| 1 | // =============================================================== // |
|---|
| 2 | // // |
|---|
| 3 | // File : ad_transpro.cxx // |
|---|
| 4 | // Purpose : // |
|---|
| 5 | // // |
|---|
| 6 | // Institute of Microbiology (Technical University Munich) // |
|---|
| 7 | // http://www.arb-home.de/ // |
|---|
| 8 | // // |
|---|
| 9 | // =============================================================== // |
|---|
| 10 | |
|---|
| 11 | #include "NT_local.h" |
|---|
| 12 | |
|---|
| 13 | #include <awt_sel_boxes.hxx> |
|---|
| 14 | #include <Translate.hxx> |
|---|
| 15 | #include <AP_codon_table.hxx> |
|---|
| 16 | #include <AP_pro_a_nucs.hxx> |
|---|
| 17 | #include <aw_awars.hxx> |
|---|
| 18 | #include <aw_root.hxx> |
|---|
| 19 | #include <aw_question.hxx> |
|---|
| 20 | #include <aw_msg.hxx> |
|---|
| 21 | #include <arb_progress.h> |
|---|
| 22 | #include <arbdbt.h> |
|---|
| 23 | #include <cctype> |
|---|
| 24 | #include <arb_defs.h> |
|---|
| 25 | |
|---|
| 26 | static GB_ERROR arb_r2a(GBDATA *gb_main, bool use_entries, bool save_entries, int selected_startpos, |
|---|
| 27 | bool translate_all, const char *ali_source, const char *ali_dest) |
|---|
| 28 | { |
|---|
| 29 | // if use_entries == true -> use fields 'codon_start' and 'transl_table' for translation |
|---|
| 30 | // (selected_startpos and AWAR_PROTEIN_TYPE are only used both fields are missing, |
|---|
| 31 | // if only one is missing, now an error occurs) |
|---|
| 32 | // if use_entries == false -> always use selected_startpos and AWAR_PROTEIN_TYPE |
|---|
| 33 | // if translate_all == true -> a selected_startpos > 1 produces a leading 'X' in protein data |
|---|
| 34 | // (otherwise nucleotides in front of the starting pos are simply ignored) |
|---|
| 35 | |
|---|
| 36 | nt_assert(selected_startpos >= 0 && selected_startpos < 3); |
|---|
| 37 | |
|---|
| 38 | GB_ERROR error = 0; |
|---|
| 39 | char *to_free = 0; |
|---|
| 40 | |
|---|
| 41 | // check/create alignments |
|---|
| 42 | { |
|---|
| 43 | GBDATA *gb_source = GBT_get_alignment(gb_main, ali_source); |
|---|
| 44 | if (!gb_source) { |
|---|
| 45 | error = GBS_global_string("No valid source alignment (%s)", GB_await_error()); |
|---|
| 46 | } |
|---|
| 47 | else { |
|---|
| 48 | GBDATA *gb_dest = GBT_get_alignment(gb_main, ali_dest); |
|---|
| 49 | if (!gb_dest) { |
|---|
| 50 | GB_clear_error(); |
|---|
| 51 | const char *msg = GBS_global_string("You have not selected a destination alignment\n" |
|---|
| 52 | "Shall I create one ('%s_pro') for you?", ali_source); |
|---|
| 53 | if (!aw_ask_sure("create_protein_ali", msg)) { |
|---|
| 54 | error = "Cancelled by user"; |
|---|
| 55 | } |
|---|
| 56 | else { |
|---|
| 57 | long slen = GBT_get_alignment_len(gb_main, ali_source); |
|---|
| 58 | to_free = GBS_global_string_copy("%s_pro", ali_source); |
|---|
| 59 | ali_dest = to_free; |
|---|
| 60 | gb_dest = GBT_create_alignment(gb_main, ali_dest, slen/3+1, 0, 1, "ami"); |
|---|
| 61 | |
|---|
| 62 | if (!gb_dest) error = GB_await_error(); |
|---|
| 63 | else { |
|---|
| 64 | char *fname = GBS_global_string_copy("%s/data", ali_dest); |
|---|
| 65 | error = GBT_add_new_changekey(gb_main, fname, GB_STRING); |
|---|
| 66 | free(fname); |
|---|
| 67 | } |
|---|
| 68 | } |
|---|
| 69 | } |
|---|
| 70 | } |
|---|
| 71 | } |
|---|
| 72 | |
|---|
| 73 | int no_data = 0; // count species w/o data |
|---|
| 74 | int spec_no_transl_info = 0; // counts species w/o or with illegal transl_table and/or codon_start |
|---|
| 75 | int count = 0; // count translated species |
|---|
| 76 | int stops = 0; // count overall stop codons |
|---|
| 77 | int selected_ttable = -1; |
|---|
| 78 | |
|---|
| 79 | if (!error) { |
|---|
| 80 | arb_progress progress("Translating", GBT_count_marked_species(gb_main)); |
|---|
| 81 | |
|---|
| 82 | bool table_used[AWT_CODON_TABLES]; |
|---|
| 83 | memset(table_used, 0, sizeof(table_used)); |
|---|
| 84 | selected_ttable = *GBT_read_int(gb_main, AWAR_PROTEIN_TYPE); // read selected table |
|---|
| 85 | |
|---|
| 86 | if (use_entries) { |
|---|
| 87 | for (GBDATA *gb_species = GBT_first_marked_species(gb_main); |
|---|
| 88 | gb_species && !error; |
|---|
| 89 | gb_species = GBT_next_marked_species(gb_species)) |
|---|
| 90 | { |
|---|
| 91 | int arb_table, codon_start; |
|---|
| 92 | error = AWT_getTranslationInfo(gb_species, arb_table, codon_start); |
|---|
| 93 | |
|---|
| 94 | if (!error) { |
|---|
| 95 | if (arb_table == -1) arb_table = selected_ttable; // no transl_table entry -> default to selected standard code |
|---|
| 96 | table_used[arb_table] = true; |
|---|
| 97 | } |
|---|
| 98 | } |
|---|
| 99 | } |
|---|
| 100 | else { |
|---|
| 101 | table_used[selected_ttable] = true; // and mark it used |
|---|
| 102 | } |
|---|
| 103 | |
|---|
| 104 | for (int table = 0; table<AWT_CODON_TABLES && !error; ++table) { |
|---|
| 105 | if (!table_used[table]) continue; |
|---|
| 106 | |
|---|
| 107 | for (GBDATA *gb_species = GBT_first_marked_species(gb_main); |
|---|
| 108 | gb_species && !error; |
|---|
| 109 | gb_species = GBT_next_marked_species(gb_species)) |
|---|
| 110 | { |
|---|
| 111 | bool found_transl_info = false; |
|---|
| 112 | int startpos = selected_startpos; |
|---|
| 113 | |
|---|
| 114 | if (use_entries) { // if entries are used, test if field 'transl_table' matches current table |
|---|
| 115 | int sp_arb_table, sp_codon_start; |
|---|
| 116 | |
|---|
| 117 | error = AWT_getTranslationInfo(gb_species, sp_arb_table, sp_codon_start); |
|---|
| 118 | |
|---|
| 119 | nt_assert(!error); // should already have been handled after first call to AWT_getTranslationInfo above |
|---|
| 120 | |
|---|
| 121 | if (sp_arb_table == -1) { // no table in DB |
|---|
| 122 | nt_assert(sp_codon_start == -1); // either both should be defined or none |
|---|
| 123 | sp_arb_table = selected_ttable; // use selected translation table as default (if 'transl_table' field is missing) |
|---|
| 124 | sp_codon_start = selected_startpos; // use selected codon startpos (if 'codon_start' field is missing) |
|---|
| 125 | } |
|---|
| 126 | else { |
|---|
| 127 | nt_assert(sp_codon_start != -1); // either both should be defined or none |
|---|
| 128 | found_transl_info = true; |
|---|
| 129 | } |
|---|
| 130 | |
|---|
| 131 | if (sp_arb_table != table) continue; // species has not current transl_table |
|---|
| 132 | |
|---|
| 133 | startpos = sp_codon_start; |
|---|
| 134 | } |
|---|
| 135 | |
|---|
| 136 | GBDATA *gb_source = GB_entry(gb_species, ali_source); |
|---|
| 137 | if (!gb_source) { ++no_data; } |
|---|
| 138 | else { |
|---|
| 139 | GBDATA *gb_source_data = GB_entry(gb_source, "data"); |
|---|
| 140 | if (!gb_source_data) { ++no_data; } |
|---|
| 141 | else { |
|---|
| 142 | char *data = GB_read_string(gb_source_data); |
|---|
| 143 | if (!data) { |
|---|
| 144 | GB_print_error(); // cannot read data (ignore species) |
|---|
| 145 | ++no_data; |
|---|
| 146 | } |
|---|
| 147 | else { |
|---|
| 148 | if (!found_transl_info) ++spec_no_transl_info; // count species with missing info |
|---|
| 149 | |
|---|
| 150 | stops += AWT_pro_a_nucs_convert(table, data, GB_read_string_count(gb_source_data), startpos, translate_all, false, false, 0); // do the translation |
|---|
| 151 | ++count; |
|---|
| 152 | |
|---|
| 153 | GBDATA *gb_dest_data = GBT_add_data(gb_species, ali_dest, "data", GB_STRING); |
|---|
| 154 | if (!gb_dest_data) error = GB_await_error(); |
|---|
| 155 | else error = GB_write_string(gb_dest_data, data); |
|---|
| 156 | |
|---|
| 157 | |
|---|
| 158 | if (!error && save_entries && !found_transl_info) { |
|---|
| 159 | error = AWT_saveTranslationInfo(gb_species, selected_ttable, startpos); |
|---|
| 160 | } |
|---|
| 161 | |
|---|
| 162 | free(data); |
|---|
| 163 | } |
|---|
| 164 | } |
|---|
| 165 | } |
|---|
| 166 | progress.inc_and_check_user_abort(error); |
|---|
| 167 | } |
|---|
| 168 | } |
|---|
| 169 | } |
|---|
| 170 | |
|---|
| 171 | if (!error) { |
|---|
| 172 | if (use_entries) { // use 'transl_table' and 'codon_start' fields ? |
|---|
| 173 | if (spec_no_transl_info) { |
|---|
| 174 | int embl_transl_table = AWT_arb_code_nr_2_embl_transl_table(selected_ttable); |
|---|
| 175 | aw_message(GBS_global_string("%i taxa had no valid translation info (fields 'transl_table' and 'codon_start')\n" |
|---|
| 176 | "Defaults (%i and %i) have been used%s.", |
|---|
| 177 | spec_no_transl_info, |
|---|
| 178 | embl_transl_table, selected_startpos+1, |
|---|
| 179 | save_entries ? " and written to DB entries" : "")); |
|---|
| 180 | } |
|---|
| 181 | else { // all entries were present |
|---|
| 182 | aw_message("codon_start and transl_table entries were found for all translated taxa"); |
|---|
| 183 | } |
|---|
| 184 | } |
|---|
| 185 | |
|---|
| 186 | if (no_data>0) { |
|---|
| 187 | aw_message(GBS_global_string("%i taxa had no data in '%s'", no_data, ali_source)); |
|---|
| 188 | } |
|---|
| 189 | if ((count+no_data) == 0) { |
|---|
| 190 | aw_message("Please mark species to translate"); |
|---|
| 191 | } |
|---|
| 192 | else { |
|---|
| 193 | aw_message(GBS_global_string("%i taxa converted\n %f stops per sequence found", |
|---|
| 194 | count, (double)stops/(double)count)); |
|---|
| 195 | } |
|---|
| 196 | } |
|---|
| 197 | |
|---|
| 198 | free(to_free); |
|---|
| 199 | |
|---|
| 200 | return error; |
|---|
| 201 | } |
|---|
| 202 | |
|---|
| 203 | #define AWAR_TRANSPRO_PREFIX "transpro/" |
|---|
| 204 | #define AWAR_TRANSPRO_SOURCE AWAR_TRANSPRO_PREFIX "source" |
|---|
| 205 | #define AWAR_TRANSPRO_DEST AWAR_TRANSPRO_PREFIX "dest" |
|---|
| 206 | #define AWAR_TRANSPRO_POS AWAR_TRANSPRO_PREFIX "pos" // [0..N-1] |
|---|
| 207 | #define AWAR_TRANSPRO_MODE AWAR_TRANSPRO_PREFIX "mode" |
|---|
| 208 | #define AWAR_TRANSPRO_XSTART AWAR_TRANSPRO_PREFIX "xstart" |
|---|
| 209 | #define AWAR_TRANSPRO_WRITE AWAR_TRANSPRO_PREFIX "write" |
|---|
| 210 | |
|---|
| 211 | static void transpro_event(AW_window *aww) { |
|---|
| 212 | GB_ERROR error = GB_begin_transaction(GLOBAL.gb_main); |
|---|
| 213 | if (!error) { |
|---|
| 214 | #if defined(DEBUG) && 0 |
|---|
| 215 | test_AWT_get_codons(); |
|---|
| 216 | #endif |
|---|
| 217 | AW_root *aw_root = aww->get_root(); |
|---|
| 218 | char *ali_source = aw_root->awar(AWAR_TRANSPRO_SOURCE)->read_string(); |
|---|
| 219 | char *ali_dest = aw_root->awar(AWAR_TRANSPRO_DEST)->read_string(); |
|---|
| 220 | char *mode = aw_root->awar(AWAR_TRANSPRO_MODE)->read_string(); |
|---|
| 221 | int startpos = aw_root->awar(AWAR_TRANSPRO_POS)->read_int(); |
|---|
| 222 | bool save2fields = aw_root->awar(AWAR_TRANSPRO_WRITE)->read_int(); |
|---|
| 223 | bool translate_all = aw_root->awar(AWAR_TRANSPRO_XSTART)->read_int(); |
|---|
| 224 | |
|---|
| 225 | error = arb_r2a(GLOBAL.gb_main, strcmp(mode, "fields") == 0, save2fields, startpos, translate_all, ali_source, ali_dest); |
|---|
| 226 | if (!error) error = GBT_check_data(GLOBAL.gb_main, 0); |
|---|
| 227 | |
|---|
| 228 | free(mode); |
|---|
| 229 | free(ali_dest); |
|---|
| 230 | free(ali_source); |
|---|
| 231 | } |
|---|
| 232 | GB_end_transaction_show_error(GLOBAL.gb_main, error, aw_message); |
|---|
| 233 | } |
|---|
| 234 | |
|---|
| 235 | static void nt_trans_cursorpos_changed(AW_root *awr) { |
|---|
| 236 | int pos = bio2info(awr->awar(AWAR_CURSOR_POSITION)->read_int()); |
|---|
| 237 | pos = pos % 3; |
|---|
| 238 | awr->awar(AWAR_TRANSPRO_POS)->write_int(pos); |
|---|
| 239 | } |
|---|
| 240 | |
|---|
| 241 | AW_window *NT_create_dna_2_pro_window(AW_root *root) { |
|---|
| 242 | GB_transaction ta(GLOBAL.gb_main); |
|---|
| 243 | |
|---|
| 244 | AW_window_simple *aws = new AW_window_simple; |
|---|
| 245 | aws->init(root, "TRANSLATE_DNA_TO_PRO", "TRANSLATE DNA TO PRO"); |
|---|
| 246 | |
|---|
| 247 | aws->load_xfig("transpro.fig"); |
|---|
| 248 | |
|---|
| 249 | aws->callback((AW_CB0)AW_POPDOWN); |
|---|
| 250 | aws->at("close"); |
|---|
| 251 | aws->create_button("CLOSE", "CLOSE", "C"); |
|---|
| 252 | |
|---|
| 253 | aws->callback(makeHelpCallback("translate_dna_2_pro.hlp")); |
|---|
| 254 | aws->at("help"); |
|---|
| 255 | aws->create_button("HELP", "HELP", "H"); |
|---|
| 256 | |
|---|
| 257 | aws->at("source"); |
|---|
| 258 | awt_create_selection_list_on_alignments(GLOBAL.gb_main, (AW_window *)aws, AWAR_TRANSPRO_SOURCE, "dna=:rna="); |
|---|
| 259 | |
|---|
| 260 | aws->at("dest"); |
|---|
| 261 | awt_create_selection_list_on_alignments(GLOBAL.gb_main, (AW_window *)aws, AWAR_TRANSPRO_DEST, "pro=:ami="); |
|---|
| 262 | |
|---|
| 263 | root->awar_int(AWAR_PROTEIN_TYPE, AWAR_PROTEIN_TYPE_bacterial_code_index, GLOBAL.gb_main); |
|---|
| 264 | aws->at("table"); |
|---|
| 265 | aws->create_option_menu(AWAR_PROTEIN_TYPE, true); |
|---|
| 266 | for (int code_nr=0; code_nr<AWT_CODON_TABLES; code_nr++) { |
|---|
| 267 | aws->insert_option(AWT_get_codon_code_name(code_nr), "", code_nr); |
|---|
| 268 | } |
|---|
| 269 | aws->update_option_menu(); |
|---|
| 270 | |
|---|
| 271 | aws->at("mode"); |
|---|
| 272 | aws->create_toggle_field(AWAR_TRANSPRO_MODE, 0, ""); |
|---|
| 273 | aws->insert_toggle("from fields 'codon_start' and 'transl_table'", "", "fields"); |
|---|
| 274 | aws->insert_default_toggle("use settings below (same for all species):", "", "settings"); |
|---|
| 275 | aws->update_toggle_field(); |
|---|
| 276 | |
|---|
| 277 | aws->at("pos"); |
|---|
| 278 | aws->create_option_menu(AWAR_TRANSPRO_POS, true); |
|---|
| 279 | for (int p = 1; p <= 3; ++p) { |
|---|
| 280 | char label[2] = { char(p+'0'), 0 }; |
|---|
| 281 | aws->insert_option(label, label, bio2info(p)); |
|---|
| 282 | } |
|---|
| 283 | aws->update_option_menu(); |
|---|
| 284 | aws->get_root()->awar_int(AWAR_CURSOR_POSITION)->add_callback(nt_trans_cursorpos_changed); |
|---|
| 285 | |
|---|
| 286 | aws->at("write"); |
|---|
| 287 | aws->label("Save settings (to 'codon_start'+'transl_table')"); |
|---|
| 288 | aws->create_toggle(AWAR_TRANSPRO_WRITE); |
|---|
| 289 | |
|---|
| 290 | aws->at("start"); |
|---|
| 291 | aws->label("Translate all data"); |
|---|
| 292 | aws->create_toggle(AWAR_TRANSPRO_XSTART); |
|---|
| 293 | |
|---|
| 294 | aws->at("translate"); |
|---|
| 295 | aws->callback(transpro_event); |
|---|
| 296 | aws->highlight(); |
|---|
| 297 | aws->create_button("TRANSLATE", "TRANSLATE", "T"); |
|---|
| 298 | |
|---|
| 299 | aws->window_fit(); |
|---|
| 300 | |
|---|
| 301 | return aws; |
|---|
| 302 | } |
|---|
| 303 | |
|---|
| 304 | // Realign a dna alignment with a given protein source |
|---|
| 305 | |
|---|
| 306 | static int synchronizeCodons(const char *proteins, const char *dna, int minCatchUp, int maxCatchUp, int *foundCatchUp, |
|---|
| 307 | const AWT_allowedCode& initially_allowed_code, AWT_allowedCode& allowed_code_left) { |
|---|
| 308 | |
|---|
| 309 | for (int catchUp=minCatchUp; catchUp<=maxCatchUp; catchUp++) { |
|---|
| 310 | const char *dna_start = dna+catchUp; |
|---|
| 311 | AWT_allowedCode allowed_code; |
|---|
| 312 | allowed_code = initially_allowed_code; |
|---|
| 313 | |
|---|
| 314 | for (int p=0; ; p++) { |
|---|
| 315 | char prot = proteins[p]; |
|---|
| 316 | |
|---|
| 317 | if (!prot) { // all proteins were synchronized |
|---|
| 318 | *foundCatchUp = catchUp; |
|---|
| 319 | return 1; |
|---|
| 320 | } |
|---|
| 321 | |
|---|
| 322 | if (!AWT_is_codon(prot, dna_start, allowed_code, allowed_code_left)) break; |
|---|
| 323 | |
|---|
| 324 | allowed_code = allowed_code_left; // if synchronized: use left codes as allowed codes! |
|---|
| 325 | dna_start += 3; |
|---|
| 326 | } |
|---|
| 327 | } |
|---|
| 328 | |
|---|
| 329 | return 0; |
|---|
| 330 | } |
|---|
| 331 | |
|---|
| 332 | #define SYNC_LENGTH 4 |
|---|
| 333 | // every X in amino-alignment, it represents 1 to 3 bases in DNA-Alignment |
|---|
| 334 | // SYNC_LENGTH is the # of codons which will be synchronized (ahead!) |
|---|
| 335 | // before deciding "X was realigned correctly" |
|---|
| 336 | |
|---|
| 337 | static GB_ERROR arb_transdna(GBDATA *gb_main, char *ali_source, char *ali_dest, long *neededLength) { |
|---|
| 338 | AP_initialize_codon_tables(); |
|---|
| 339 | |
|---|
| 340 | GBDATA *gb_source = GBT_get_alignment(gb_main, ali_source); if (!gb_source) return "Please select a valid source alignment"; |
|---|
| 341 | GBDATA *gb_dest = GBT_get_alignment(gb_main, ali_dest); if (!gb_dest) return "Please select a valid destination alignment"; |
|---|
| 342 | |
|---|
| 343 | long ali_len = GBT_get_alignment_len(gb_main, ali_dest); |
|---|
| 344 | long max_wanted_ali_len = 0; |
|---|
| 345 | GB_ERROR error = 0; |
|---|
| 346 | |
|---|
| 347 | long no_of_marked_species = GBT_count_marked_species(gb_main); |
|---|
| 348 | long no_of_realigned_species = 0; |
|---|
| 349 | |
|---|
| 350 | arb_progress progress("Re-aligner", no_of_marked_species); |
|---|
| 351 | progress.auto_subtitles("Re-aligning species"); |
|---|
| 352 | |
|---|
| 353 | int ignore_fail_pos = 0; |
|---|
| 354 | |
|---|
| 355 | for (GBDATA *gb_species = GBT_first_marked_species(gb_main); |
|---|
| 356 | !error && gb_species; |
|---|
| 357 | gb_species = GBT_next_marked_species(gb_species)) |
|---|
| 358 | { |
|---|
| 359 | gb_source = GB_entry(gb_species, ali_source); if (!gb_source) continue; |
|---|
| 360 | GBDATA *gb_source_data = GB_entry(gb_source, "data"); if (!gb_source_data) continue; |
|---|
| 361 | gb_dest = GB_entry(gb_species, ali_dest); if (!gb_dest) continue; |
|---|
| 362 | GBDATA *gb_dest_data = GB_entry(gb_dest, "data"); if (!gb_dest_data) continue; |
|---|
| 363 | |
|---|
| 364 | char *source = GB_read_string(gb_source_data); if (!source) { GB_print_error(); continue; } |
|---|
| 365 | char *dest = GB_read_string(gb_dest_data); if (!dest) { GB_print_error(); continue; } |
|---|
| 366 | |
|---|
| 367 | long source_len = GB_read_string_count(gb_source_data); |
|---|
| 368 | long dest_len = GB_read_string_count(gb_dest_data); |
|---|
| 369 | |
|---|
| 370 | // compress destination DNA (=remove align-characters): |
|---|
| 371 | char *compressed_dest = (char*)malloc(dest_len+1); |
|---|
| 372 | { |
|---|
| 373 | char *f = dest; |
|---|
| 374 | char *t = compressed_dest; |
|---|
| 375 | |
|---|
| 376 | while (1) { |
|---|
| 377 | char c = *f++; |
|---|
| 378 | if (!c) break; |
|---|
| 379 | if (c!='.' && c!='-') *t++ = c; |
|---|
| 380 | } |
|---|
| 381 | *t = 0; |
|---|
| 382 | } |
|---|
| 383 | |
|---|
| 384 | int failed = 0; |
|---|
| 385 | const char *fail_reason = 0; |
|---|
| 386 | |
|---|
| 387 | long wanted_ali_len = source_len*3L; |
|---|
| 388 | char *buffer = (char*)malloc(ali_len+1); |
|---|
| 389 | if (ali_len<wanted_ali_len) { |
|---|
| 390 | failed = 1; |
|---|
| 391 | fail_reason = GBS_global_string("Alignment '%s' is too short (increase its length to %li)", ali_dest, wanted_ali_len); |
|---|
| 392 | ignore_fail_pos = 1; |
|---|
| 393 | |
|---|
| 394 | if (wanted_ali_len>max_wanted_ali_len) max_wanted_ali_len = wanted_ali_len; |
|---|
| 395 | } |
|---|
| 396 | |
|---|
| 397 | AWT_allowedCode allowed_code; // default: all allowed |
|---|
| 398 | |
|---|
| 399 | if (!failed) { |
|---|
| 400 | int arb_transl_table, codon_start; |
|---|
| 401 | GB_ERROR local_error = AWT_getTranslationInfo(gb_species, arb_transl_table, codon_start); |
|---|
| 402 | if (local_error) { |
|---|
| 403 | failed = 1; |
|---|
| 404 | fail_reason = GBS_global_string("Error while reading 'transl_table' (%s)", local_error); |
|---|
| 405 | ignore_fail_pos = 1; |
|---|
| 406 | } |
|---|
| 407 | else if (arb_transl_table >= 0) { |
|---|
| 408 | // we found a 'transl_table' entry -> restrict used code to the code stored there |
|---|
| 409 | allowed_code.forbidAllBut(arb_transl_table); |
|---|
| 410 | } |
|---|
| 411 | } |
|---|
| 412 | |
|---|
| 413 | char *d = compressed_dest; |
|---|
| 414 | char *s = source; |
|---|
| 415 | |
|---|
| 416 | if (!failed) { |
|---|
| 417 | char *p = buffer; |
|---|
| 418 | int x_count = 0; |
|---|
| 419 | const char *x_start = 0; |
|---|
| 420 | |
|---|
| 421 | for (;;) { |
|---|
| 422 | char c = *s++; |
|---|
| 423 | if (!c) { |
|---|
| 424 | if (x_count) { |
|---|
| 425 | int off = -(x_count*3); |
|---|
| 426 | while (d[0]) { |
|---|
| 427 | p[off++] = *d++; |
|---|
| 428 | } |
|---|
| 429 | } |
|---|
| 430 | break; |
|---|
| 431 | } |
|---|
| 432 | |
|---|
| 433 | if (c=='.' || c=='-') { |
|---|
| 434 | p[0] = p[1] = p[2] = c; |
|---|
| 435 | p += 3; |
|---|
| 436 | } |
|---|
| 437 | else if (toupper(c)=='X') { // one X represents 1 to 3 DNAs |
|---|
| 438 | x_start = s-1; |
|---|
| 439 | x_count = 1; |
|---|
| 440 | int gap_count = 0; |
|---|
| 441 | |
|---|
| 442 | for (;;) { |
|---|
| 443 | char c2 = toupper(s[0]); |
|---|
| 444 | |
|---|
| 445 | if (c2=='X') { |
|---|
| 446 | x_count++; |
|---|
| 447 | } |
|---|
| 448 | else { |
|---|
| 449 | if (c2!='.' && c2!='-') break; |
|---|
| 450 | gap_count++; |
|---|
| 451 | } |
|---|
| 452 | s++; |
|---|
| 453 | } |
|---|
| 454 | |
|---|
| 455 | int setgap = (x_count+gap_count)*3; |
|---|
| 456 | memset(p, '.', setgap); |
|---|
| 457 | p += setgap; |
|---|
| 458 | } |
|---|
| 459 | else { |
|---|
| 460 | AWT_allowedCode allowed_code_left; |
|---|
| 461 | |
|---|
| 462 | if (x_count) { // synchronize |
|---|
| 463 | char protein[SYNC_LENGTH+1]; |
|---|
| 464 | int count; |
|---|
| 465 | { |
|---|
| 466 | int off; |
|---|
| 467 | |
|---|
| 468 | protein[0] = toupper(c); |
|---|
| 469 | for (count=1, off=0; count<SYNC_LENGTH; off++) { |
|---|
| 470 | char c2 = s[off]; |
|---|
| 471 | |
|---|
| 472 | if (c2!='.' && c2!='-') { |
|---|
| 473 | c2 = toupper(c2); |
|---|
| 474 | if (c2=='X') break; // can't sync X |
|---|
| 475 | protein[count++] = c2; |
|---|
| 476 | } |
|---|
| 477 | } |
|---|
| 478 | } |
|---|
| 479 | |
|---|
| 480 | nt_assert(count>=1); |
|---|
| 481 | protein[count] = 0; |
|---|
| 482 | |
|---|
| 483 | int catchUp; |
|---|
| 484 | if (count<SYNC_LENGTH) { |
|---|
| 485 | int sync_possibilities = 0; |
|---|
| 486 | int *sync_possible_with_catchup = new int[x_count*3+1]; |
|---|
| 487 | int maxCatchup = x_count*3; |
|---|
| 488 | |
|---|
| 489 | catchUp = x_count-1; |
|---|
| 490 | for (;;) { |
|---|
| 491 | if (!synchronizeCodons(protein, d, catchUp+1, maxCatchup, &catchUp, allowed_code, allowed_code_left)) { |
|---|
| 492 | break; |
|---|
| 493 | } |
|---|
| 494 | sync_possible_with_catchup[sync_possibilities++] = catchUp; |
|---|
| 495 | } |
|---|
| 496 | |
|---|
| 497 | if (sync_possibilities==0) { |
|---|
| 498 | delete [] sync_possible_with_catchup; |
|---|
| 499 | failed = 1; |
|---|
| 500 | fail_reason = "Can't synchronize after 'X'"; |
|---|
| 501 | break; |
|---|
| 502 | } |
|---|
| 503 | if (sync_possibilities>1) { |
|---|
| 504 | delete [] sync_possible_with_catchup; |
|---|
| 505 | failed = 1; |
|---|
| 506 | fail_reason = "Not enough data behind 'X' (please contact ARB-Support)"; |
|---|
| 507 | break; |
|---|
| 508 | } |
|---|
| 509 | |
|---|
| 510 | nt_assert(sync_possibilities==1); |
|---|
| 511 | catchUp = sync_possible_with_catchup[0]; |
|---|
| 512 | delete [] sync_possible_with_catchup; |
|---|
| 513 | } |
|---|
| 514 | else if (!synchronizeCodons(protein, d, x_count, x_count*3, &catchUp, allowed_code, allowed_code_left)) { |
|---|
| 515 | failed = 1; |
|---|
| 516 | fail_reason = "Can't synchronize after 'X'"; |
|---|
| 517 | break; |
|---|
| 518 | } |
|---|
| 519 | |
|---|
| 520 | allowed_code = allowed_code_left; |
|---|
| 521 | |
|---|
| 522 | // copy 'catchUp' characters (they are the content of the found Xs): |
|---|
| 523 | { |
|---|
| 524 | const char *after = s-1; |
|---|
| 525 | const char *i; |
|---|
| 526 | int off = int(after-x_start); |
|---|
| 527 | nt_assert(off>=x_count); |
|---|
| 528 | off = -(off*3); |
|---|
| 529 | int x_rest = x_count; |
|---|
| 530 | |
|---|
| 531 | for (i=x_start; i<after; i++) { |
|---|
| 532 | switch (i[0]) { |
|---|
| 533 | case 'x': |
|---|
| 534 | case 'X': |
|---|
| 535 | { |
|---|
| 536 | int take_per_X = catchUp/x_rest; |
|---|
| 537 | int o; |
|---|
| 538 | for (o=0; o<3; o++) { |
|---|
| 539 | if (o<take_per_X) { |
|---|
| 540 | p[off++] = *d++; |
|---|
| 541 | } |
|---|
| 542 | else { |
|---|
| 543 | p[off++] = '.'; |
|---|
| 544 | } |
|---|
| 545 | } |
|---|
| 546 | x_rest--; |
|---|
| 547 | break; |
|---|
| 548 | } |
|---|
| 549 | case '.': |
|---|
| 550 | case '-': |
|---|
| 551 | p[off++] = i[0]; |
|---|
| 552 | p[off++] = i[0]; |
|---|
| 553 | p[off++] = i[0]; |
|---|
| 554 | break; |
|---|
| 555 | default: |
|---|
| 556 | nt_assert(0); |
|---|
| 557 | break; |
|---|
| 558 | } |
|---|
| 559 | } |
|---|
| 560 | } |
|---|
| 561 | x_count = 0; |
|---|
| 562 | } |
|---|
| 563 | else { |
|---|
| 564 | const char *why_fail; |
|---|
| 565 | if (!AWT_is_codon(c, d, allowed_code, allowed_code_left, &why_fail)) { |
|---|
| 566 | failed = 1; |
|---|
| 567 | fail_reason = GBS_global_string("Not a codon (%s)", why_fail); |
|---|
| 568 | break; |
|---|
| 569 | } |
|---|
| 570 | |
|---|
| 571 | allowed_code = allowed_code_left; |
|---|
| 572 | } |
|---|
| 573 | |
|---|
| 574 | // copy one codon: |
|---|
| 575 | p[0] = d[0]; |
|---|
| 576 | p[1] = d[1]; |
|---|
| 577 | p[2] = d[2]; |
|---|
| 578 | |
|---|
| 579 | p += 3; |
|---|
| 580 | d += 3; |
|---|
| 581 | } |
|---|
| 582 | } |
|---|
| 583 | |
|---|
| 584 | if (!failed) { |
|---|
| 585 | int len = p-buffer; |
|---|
| 586 | int rest = ali_len-len; |
|---|
| 587 | |
|---|
| 588 | memset(p, '.', rest); |
|---|
| 589 | p += rest; |
|---|
| 590 | p[0] = 0; |
|---|
| 591 | } |
|---|
| 592 | } |
|---|
| 593 | |
|---|
| 594 | if (failed) { |
|---|
| 595 | int source_fail_pos = (s-1)-source+1; |
|---|
| 596 | int dest_fail_pos = 0; |
|---|
| 597 | { |
|---|
| 598 | int fail_d_base_count = d-compressed_dest; |
|---|
| 599 | char *dp = dest; |
|---|
| 600 | |
|---|
| 601 | for (;;) { |
|---|
| 602 | char c = *dp++; |
|---|
| 603 | |
|---|
| 604 | if (!c) { |
|---|
| 605 | nt_assert(c); |
|---|
| 606 | break; |
|---|
| 607 | } |
|---|
| 608 | if (c!='.' && c!='-') { |
|---|
| 609 | if (!fail_d_base_count) { |
|---|
| 610 | dest_fail_pos = (dp-1)-dest+1; |
|---|
| 611 | break; |
|---|
| 612 | } |
|---|
| 613 | fail_d_base_count--; |
|---|
| 614 | } |
|---|
| 615 | } |
|---|
| 616 | } |
|---|
| 617 | |
|---|
| 618 | { |
|---|
| 619 | char *dup_fail_reason = strdup(fail_reason); // otherwise it may be destroyed by GBS_global_string |
|---|
| 620 | aw_message(GBS_global_string("Automatic re-align failed for '%s'", GBT_read_name(gb_species))); |
|---|
| 621 | |
|---|
| 622 | if (ignore_fail_pos) { |
|---|
| 623 | aw_message(GBS_global_string("Reason: %s", dup_fail_reason)); |
|---|
| 624 | } |
|---|
| 625 | else { |
|---|
| 626 | aw_message(GBS_global_string("Reason: %s at %s:%i / %s:%i", dup_fail_reason, ali_source, source_fail_pos, ali_dest, dest_fail_pos)); |
|---|
| 627 | } |
|---|
| 628 | |
|---|
| 629 | free(dup_fail_reason); |
|---|
| 630 | } |
|---|
| 631 | |
|---|
| 632 | } |
|---|
| 633 | else { |
|---|
| 634 | nt_assert(strlen(buffer) == (unsigned)ali_len); |
|---|
| 635 | |
|---|
| 636 | // re-alignment successful |
|---|
| 637 | error = GB_write_string(gb_dest_data, buffer); |
|---|
| 638 | |
|---|
| 639 | if (!error) { |
|---|
| 640 | int explicit_table_known = allowed_code.explicit_table(); |
|---|
| 641 | |
|---|
| 642 | if (explicit_table_known >= 0) { // we know the exact code -> write codon_start and transl_table |
|---|
| 643 | const int codon_start = 1; // by definition (after realignment) |
|---|
| 644 | error = AWT_saveTranslationInfo(gb_species, explicit_table_known, codon_start); |
|---|
| 645 | } |
|---|
| 646 | else { // we dont know the exact code -> delete codon_start and transl_table |
|---|
| 647 | error = AWT_removeTranslationInfo(gb_species); |
|---|
| 648 | } |
|---|
| 649 | } |
|---|
| 650 | } |
|---|
| 651 | |
|---|
| 652 | free(buffer); |
|---|
| 653 | free(compressed_dest); |
|---|
| 654 | free(dest); |
|---|
| 655 | free(source); |
|---|
| 656 | |
|---|
| 657 | progress.inc_and_check_user_abort(error); |
|---|
| 658 | no_of_realigned_species++; |
|---|
| 659 | } |
|---|
| 660 | |
|---|
| 661 | *neededLength = max_wanted_ali_len; |
|---|
| 662 | |
|---|
| 663 | if (!error) { |
|---|
| 664 | int not_realigned = no_of_marked_species - no_of_realigned_species; |
|---|
| 665 | if (not_realigned>0) { |
|---|
| 666 | aw_message(GBS_global_string("Did not try to realign %i species (source/dest alignment missing?)", not_realigned)); |
|---|
| 667 | } |
|---|
| 668 | } |
|---|
| 669 | |
|---|
| 670 | if (!error) error = GBT_check_data(gb_main,ali_dest); |
|---|
| 671 | |
|---|
| 672 | return error; |
|---|
| 673 | } |
|---|
| 674 | |
|---|
| 675 | #undef SYNC_LENGTH |
|---|
| 676 | |
|---|
| 677 | |
|---|
| 678 | static void transdna_event(AW_window *aww) { |
|---|
| 679 | AW_root *aw_root = aww->get_root(); |
|---|
| 680 | char *ali_source = aw_root->awar(AWAR_TRANSPRO_DEST)->read_string(); |
|---|
| 681 | char *ali_dest = aw_root->awar(AWAR_TRANSPRO_SOURCE)->read_string(); |
|---|
| 682 | long neededLength = -1; |
|---|
| 683 | bool retrying = false; |
|---|
| 684 | GB_ERROR error = 0; |
|---|
| 685 | |
|---|
| 686 | while (!error && neededLength) { |
|---|
| 687 | error = GB_begin_transaction(GLOBAL.gb_main); |
|---|
| 688 | if (!error) error = arb_transdna(GLOBAL.gb_main, ali_source, ali_dest, &neededLength); |
|---|
| 689 | error = GB_end_transaction(GLOBAL.gb_main, error); |
|---|
| 690 | |
|---|
| 691 | if (!error && neededLength>0) { |
|---|
| 692 | if (retrying || !aw_ask_sure("increase_ali_length", GBS_global_string("Increase length of '%s' to %li?", ali_dest, neededLength))) { |
|---|
| 693 | error = GBS_global_string("Missing %li columns in alignment '%s'", neededLength, ali_dest); |
|---|
| 694 | } |
|---|
| 695 | else { |
|---|
| 696 | error = GB_begin_transaction(GLOBAL.gb_main); |
|---|
| 697 | if (!error) error = GBT_set_alignment_len(GLOBAL.gb_main, ali_dest, neededLength); // @@@ has no effect ? ? why ? |
|---|
| 698 | error = GB_end_transaction(GLOBAL.gb_main, error); |
|---|
| 699 | |
|---|
| 700 | if (!error) { |
|---|
| 701 | aw_message(GBS_global_string("Alignment length of '%s' has been set to %li\n" |
|---|
| 702 | "running re-aligner again!", |
|---|
| 703 | ali_dest, neededLength)); |
|---|
| 704 | retrying = true; |
|---|
| 705 | neededLength = -1; |
|---|
| 706 | } |
|---|
| 707 | } |
|---|
| 708 | } |
|---|
| 709 | else { |
|---|
| 710 | neededLength = 0; |
|---|
| 711 | } |
|---|
| 712 | } |
|---|
| 713 | |
|---|
| 714 | if (error) aw_message(error); |
|---|
| 715 | free(ali_dest); |
|---|
| 716 | free(ali_source); |
|---|
| 717 | } |
|---|
| 718 | |
|---|
| 719 | AW_window *NT_create_realign_dna_window(AW_root *root) { |
|---|
| 720 | AW_window_simple *aws = new AW_window_simple; |
|---|
| 721 | aws->init(root, "REALIGN_DNA", "REALIGN DNA"); |
|---|
| 722 | |
|---|
| 723 | aws->load_xfig("transdna.fig"); |
|---|
| 724 | |
|---|
| 725 | aws->callback((AW_CB0)AW_POPDOWN); |
|---|
| 726 | aws->at("close"); |
|---|
| 727 | aws->create_button("CLOSE", "CLOSE", "C"); |
|---|
| 728 | |
|---|
| 729 | aws->callback(makeHelpCallback("realign_dna.hlp")); |
|---|
| 730 | aws->at("help"); |
|---|
| 731 | aws->create_button("HELP", "HELP", "H"); |
|---|
| 732 | |
|---|
| 733 | aws->at("source"); |
|---|
| 734 | awt_create_selection_list_on_alignments(GLOBAL.gb_main, (AW_window *)aws, AWAR_TRANSPRO_SOURCE, "dna=:rna="); |
|---|
| 735 | aws->at("dest"); |
|---|
| 736 | awt_create_selection_list_on_alignments(GLOBAL.gb_main, (AW_window *)aws, AWAR_TRANSPRO_DEST, "pro=:ami="); |
|---|
| 737 | |
|---|
| 738 | aws->at("realign"); |
|---|
| 739 | aws->callback(transdna_event); |
|---|
| 740 | aws->highlight(); |
|---|
| 741 | aws->create_button("REALIGN", "REALIGN", "T"); |
|---|
| 742 | |
|---|
| 743 | return aws; |
|---|
| 744 | } |
|---|
| 745 | |
|---|
| 746 | |
|---|
| 747 | void NT_create_transpro_variables(AW_root *root, AW_default db1) |
|---|
| 748 | { |
|---|
| 749 | root->awar_string(AWAR_TRANSPRO_SOURCE, "", db1); |
|---|
| 750 | root->awar_string(AWAR_TRANSPRO_DEST, "", db1); |
|---|
| 751 | root->awar_string(AWAR_TRANSPRO_MODE, "settings", db1); |
|---|
| 752 | root->awar_int(AWAR_TRANSPRO_POS, 0, db1); |
|---|
| 753 | root->awar_int(AWAR_TRANSPRO_XSTART, 1, db1); |
|---|
| 754 | root->awar_int(AWAR_TRANSPRO_WRITE, 0, db1); |
|---|
| 755 | } |
|---|