_sel_boxes.hxx> |
---|
16 | #include <aw_awar.hxx> |
---|
17 | #include <aw_msg.hxx> |
---|
18 | #include <aw_root.hxx> |
---|
19 | #include <arbdbt.h> |
---|
20 | #include <cctype> |
---|
21 | |
---|
22 | using namespace std; |
---|
23 | |
---|
24 | #define AWAR_DOT_BASE "dotmiss/" |
---|
25 | #define AWAR_DOT_SAI AWAR_DOT_BASE "sainame" // selected SAI |
---|
26 | #define AWAR_DOT_SAI_CHARS AWAR_DOT_BASE "chars" // handle columns where SAI contains one of these chars |
---|
27 | #define AWAR_DOT_MARKED AWAR_DOT_BASE "marked" // handle marked only? |
---|
28 | |
---|
29 | struct dot_insert_stat { |
---|
30 | size_t pos_count; |
---|
31 | size_t *position; // contains 'pos_count' positions, where dots get inserted if sequence contains '-' |
---|
32 | |
---|
33 | bool marked_only; |
---|
34 | |
---|
35 | // statistics: |
---|
36 | size_t dots_inserted; |
---|
37 | size_t already_there; |
---|
38 | size_t sequences_checked; |
---|
39 | }; |
---|
40 | |
---|
41 | static ARB_ERROR dot_sequence_by_consensus(ED4_base *base, AW_CL cl_insert_stat) { |
---|
42 | ARB_ERROR error = 0; |
---|
43 | |
---|
44 | if (base->is_sequence_info_terminal()) { |
---|
45 | ED4_sequence_info_terminal *seq_term = base->to_sequence_info_terminal(); |
---|
46 | if (seq_term) { |
---|
47 | GBDATA *gb_ali = seq_term->data(); |
---|
48 | if (gb_ali) { |
---|
49 | GBDATA *gb_species = GB_get_grandfather(gb_ali); |
---|
50 | bool marked = GB_read_flag(gb_species); |
---|
51 | dot_insert_stat& stat = *reinterpret_cast<dot_insert_stat *>(cl_insert_stat); |
---|
52 | |
---|
53 | if (marked || !stat.marked_only) { |
---|
54 | char *sequence = GB_read_string(gb_ali); |
---|
55 | |
---|
56 | if (!sequence) { |
---|
57 | GB_ERROR err = GB_await_error(); |
---|
58 | error = GBS_global_string("No sequence found for '%s'\n(Reason: %s)", |
---|
59 | GBT_read_name(gb_species), err); |
---|
60 | } |
---|
61 | else { |
---|
62 | size_t length = GB_read_string_count(gb_ali); |
---|
63 | size_t old_dots_inserted = stat.dots_inserted; |
---|
64 | |
---|
65 | for (size_t p = 0; p<stat.pos_count; p++) { |
---|
66 | size_t pos = stat.position[p]; |
---|
67 | |
---|
68 | if (pos<length) { |
---|
69 | switch (sequence[pos]) { |
---|
70 | case '-': |
---|
71 | sequence[pos] = '.'; |
---|
72 | stat.dots_inserted++; |
---|
73 | break; |
---|
74 | |
---|
75 | case '.': |
---|
76 | stat.already_there++; |
---|
77 | break; |
---|
78 | |
---|
79 | default: |
---|
80 | break; |
---|
81 | } |
---|
82 | } |
---|
83 | } |
---|
84 | |
---|
85 | if (stat.dots_inserted > old_dots_inserted) { // did sequence change? |
---|
86 | error = GB_write_string(gb_ali, sequence); |
---|
87 | } |
---|
88 | |
---|
89 | free(sequence); |
---|
90 | stat.sequences_checked++; |
---|
91 | } |
---|
92 | } |
---|
93 | } |
---|
94 | } |
---|
95 | } |
---|
96 | |
---|
97 | return error; |
---|
98 | } |
---|
99 | |
---|
100 | static void dot_missing_bases(AW_window *aww) { |
---|
101 | ED4_MostRecentWinContext context; |
---|
102 | |
---|
103 | ED4_cursor *cursor = ¤t_cursor(); |
---|
104 | ARB_ERROR error = 0; |
---|
105 | |
---|
106 | if (!cursor->in_consensus_terminal()) { |
---|
107 | error = "No consensus selected"; |
---|
108 | } |
---|
109 | else { |
---|
110 | AW_root *aw_root = aww->get_root(); |
---|
111 | |
---|
112 | dot_insert_stat stat; |
---|
113 | stat.dots_inserted = 0; |
---|
114 | stat.already_there = 0; |
---|
115 | stat.position = 0; |
---|
116 | stat.sequences_checked = 0; |
---|
117 | stat.marked_only = aw_root->awar(AWAR_DOT_MARKED)->read_int(); |
---|
118 | |
---|
119 | ED4_group_manager *group_manager = cursor->owner_of_cursor->get_parent(ED4_L_GROUP)->to_group_manager(); |
---|
120 | { |
---|
121 | // build list of positions where consensus contains upper case characters: |
---|
122 | char *consensus = group_manager->table().build_consensus_string(); |
---|
123 | for (int pass = 1; pass <= 2; pass++) { |
---|
124 | stat.pos_count = 0; |
---|
125 | for (int pos = 0; consensus[pos]; pos++) { |
---|
126 | if (isupper(consensus[pos])) { |
---|
127 | if (pass == 2) stat.position[stat.pos_count] = pos; |
---|
128 | stat.pos_count++; |
---|
129 | } |
---|
130 | } |
---|
131 | |
---|
132 | if (pass == 1) stat.position = (size_t*)malloc(stat.pos_count * sizeof(*stat.position)); |
---|
133 | } |
---|
134 | free(consensus); |
---|
135 | } |
---|
136 | |
---|
137 | if (!stat.pos_count) { |
---|
138 | error = "No consensus column contains upper case characters"; |
---|
139 | } |
---|
140 | else { |
---|
141 | // if SAI is selected, reduce list of affected positions |
---|
142 | char *sai = 0; |
---|
143 | size_t sai_len = -1; |
---|
144 | { |
---|
145 | GB_transaction ta(GLOBAL_gb_main); |
---|
146 | char *sai_name = aw_root->awar(AWAR_DOT_SAI)->read_string(); |
---|
147 | |
---|
148 | if (sai_name && sai_name[0]) { |
---|
149 | GBDATA *gb_sai = GBT_expect_SAI(GLOBAL_gb_main, sai_name); |
---|
150 | if (!gb_sai) error = GB_await_error(); |
---|
151 | else { |
---|
152 | GBDATA *gb_ali = GBT_find_sequence(gb_sai, ED4_ROOT->alignment_name); |
---|
153 | if (!gb_ali) { |
---|
154 | error = GBS_global_string("SAI '%s' has no data in '%s'", sai_name, ED4_ROOT->alignment_name); |
---|
155 | } |
---|
156 | else { |
---|
157 | sai = GB_read_string(gb_ali); |
---|
158 | sai_len = GB_read_string_count(gb_ali); |
---|
159 | } |
---|
160 | } |
---|
161 | } |
---|
162 | free(sai_name); |
---|
163 | error = ta.close(error); |
---|
164 | } |
---|
165 | |
---|
166 | if (sai) { // SAI is selected |
---|
167 | if (!error) { |
---|
168 | char *sai_chars = aw_root->awar(AWAR_DOT_SAI_CHARS)->read_string(); |
---|
169 | if (sai_chars[0] == 0) error = "No SAI characters given -> no column selectable"; |
---|
170 | else { |
---|
171 | size_t k = 0; |
---|
172 | size_t p; |
---|
173 | for (p = 0; p<stat.pos_count && stat.position[p]<sai_len; p++) { |
---|
174 | size_t pos = stat.position[p]; |
---|
175 | if (strchr(sai_chars, sai[pos]) != NULL) { // SAI contains one of the 'sai_chars' |
---|
176 | stat.position[k++] = pos; // use current position |
---|
177 | } |
---|
178 | } |
---|
179 | stat.pos_count = k; |
---|
180 | |
---|
181 | if (!stat.pos_count) error = "SAI selects other columns than consensus. Nothing to do."; |
---|
182 | } |
---|
183 | free(sai_chars); |
---|
184 | } |
---|
185 | free(sai); |
---|
186 | } |
---|
187 | } |
---|
188 | |
---|
189 | if (!error) { |
---|
190 | e4_assert(stat.pos_count); |
---|
191 | GB_transaction ta(GLOBAL_gb_main); |
---|
192 | error = group_manager->route_down_hierarchy(dot_sequence_by_consensus, (AW_CL)&stat); |
---|
193 | |
---|
194 | if (stat.sequences_checked == 0 && !error) { |
---|
195 | error = GBS_global_string("Group contains no %ssequences", stat.marked_only ? "marked " : ""); |
---|
196 | } |
---|
197 | |
---|
198 | if (!error) { |
---|
199 | const char *present = ""; |
---|
200 | if (stat.already_there) { |
---|
201 | present = GBS_global_string("Dots already present: %zu ", stat.already_there); |
---|
202 | } |
---|
203 | |
---|
204 | const char *changed = stat.dots_inserted |
---|
205 | ? GBS_global_string("Gaps changed into dots: %zu", stat.dots_inserted) |
---|
206 | : "No gaps were changed into dots."; |
---|
207 | |
---|
208 | aw_message(GBS_global_string("%s%s", present, changed)); |
---|
209 | } |
---|
210 | |
---|
211 | error = ta.close(error); |
---|
212 | } |
---|
213 | } |
---|
214 | |
---|
215 | aw_message_if(error); |
---|
216 | } |
---|
217 | |
---|
218 | void ED4_create_dot_missing_bases_awars(AW_root *aw_root, AW_default aw_def) { |
---|
219 | aw_root->awar_int(AWAR_DOT_MARKED, 0, aw_def); |
---|
220 | aw_root->awar_string(AWAR_DOT_SAI, "", aw_def); |
---|
221 | aw_root->awar_string(AWAR_DOT_SAI_CHARS, "", aw_def); |
---|
222 | } |
---|
223 | |
---|
224 | void ED4_popup_dot_missing_bases_window(AW_window *editor_window, AW_CL, AW_CL) { |
---|
225 | AW_root *aw_root = editor_window->get_root(); |
---|
226 | static AW_window_simple *aws = 0; |
---|
227 | |
---|
228 | ED4_LocalWinContext uses(editor_window); |
---|
229 | |
---|
230 | if (!aws) { |
---|
231 | aws = new AW_window_simple; |
---|
232 | |
---|
233 | aws->init(aw_root, "DOT_MISS_BASES", "Dot potentially missing bases"); |
---|
234 | aws->load_xfig("edit4/missbase.fig"); |
---|
235 | |
---|
236 | aws->button_length(10); |
---|
237 | |
---|
238 | aws->at("close"); |
---|
239 | aws->callback(AW_POPDOWN); |
---|
240 | aws->create_button("CLOSE", "CLOSE", "C"); |
---|
241 | |
---|
242 | aws->at("help"); |
---|
243 | aws->callback(makeHelpCallback("missbase.hlp")); |
---|
244 | aws->create_button("HELP", "HELP", "H"); |
---|
245 | |
---|
246 | aws->at("marked"); |
---|
247 | aws->label("Marked species only"); |
---|
248 | aws->create_toggle(AWAR_DOT_MARKED); |
---|
249 | |
---|
250 | aws->at("SAI"); |
---|
251 | aws->button_length(30); |
---|
252 | awt_create_SAI_selection_button(GLOBAL_gb_main, aws, AWAR_DOT_SAI); |
---|
253 | |
---|
254 | aws->at("SAI_chars"); |
---|
255 | aws->label("contains one of"); |
---|
256 | aws->create_input_field(AWAR_DOT_SAI_CHARS, 20); |
---|
257 | |
---|
258 | aws->button_length(10); |
---|
259 | |
---|
260 | aws->at("cons_def"); |
---|
261 | aws->label("Change definition of"); |
---|
262 | aws->callback(ED4_create_consensus_definition_window); |
---|
263 | aws->create_button("CONS_DEF", "Consensus", "C"); |
---|
264 | |
---|
265 | aws->at("go"); |
---|
266 | aws->callback(dot_missing_bases); |
---|
267 | aws->create_button("GO", "GO", "G"); |
---|
268 | } |
---|
269 | |
---|
270 | e4_assert(aws); |
---|
271 | |
---|
272 | aws->activate(); |
---|
273 | } |
---|
274 | |
---|