| 1 | // ========================================================= // |
|---|
| 2 | // // |
|---|
| 3 | // File : NT_split_ali.cxx // |
|---|
| 4 | // Purpose : split alignments // |
|---|
| 5 | // // |
|---|
| 6 | // Coded by Ralf Westram (coder@reallysoft.de) in Jun 22 // |
|---|
| 7 | // http://www.arb-home.de/ // |
|---|
| 8 | // // |
|---|
| 9 | // ========================================================= // |
|---|
| 10 | |
|---|
| 11 | #include "NT_local_proto.h" |
|---|
| 12 | |
|---|
| 13 | #include <arbdbt.h> |
|---|
| 14 | #include <arb_progress.h> |
|---|
| 15 | #include <arb_defs.h> |
|---|
| 16 | #include <arb_strbuf.h> |
|---|
| 17 | #include <pos_range.h> |
|---|
| 18 | |
|---|
| 19 | #include <aw_root.hxx> |
|---|
| 20 | #include <aw_window.hxx> |
|---|
| 21 | |
|---|
| 22 | #include <limits> |
|---|
| 23 | #include <vector> |
|---|
| 24 | #include <insdel.h> |
|---|
| 25 | |
|---|
| 26 | using namespace std; |
|---|
| 27 | |
|---|
| 28 | typedef vector<ExplicitRange> RangeVec; |
|---|
| 29 | |
|---|
| 30 | static const int END__ = numeric_limits<int>::max(); |
|---|
| 31 | |
|---|
| 32 | static GB_ERROR calculate_and_check_ranges(int ali_source_len, const int *splitBefore, RangeVec& ranges) { |
|---|
| 33 | GB_ERROR error = NULp; |
|---|
| 34 | ExplicitRange rest(ali_source_len); |
|---|
| 35 | for (int p = 0; splitBefore[p] != END__ && !error; ++p) { |
|---|
| 36 | ExplicitRange next = intersection(ExplicitRange::prior(splitBefore[p]), rest); |
|---|
| 37 | if (next.is_empty()) { |
|---|
| 38 | if (rest.is_empty()) error = GBS_global_string("position %i exceeds overall length (=%i)", p ? info2bio(splitBefore[p-1]) : 0, ali_source_len); |
|---|
| 39 | else if (splitBefore[p]<rest.start()) error = "invalid position order"; |
|---|
| 40 | else error = "empty range dectected"; |
|---|
| 41 | } |
|---|
| 42 | else { |
|---|
| 43 | ranges.push_back(next); |
|---|
| 44 | rest = intersection(rest, next.after()); |
|---|
| 45 | } |
|---|
| 46 | } |
|---|
| 47 | |
|---|
| 48 | if (!error) { |
|---|
| 49 | if (rest.is_empty()) error = GBS_global_string("last position exceeds overall length (=%i)", ali_source_len); |
|---|
| 50 | else ranges.push_back(rest); // append final range |
|---|
| 51 | } |
|---|
| 52 | |
|---|
| 53 | return error; |
|---|
| 54 | } |
|---|
| 55 | |
|---|
| 56 | CONSTEXPR_INLINE int digits(int parts) { // when using numbers from [1 .. parts] |
|---|
| 57 | return parts<0 ? digits(-parts) : (parts<10 ? 1 : (parts<100 ? 2 : 3)); // works up to 999 |
|---|
| 58 | } |
|---|
| 59 | |
|---|
| 60 | inline const char *target_ali_name(const char *ali_target_prefix, int p, int digits) { |
|---|
| 61 | static SmartCharPtr ali; |
|---|
| 62 | ali = GBS_global_string_copy("%s_%0*i", ali_target_prefix, digits, p+1); |
|---|
| 63 | return &*ali; |
|---|
| 64 | } |
|---|
| 65 | |
|---|
| 66 | inline char *build_split_id(const RangeVec& ranges) { |
|---|
| 67 | GBS_strstruct buf(100); |
|---|
| 68 | for (size_t r = 0; r<ranges.size(); ++r) { |
|---|
| 69 | if (r>0) buf.put('-'); |
|---|
| 70 | buf.putlong(ranges[r].size()); |
|---|
| 71 | } |
|---|
| 72 | return buf.get_copy(); |
|---|
| 73 | } |
|---|
| 74 | |
|---|
| 75 | static GB_ERROR split_data_into_parts(GBDATA *gb_data, const RangeVec& ranges, const char *ali_target_prefix) { |
|---|
| 76 | GB_ERROR error = NULp; |
|---|
| 77 | char *seq = GB_read_as_string(gb_data); |
|---|
| 78 | const char *key = GB_read_key_pntr(gb_data); |
|---|
| 79 | GB_TYPES type = GB_read_type(gb_data); |
|---|
| 80 | int security = GB_read_security_write(gb_data); |
|---|
| 81 | size_t seq_len = strlen(seq); |
|---|
| 82 | const int parts = ranges.size(); |
|---|
| 83 | GBDATA *gb_item = GB_get_grandfather(gb_data); |
|---|
| 84 | |
|---|
| 85 | // [Note: the way elements are handled here is a bit q&d; see EditedTerminal::apply for howto properly handle all data types] |
|---|
| 86 | |
|---|
| 87 | // write splitted sequence data |
|---|
| 88 | for (int p = 0; p<parts && !error; ++p) { |
|---|
| 89 | const char *ali = target_ali_name(ali_target_prefix, p, digits(parts)); |
|---|
| 90 | GBDATA *gb_part_data = GBT_create_sequence_data(gb_item, ali, key, type, security); |
|---|
| 91 | |
|---|
| 92 | if (!gb_part_data) { |
|---|
| 93 | UNCOVERED(); |
|---|
| 94 | error = GB_await_error(); |
|---|
| 95 | } |
|---|
| 96 | else { |
|---|
| 97 | char *part = ranges[p].dup_corresponding_part(seq, seq_len); |
|---|
| 98 | error = GB_write_autoconv_string(gb_part_data, part); |
|---|
| 99 | free(part); |
|---|
| 100 | } |
|---|
| 101 | } |
|---|
| 102 | |
|---|
| 103 | free(seq); |
|---|
| 104 | |
|---|
| 105 | return error; |
|---|
| 106 | } |
|---|
| 107 | |
|---|
| 108 | static GB_ERROR copy_field_into_partial_alignments(GBDATA *gb_data, const int parts, const char *ali_target_prefix) { |
|---|
| 109 | GB_ERROR error = NULp; |
|---|
| 110 | GB_TYPES type = GB_read_type(gb_data); |
|---|
| 111 | GBDATA *gb_item = GB_get_grandfather(gb_data); |
|---|
| 112 | const char *key = GB_read_key_pntr(gb_data); |
|---|
| 113 | |
|---|
| 114 | char *target_val; |
|---|
| 115 | if (type == GB_STRING) target_val = GBS_global_string_copy("[splitted] %s", GB_read_char_pntr(gb_data)); |
|---|
| 116 | else { |
|---|
| 117 | target_val = GBS_global_string_copy("[split skipped copy of field '%s' (type: %i)]", key, type); |
|---|
| 118 | type = GB_STRING; |
|---|
| 119 | } |
|---|
| 120 | |
|---|
| 121 | const int security = GB_read_security_write(gb_data); |
|---|
| 122 | for (int p = 0; p<parts && !error; ++p) { |
|---|
| 123 | const char *ali = target_ali_name(ali_target_prefix, p, digits(parts)); |
|---|
| 124 | GBDATA *gb_part_data = GBT_create_sequence_data(gb_item, ali, key, type, security); |
|---|
| 125 | |
|---|
| 126 | error = GB_write_string(gb_part_data, target_val); |
|---|
| 127 | } |
|---|
| 128 | free(target_val); |
|---|
| 129 | |
|---|
| 130 | return error; |
|---|
| 131 | } |
|---|
| 132 | |
|---|
| 133 | static GB_ERROR split_alignment(GBDATA *gb_main, const char *ali_source, const char *ali_target_prefix, const int *splitBefore, bool markedOnly, bool overwriteAlignments) { |
|---|
| 134 | /*! splits one alignment into multiple alignments. |
|---|
| 135 | * @param ali_source name of source alignment |
|---|
| 136 | * @param ali_target_prefix prefix of target alignment. will be extended by '_1', '_2', ... |
|---|
| 137 | * @param splitBefore list of positions [0..N-1] where 2nd, 3rd, ... part shall begin. Terminate with 'END__'. |
|---|
| 138 | * @param markedOnly true->operate only on marked species, false->operate on all species |
|---|
| 139 | * @param overwriteAlignments true->delete+recreate existing alignments, false->bail-out with error |
|---|
| 140 | * @return error if sth went wrong. |
|---|
| 141 | */ |
|---|
| 142 | |
|---|
| 143 | GB_ERROR error = NULp; |
|---|
| 144 | GB_transaction ta(gb_main); |
|---|
| 145 | |
|---|
| 146 | arb_progress progress_main("Splitting sequences"); |
|---|
| 147 | |
|---|
| 148 | // check source alignment exists: |
|---|
| 149 | long ali_source_len = GBT_get_alignment_len(gb_main, ali_source); |
|---|
| 150 | if (ali_source_len<=0) { |
|---|
| 151 | error = GB_await_error(); |
|---|
| 152 | } |
|---|
| 153 | else { |
|---|
| 154 | RangeVec ranges; |
|---|
| 155 | error = calculate_and_check_ranges(ali_source_len, splitBefore, ranges); |
|---|
| 156 | |
|---|
| 157 | const int parts = ranges.size(); |
|---|
| 158 | if (parts<2 && !error) { |
|---|
| 159 | error = GBS_global_string("Not enough parts (%i) specified to split alignment", parts); |
|---|
| 160 | } |
|---|
| 161 | |
|---|
| 162 | // @@@ read the following values from source alignment: |
|---|
| 163 | bool is_aligned = true; // @@@ if source alignment is not formatted -> abort with error (for safety and to avoid special cases here) |
|---|
| 164 | int security = 3; |
|---|
| 165 | const char *ali_type = "rna"; |
|---|
| 166 | |
|---|
| 167 | char *split_id = build_split_id(ranges); |
|---|
| 168 | |
|---|
| 169 | // create alignments: |
|---|
| 170 | for (int p = 0; p<parts && !error; ++p) { |
|---|
| 171 | const char *ali = target_ali_name(ali_target_prefix, p, digits(parts)); |
|---|
| 172 | { |
|---|
| 173 | GBDATA *gb_ali = GBT_get_alignment(gb_main, ali); |
|---|
| 174 | if (!gb_ali) { |
|---|
| 175 | GB_clear_error(); // from GBT_get_alignment |
|---|
| 176 | } |
|---|
| 177 | else { |
|---|
| 178 | if (overwriteAlignments) { |
|---|
| 179 | error = GBT_delete_alignment(gb_main, ali); // delete existing alignment |
|---|
| 180 | } |
|---|
| 181 | else { |
|---|
| 182 | error = GBS_global_string("alignment '%s' already exists", ali); |
|---|
| 183 | } |
|---|
| 184 | } |
|---|
| 185 | } |
|---|
| 186 | if (!error) { |
|---|
| 187 | // create target alignment: |
|---|
| 188 | const ExplicitRange& range = ranges[p]; |
|---|
| 189 | char *why_created = GBS_global_string_copy("while splitting %s into %s [%i-%i]", ali_source, split_id, info2bio(range.start()), info2bio(range.end())); |
|---|
| 190 | |
|---|
| 191 | GBDATA *gb_ali = GBT_create_new_alignment(gb_main, ali, ranges[p].size(), is_aligned, security, ali_type, why_created); |
|---|
| 192 | if (!gb_ali) { |
|---|
| 193 | error = GB_await_error(); |
|---|
| 194 | } |
|---|
| 195 | else { |
|---|
| 196 | error = GBT_add_alignment_changekeys(gb_main, ali); |
|---|
| 197 | } |
|---|
| 198 | |
|---|
| 199 | free(why_created); |
|---|
| 200 | } |
|---|
| 201 | } |
|---|
| 202 | |
|---|
| 203 | if (!error) { |
|---|
| 204 | size_t data_count = 0; // count species with sequence data in alignment (during pass 1) |
|---|
| 205 | size_t sai_data_count = 0; // count sequence associated data in alignment containers of all SAIs (during pass 1) |
|---|
| 206 | size_t sequences_splitted = 0; // count splitted sequences (species) |
|---|
| 207 | size_t sai_data_splitted = 0; // count splitted data entries (SAI; may be multiple per SAI) |
|---|
| 208 | |
|---|
| 209 | SmartPtr<arb_progress> progress; |
|---|
| 210 | |
|---|
| 211 | for (int pass = 1; pass<=2 && !error; ++pass) { |
|---|
| 212 | if (pass == 2) { |
|---|
| 213 | progress = new arb_progress(data_count+sai_data_count); |
|---|
| 214 | progress->auto_subtitles("seq"); |
|---|
| 215 | } |
|---|
| 216 | |
|---|
| 217 | // generate sequence data in splitted alignments: |
|---|
| 218 | for (GBDATA *gb_species = markedOnly ? GBT_first_marked_species(gb_main) : GBT_first_species(gb_main); |
|---|
| 219 | gb_species && !error; |
|---|
| 220 | gb_species = markedOnly ? GBT_next_marked_species(gb_species) : GBT_next_species(gb_species)) |
|---|
| 221 | { |
|---|
| 222 | GBDATA *gb_seq = GBT_find_sequence(gb_species, ali_source); |
|---|
| 223 | if (gb_seq) { |
|---|
| 224 | if (pass == 1) ++data_count; |
|---|
| 225 | else { |
|---|
| 226 | error = split_data_into_parts(gb_seq, ranges, ali_target_prefix); |
|---|
| 227 | if (!error) ++sequences_splitted; // count splits |
|---|
| 228 | progress->inc_and_check_user_abort(error); |
|---|
| 229 | } |
|---|
| 230 | } |
|---|
| 231 | } |
|---|
| 232 | |
|---|
| 233 | for (GBDATA *gb_sai = GBT_first_SAI(gb_main); gb_sai && !error; gb_sai = GBT_next_SAI(gb_sai)) { |
|---|
| 234 | GBDATA *gb_ali = GB_entry(gb_sai, ali_source); |
|---|
| 235 | if (gb_ali) { |
|---|
| 236 | for (GBDATA *gb_data = GB_child(gb_ali); gb_data && !error; gb_data = GB_nextChild(gb_data)) { |
|---|
| 237 | if (ARB_is_alignment_relative_data(gb_data)) { |
|---|
| 238 | if (pass == 1) { |
|---|
| 239 | ++sai_data_count; |
|---|
| 240 | } |
|---|
| 241 | else { |
|---|
| 242 | error = split_data_into_parts(gb_data, ranges, ali_target_prefix); |
|---|
| 243 | if (!error) ++sai_data_splitted; // count splits |
|---|
| 244 | progress->inc_and_check_user_abort(error); |
|---|
| 245 | } |
|---|
| 246 | } |
|---|
| 247 | else if (pass == 2) { |
|---|
| 248 | error = copy_field_into_partial_alignments(gb_data, parts, ali_target_prefix); |
|---|
| 249 | } |
|---|
| 250 | } |
|---|
| 251 | } |
|---|
| 252 | } |
|---|
| 253 | } |
|---|
| 254 | |
|---|
| 255 | if (!error) { |
|---|
| 256 | size_t overall_data_splitted = sequences_splitted + sai_data_splitted; |
|---|
| 257 | if (overall_data_splitted) { |
|---|
| 258 | size_t split_count = overall_data_splitted*parts; |
|---|
| 259 | fprintf(stdout, "Summary: splitted %zu element%s into %zu parts.\n", overall_data_splitted, plural(overall_data_splitted), split_count); |
|---|
| 260 | } |
|---|
| 261 | else { |
|---|
| 262 | UNCOVERED(); |
|---|
| 263 | error = "No sequences were splitted (none of the handled species contained data in alignment; check settings)"; |
|---|
| 264 | } |
|---|
| 265 | } |
|---|
| 266 | } |
|---|
| 267 | |
|---|
| 268 | free(split_id); |
|---|
| 269 | } |
|---|
| 270 | error = ta.close(error); |
|---|
| 271 | |
|---|
| 272 | return error; |
|---|
| 273 | } |
|---|
| 274 | |
|---|
| 275 | // -------------------------------------------------------------------------------- |
|---|
| 276 | |
|---|
| 277 | #ifdef UNIT_TESTS |
|---|
| 278 | #include <arb_diff.h> |
|---|
| 279 | #include <arb_file.h> |
|---|
| 280 | #ifndef TEST_UNIT_H |
|---|
| 281 | #include <test_unit.h> |
|---|
| 282 | #endif |
|---|
| 283 | |
|---|
| 284 | // #define TEST_AUTO_UPDATE // uncomment to auto-update expected result DB |
|---|
| 285 | |
|---|
| 286 | void TEST_split_alignment() { |
|---|
| 287 | GB_shell shell; |
|---|
| 288 | GBDATA *gb_main = GB_open("TEST_prot_tiny.arb", "rw"); // source: ../UNIT_TESTER/run/TEST_prot_tiny.arb |
|---|
| 289 | // target: ../UNIT_TESTER/run/TEST_split_expected.arb |
|---|
| 290 | |
|---|
| 291 | // alignments existing in DB: |
|---|
| 292 | // -------------------------- |
|---|
| 293 | // alignment_name "ali_dna" alignment_len %i 337 |
|---|
| 294 | // alignment_name "ali_prot" alignment_len %i 112 |
|---|
| 295 | // alignment_name "ali_dna_incomplete" alignment_len %i 337 |
|---|
| 296 | |
|---|
| 297 | { |
|---|
| 298 | GB_securityLevel raised(gb_main, 3); |
|---|
| 299 | |
|---|
| 300 | // split some alignment |
|---|
| 301 | { |
|---|
| 302 | GB_transaction ta(gb_main); |
|---|
| 303 | TEST_EXPECT_NO_ERROR(GBT_restore_marked_species(gb_main, "CytLyti6;BctFra12;StrCoel9;MucRace2")); |
|---|
| 304 | |
|---|
| 305 | int split[] = { 41, 175, END__ }; |
|---|
| 306 | TEST_EXPECT_NO_ERROR(split_alignment(gb_main, "ali_dna", "ali_dna_split1", split, true, false)); |
|---|
| 307 | } |
|---|
| 308 | { |
|---|
| 309 | GB_transaction ta(gb_main); |
|---|
| 310 | TEST_EXPECT_NO_ERROR(GBT_restore_marked_species(gb_main, "CytLyti6;TaxOcell;BctFra12;StrRamo3")); |
|---|
| 311 | |
|---|
| 312 | int split[] = { 168, END__ }; // approx. split in the middle of the sequence |
|---|
| 313 | TEST_EXPECT_NO_ERROR(split_alignment(gb_main, "ali_dna", "ali_dna_split2", split, true, false)); |
|---|
| 314 | TEST_EXPECT_NO_ERROR(split_alignment(gb_main, "ali_dna", "ali_overwrite", split, true, false)); // created to be overwritten below |
|---|
| 315 | } |
|---|
| 316 | { |
|---|
| 317 | GB_transaction ta(gb_main); |
|---|
| 318 | TEST_EXPECT_NO_ERROR(GBT_restore_marked_species(gb_main, "TaxOcell")); |
|---|
| 319 | |
|---|
| 320 | int split[] = { 1, 160, 161, 336, END__ }; // test short parts (at 5'-end, in the middle and at 3'-end) |
|---|
| 321 | TEST_EXPECT_NO_ERROR(split_alignment(gb_main, "ali_dna", "ali_dna_split3", split, true, false)); |
|---|
| 322 | } |
|---|
| 323 | { |
|---|
| 324 | GB_transaction ta(gb_main); |
|---|
| 325 | |
|---|
| 326 | int split[] = { 25, 65, END__ }; |
|---|
| 327 | TEST_EXPECT_NO_ERROR(split_alignment(gb_main, "ali_prot", "ali_prot_split", split, false, false)); |
|---|
| 328 | } |
|---|
| 329 | { // split alignment not occurring in all species (applied to all species) |
|---|
| 330 | GB_transaction ta(gb_main); |
|---|
| 331 | |
|---|
| 332 | int split[] = { 150, END__ }; |
|---|
| 333 | TEST_EXPECT_NO_ERROR(split_alignment(gb_main, "ali_dna_incomplete", "ali_dna_incomplete_split", split, false, false)); |
|---|
| 334 | } |
|---|
| 335 | |
|---|
| 336 | |
|---|
| 337 | // provoke errors (invalid ali positions): |
|---|
| 338 | { |
|---|
| 339 | int split[] = { END__ }; |
|---|
| 340 | TEST_EXPECT_ERROR_CONTAINS(split_alignment(gb_main, "ali_dna", "ali_any", split, true, false), |
|---|
| 341 | "Not enough parts (1) specified to split alignment"); |
|---|
| 342 | } |
|---|
| 343 | { |
|---|
| 344 | int split[] = { 41, 470, END__ }; // requests invalid split (beyond 3-end) -> error wanted |
|---|
| 345 | TEST_EXPECT_ERROR_CONTAINS(split_alignment(gb_main, "ali_dna", "ali_any", split, true, false), |
|---|
| 346 | "last position exceeds overall length (=337)"); |
|---|
| 347 | } |
|---|
| 348 | |
|---|
| 349 | { |
|---|
| 350 | int split[] = { 336, 4100, END__ }; // tests that offset 336 is a legal split position |
|---|
| 351 | TEST_EXPECT_ERROR_CONTAINS(split_alignment(gb_main, "ali_dna", "ali_any", split, true, false), |
|---|
| 352 | "last position exceeds overall length (=337)"); // 2nd offset causes this failure |
|---|
| 353 | } |
|---|
| 354 | { |
|---|
| 355 | int split[] = { 337, 4100, END__ }; // test that offset 337 is NO legal split position (no data behind) |
|---|
| 356 | TEST_EXPECT_ERROR_CONTAINS(split_alignment(gb_main, "ali_dna", "ali_any", split, true, false), |
|---|
| 357 | "position 338 exceeds overall length (=337)"); |
|---|
| 358 | } |
|---|
| 359 | { |
|---|
| 360 | int split[] = { 338, 4100, END__ }; // requests invalid split (beyond 3-end) with first offset -> error wanted |
|---|
| 361 | TEST_EXPECT_ERROR_CONTAINS(split_alignment(gb_main, "ali_dna", "ali_any", split, true, false), |
|---|
| 362 | "position 339 exceeds overall length (=337)"); |
|---|
| 363 | } |
|---|
| 364 | { |
|---|
| 365 | int split[] = { 0, 4100, END__ }; // requests invalid split (at 5-end) with first offset -> error wanted |
|---|
| 366 | TEST_EXPECT_ERROR_CONTAINS(split_alignment(gb_main, "ali_dna", "ali_any", split, true, false), |
|---|
| 367 | "empty range dectected"); |
|---|
| 368 | } |
|---|
| 369 | |
|---|
| 370 | { |
|---|
| 371 | int split[] = { 41, 30, 175, END__ }; // 2nd offset < 1st offset -> error wanted |
|---|
| 372 | TEST_EXPECT_ERROR_CONTAINS(split_alignment(gb_main, "ali_dna", "ali_any", split, true, false), |
|---|
| 373 | "invalid position order"); |
|---|
| 374 | } |
|---|
| 375 | { |
|---|
| 376 | int split[] = { 41, 41, 175, END__ }; // empty range -> error wanted |
|---|
| 377 | TEST_EXPECT_ERROR_CONTAINS(split_alignment(gb_main, "ali_dna", "ali_any", split, true, false), |
|---|
| 378 | "empty range dectected"); |
|---|
| 379 | } |
|---|
| 380 | |
|---|
| 381 | { |
|---|
| 382 | // provoke error (no such alignment) |
|---|
| 383 | int split[] = { 41, 175, END__ }; |
|---|
| 384 | TEST_EXPECT_ERROR_CONTAINS(split_alignment(gb_main, "ali_nosuch", "ali_any", split, true, false), |
|---|
| 385 | "alignment 'ali_nosuch' not found"); |
|---|
| 386 | } |
|---|
| 387 | |
|---|
| 388 | { |
|---|
| 389 | // provoke error (target alignment exists) |
|---|
| 390 | |
|---|
| 391 | const char *ali_conflict = "ali_existing_3"; |
|---|
| 392 | GBDATA *gb_ali_conflict = NULp; |
|---|
| 393 | { |
|---|
| 394 | GB_transaction ta(gb_main); |
|---|
| 395 | gb_ali_conflict = GBT_create_new_alignment(gb_main, ali_conflict, 123, 0, 0, "dna", "by TEST_split_alignment"); |
|---|
| 396 | TEST_REJECT_NULL(gb_ali_conflict); |
|---|
| 397 | } |
|---|
| 398 | |
|---|
| 399 | { |
|---|
| 400 | int split[] = { 41, 175, END__ }; |
|---|
| 401 | TEST_EXPECT_ERROR_CONTAINS(split_alignment(gb_main, "ali_dna", "ali_existing", split, true, false), |
|---|
| 402 | "alignment 'ali_existing_3' already exists"); |
|---|
| 403 | } |
|---|
| 404 | |
|---|
| 405 | // overwrite existing alignment |
|---|
| 406 | { |
|---|
| 407 | GB_transaction ta(gb_main); |
|---|
| 408 | TEST_EXPECT_NO_ERROR(GBT_restore_marked_species(gb_main, "CytLyti6;BctFra12;MucRace2")); |
|---|
| 409 | } |
|---|
| 410 | { |
|---|
| 411 | int split[] = { 170, END__ }; // approx. split in the middle of the sequence (but different than in 1st split to ali_overwrite..) |
|---|
| 412 | |
|---|
| 413 | GB_topSecurityLevel raise(gb_main); |
|---|
| 414 | TEST_EXPECT_NO_ERROR(split_alignment(gb_main, "ali_dna", "ali_overwrite", split, true, true)); |
|---|
| 415 | } |
|---|
| 416 | } |
|---|
| 417 | } |
|---|
| 418 | |
|---|
| 419 | GBT_mark_all(gb_main, 0); // clear marks before saving database |
|---|
| 420 | |
|---|
| 421 | { |
|---|
| 422 | // save as ascii with different name + compare result with expected file |
|---|
| 423 | |
|---|
| 424 | char *savename = ARB_strdup(GB_path_in_ARBHOME("UNIT_TESTER/run/TEST_split.arb")); |
|---|
| 425 | char *expected = ARB_strdup(GB_path_in_ARBHOME("UNIT_TESTER/run/TEST_split_expected.arb")); |
|---|
| 426 | |
|---|
| 427 | TEST_EXPECT_NO_ERROR(GB_save_as(gb_main, savename, "a")); |
|---|
| 428 | |
|---|
| 429 | #if defined(TEST_AUTO_UPDATE) |
|---|
| 430 | TEST_COPY_FILE(savename, expected); |
|---|
| 431 | #else // !defined(TEST_AUTO_UPDATE) |
|---|
| 432 | TEST_EXPECT_TEXTFILES_EQUAL(savename, expected); |
|---|
| 433 | #endif |
|---|
| 434 | TEST_EXPECT_ZERO_OR_SHOW_ERRNO(GB_unlink(savename)); |
|---|
| 435 | |
|---|
| 436 | free(expected); |
|---|
| 437 | free(savename); |
|---|
| 438 | } |
|---|
| 439 | |
|---|
| 440 | GB_close(gb_main); |
|---|
| 441 | } |
|---|
| 442 | |
|---|
| 443 | #endif // UNIT_TESTS |
|---|
| 444 | |
|---|
| 445 | // -------------------------------------------------------------------------------- |
|---|
| 446 | |
|---|
| 447 | #include "NT_local.h" |
|---|
| 448 | |
|---|
| 449 | #include <arb_strarray.h> |
|---|
| 450 | |
|---|
| 451 | #include <aw_awar.hxx> |
|---|
| 452 | #include <aw_awar_defs.hxx> |
|---|
| 453 | #include <aw_msg.hxx> |
|---|
| 454 | |
|---|
| 455 | |
|---|
| 456 | #define AWAR_SPLIT_ALI_BASE "split_ali/" |
|---|
| 457 | #define AWAR_SPLIT_ALI_BTMP "tmp/" AWAR_SPLIT_ALI_BASE |
|---|
| 458 | |
|---|
| 459 | #define AWAR_SPLIT_ALI_TARGET AWAR_SPLIT_ALI_BTMP "target" |
|---|
| 460 | #define AWAR_SPLIT_ALI_POSITIONS AWAR_SPLIT_ALI_BASE "pos" |
|---|
| 461 | #define AWAR_SPLIT_ALI_MARKED AWAR_SPLIT_ALI_BASE "marked" |
|---|
| 462 | #define AWAR_SPLIT_ALI_OVERWRITE AWAR_SPLIT_ALI_BASE "overwrite" |
|---|
| 463 | |
|---|
| 464 | static void create_splitAlignment_awars(AW_root *root, AW_default properties, GBDATA *gb_main) { |
|---|
| 465 | // use current alignment to init default for target alignment: |
|---|
| 466 | const char *ali_part = "ali_example_part"; |
|---|
| 467 | { |
|---|
| 468 | char *ali_source = GBT_get_default_alignment(gb_main); |
|---|
| 469 | if (!ali_source) GB_clear_error(); |
|---|
| 470 | else ali_part = GBS_global_string("%s_part", ali_source); |
|---|
| 471 | free(ali_source); |
|---|
| 472 | } |
|---|
| 473 | |
|---|
| 474 | root->awar_string(AWAR_SPLIT_ALI_TARGET, ali_part, properties); |
|---|
| 475 | root->awar_string(AWAR_SPLIT_ALI_POSITIONS, "", properties); |
|---|
| 476 | root->awar_int (AWAR_SPLIT_ALI_MARKED, 0, properties); |
|---|
| 477 | root->awar_int (AWAR_SPLIT_ALI_OVERWRITE, 0, properties); |
|---|
| 478 | |
|---|
| 479 | root->awar_string(AWAR_RANGE, "", gb_main); |
|---|
| 480 | } |
|---|
| 481 | |
|---|
| 482 | static void split_ali_cb(AW_window *aww) { |
|---|
| 483 | AW_root *awr = aww->get_root(); |
|---|
| 484 | GBDATA *gb_main = GLOBAL.gb_main; |
|---|
| 485 | GB_ERROR error = NULp; |
|---|
| 486 | |
|---|
| 487 | char *ali_source = GBT_get_default_alignment(gb_main); |
|---|
| 488 | if (!ali_source) { |
|---|
| 489 | error = GB_await_error(); |
|---|
| 490 | } |
|---|
| 491 | else { |
|---|
| 492 | char *ali_target_prefix = awr->awar(AWAR_SPLIT_ALI_TARGET)->read_string(); |
|---|
| 493 | |
|---|
| 494 | ConstStrArray numbers; |
|---|
| 495 | { |
|---|
| 496 | char *split_pos_list = awr->awar(AWAR_SPLIT_ALI_POSITIONS)->read_string(); |
|---|
| 497 | GBT_splitNdestroy_string(numbers, split_pos_list, ", ", SPLIT_DROPEMPTY); |
|---|
| 498 | } |
|---|
| 499 | |
|---|
| 500 | int split_positions[numbers.size()+1]; |
|---|
| 501 | for (size_t n = 0; n<numbers.size() && !error; ++n) { |
|---|
| 502 | const char *num = numbers[n]; |
|---|
| 503 | split_positions[n] = bio2info(atoi(num)); |
|---|
| 504 | if (split_positions[n]<1) { |
|---|
| 505 | error = GBS_global_string("invalid position '%s'", num); // shows biological position |
|---|
| 506 | } |
|---|
| 507 | } |
|---|
| 508 | split_positions[numbers.size()] = END__; |
|---|
| 509 | |
|---|
| 510 | if (!error) { |
|---|
| 511 | bool markedOnly = awr->awar(AWAR_SPLIT_ALI_MARKED)->read_int(); |
|---|
| 512 | bool overwrite = awr->awar(AWAR_SPLIT_ALI_OVERWRITE)->read_int(); |
|---|
| 513 | |
|---|
| 514 | error = split_alignment(gb_main, ali_source, ali_target_prefix, split_positions, markedOnly, overwrite); |
|---|
| 515 | } |
|---|
| 516 | |
|---|
| 517 | free(ali_target_prefix); |
|---|
| 518 | free(ali_source); |
|---|
| 519 | } |
|---|
| 520 | |
|---|
| 521 | aw_message_if(error); |
|---|
| 522 | } |
|---|
| 523 | |
|---|
| 524 | static void use_editor_range_cb(AW_window *aww) { |
|---|
| 525 | AW_root *awr = aww->get_root(); |
|---|
| 526 | GB_ERROR error = NULp; |
|---|
| 527 | const char *editor_range = awr->awar(AWAR_RANGE)->read_char_pntr(); |
|---|
| 528 | const char *hyphen = strchr(editor_range, '-'); |
|---|
| 529 | |
|---|
| 530 | if (!hyphen) { |
|---|
| 531 | if (editor_range[0]) { |
|---|
| 532 | error = "expected range to contain '-'."; |
|---|
| 533 | } |
|---|
| 534 | else { |
|---|
| 535 | error = "No block has been highlighted in sequence editor."; |
|---|
| 536 | } |
|---|
| 537 | } |
|---|
| 538 | else { |
|---|
| 539 | long i1 = atol(editor_range); |
|---|
| 540 | long i2 = atol(hyphen+1); |
|---|
| 541 | |
|---|
| 542 | const char *pos_list = GBS_global_string("%li,%li", i1, i2+1); // split BEFORE i1 + AFTER i2 |
|---|
| 543 | awr->awar(AWAR_SPLIT_ALI_POSITIONS)->write_string(pos_list); |
|---|
| 544 | } |
|---|
| 545 | |
|---|
| 546 | aw_message_if(error); |
|---|
| 547 | } |
|---|
| 548 | |
|---|
| 549 | AW_window *NT_create_splitAlignment_window(AW_root *awr) { |
|---|
| 550 | create_splitAlignment_awars(awr, AW_ROOT_DEFAULT, GLOBAL.gb_main); |
|---|
| 551 | |
|---|
| 552 | AW_window_simple *aws = new AW_window_simple; |
|---|
| 553 | aws->init(awr, "SPLIT_ALIGNMENT", "Split alignment"); |
|---|
| 554 | |
|---|
| 555 | const int PAD = 10; |
|---|
| 556 | |
|---|
| 557 | const int BUTTON_LENGTH = 10; |
|---|
| 558 | const int FIELD_LENGTH = 40; |
|---|
| 559 | const int LABEL_LENGTH = 30; |
|---|
| 560 | |
|---|
| 561 | aws->at(PAD, PAD); |
|---|
| 562 | aws->auto_space(PAD/2, PAD); |
|---|
| 563 | |
|---|
| 564 | aws->button_length(BUTTON_LENGTH); |
|---|
| 565 | |
|---|
| 566 | aws->callback(AW_POPDOWN); |
|---|
| 567 | aws->create_button("CLOSE", "Close", "C"); |
|---|
| 568 | |
|---|
| 569 | aws->callback(makeHelpCallback("split.hlp")); |
|---|
| 570 | aws->create_button("HELP", "Help", "H"); |
|---|
| 571 | |
|---|
| 572 | aws->at_newline(); |
|---|
| 573 | |
|---|
| 574 | aws->label_length(LABEL_LENGTH); |
|---|
| 575 | |
|---|
| 576 | aws->label("Target alignment prefix:"); |
|---|
| 577 | aws->create_input_field(AWAR_SPLIT_ALI_TARGET, FIELD_LENGTH); |
|---|
| 578 | aws->at_newline(); |
|---|
| 579 | |
|---|
| 580 | aws->label("Position(s) to split before:\n" |
|---|
| 581 | "(comma-separated)"); |
|---|
| 582 | aws->create_input_field(AWAR_SPLIT_ALI_POSITIONS, FIELD_LENGTH); |
|---|
| 583 | aws->at_newline(); |
|---|
| 584 | |
|---|
| 585 | aws->button_length(LABEL_LENGTH-1); |
|---|
| 586 | aws->callback(use_editor_range_cb); |
|---|
| 587 | aws->create_button("USE_EDITOR_RANGE", "Use editor range:"); |
|---|
| 588 | aws->button_length(FIELD_LENGTH); |
|---|
| 589 | aws->create_button(NULp, AWAR_RANGE, "", "-"); |
|---|
| 590 | aws->button_length(BUTTON_LENGTH); |
|---|
| 591 | aws->at_newline(); |
|---|
| 592 | |
|---|
| 593 | aws->label("Split marked only:"); |
|---|
| 594 | aws->create_toggle(AWAR_SPLIT_ALI_MARKED); |
|---|
| 595 | aws->at_newline(); |
|---|
| 596 | |
|---|
| 597 | aws->label("Overwrite existing alignments:\n" |
|---|
| 598 | "(deletes ALL data)"); |
|---|
| 599 | aws->create_toggle(AWAR_SPLIT_ALI_OVERWRITE); |
|---|
| 600 | aws->at_newline(); |
|---|
| 601 | |
|---|
| 602 | aws->callback(split_ali_cb); |
|---|
| 603 | aws->create_button("GO", "GO", "G"); |
|---|
| 604 | |
|---|
| 605 | return aws; |
|---|
| 606 | } |
|---|
| 607 | |
|---|