jalviewX
File Alignment.java
Coverage histogram
28% of files have more coverage
Classes
| Class | Line # | Actions | ||||
|---|---|---|---|---|---|---|
| Alignment | 50 | 583 | 289 | 355 |
Contributing tests
This file is covered by 398 tests.
.
Source view
| 1 | /* | |
| 2 | * Jalview - A Sequence Alignment Editor and Viewer ($$Version-Rel$$) | |
| 3 | * Copyright (C) $$Year-Rel$$ The Jalview Authors | |
| 4 | * | |
| 5 | * This file is part of Jalview. | |
| 6 | * | |
| 7 | * Jalview is free software: you can redistribute it and/or | |
| 8 | * modify it under the terms of the GNU General Public License | |
| 9 | * as published by the Free Software Foundation, either version 3 | |
| 10 | * of the License, or (at your option) any later version. | |
| 11 | * | |
| 12 | * Jalview is distributed in the hope that it will be useful, but | |
| 13 | * WITHOUT ANY WARRANTY; without even the implied warranty | |
| 14 | * of MERCHANTABILITY or FITNESS FOR A PARTICULAR | |
| 15 | * PURPOSE. See the GNU General Public License for more details. | |
| 16 | * | |
| 17 | * You should have received a copy of the GNU General Public License | |
| 18 | * along with Jalview. If not, see <http://www.gnu.org/licenses/>. | |
| 19 | * The Jalview Authors are detailed in the 'AUTHORS' file. | |
| 20 | */ | |
| 21 | package jalview.datamodel; | |
| 22 | ||
| 23 | import jalview.analysis.AlignmentUtils; | |
| 24 | import jalview.datamodel.AlignedCodonFrame.SequenceToSequenceMapping; | |
| 25 | import jalview.io.FastaFile; | |
| 26 | import jalview.util.Comparison; | |
| 27 | import jalview.util.LinkedIdentityHashSet; | |
| 28 | import jalview.util.MessageManager; | |
| 29 | ||
| 30 | import java.util.ArrayList; | |
| 31 | import java.util.Arrays; | |
| 32 | import java.util.BitSet; | |
| 33 | import java.util.Collections; | |
| 34 | import java.util.Enumeration; | |
| 35 | import java.util.HashSet; | |
| 36 | import java.util.Hashtable; | |
| 37 | import java.util.Iterator; | |
| 38 | import java.util.List; | |
| 39 | import java.util.Map; | |
| 40 | import java.util.Set; | |
| 41 | import java.util.Vector; | |
| 42 | ||
| 43 | /** | |
| 44 | * Data structure to hold and manipulate a multiple sequence alignment | |
| 45 | */ | |
| 46 | /** | |
| 47 | * @author JimP | |
| 48 | * | |
| 49 | */ | |
| 50 | public class Alignment implements AlignmentI | |
| 51 | { | |
| 52 | private Alignment dataset; | |
| 53 | ||
| 54 | private List<SequenceI> sequences; | |
| 55 | ||
| 56 | protected List<SequenceGroup> groups; | |
| 57 | ||
| 58 | protected char gapCharacter = '-'; | |
| 59 | ||
| 60 | private boolean nucleotide = true; | |
| 61 | ||
| 62 | public boolean hasRNAStructure = false; | |
| 63 | ||
| 64 | public AlignmentAnnotation[] annotations; | |
| 65 | ||
| 66 | HiddenSequences hiddenSequences; | |
| 67 | ||
| 68 | HiddenColumns hiddenCols; | |
| 69 | ||
| 70 | public Hashtable alignmentProperties; | |
| 71 | ||
| 72 | private List<AlignedCodonFrame> codonFrameList; | |
| 73 | ||
| 74 | 1211 |  private void initAlignment(SequenceI[] seqs) |
| 75 | { | |
| 76 | 1211 | groups = Collections.synchronizedList(new ArrayList<SequenceGroup>()); |
| 77 | 1211 | hiddenSequences = new HiddenSequences(this); |
| 78 | 1211 | hiddenCols = new HiddenColumns(); |
| 79 | 1211 | codonFrameList = new ArrayList<>(); |
| 80 | ||
| 81 | 1211 | nucleotide = Comparison.isNucleotide(seqs); |
| 82 | ||
| 83 | 1211 | sequences = Collections.synchronizedList(new ArrayList<SequenceI>()); |
| 84 | ||
| 85 | 30776 | for (int i = 0; i < seqs.length; i++) |
| 86 | { | |
| 87 | 29565 | sequences.add(seqs[i]); |
| 88 | } | |
| 89 | ||
| 90 | } | |
| 91 | ||
| 92 | /** | |
| 93 | * Make a 'copy' alignment - sequences have new copies of features and | |
| 94 | * annotations, but share the original dataset sequences. | |
| 95 | */ | |
| 96 | 4 | public Alignment(AlignmentI al) |
| 97 | { | |
| 98 | 4 | SequenceI[] seqs = al.getSequencesArray(); |
| 99 | 19 | for (int i = 0; i < seqs.length; i++) |
| 100 | { | |
| 101 | 15 | seqs[i] = new Sequence(seqs[i]); |
| 102 | } | |
| 103 | ||
| 104 | 4 | initAlignment(seqs); |
| 105 | ||
| 106 | /* | |
| 107 | * Share the same dataset sequence mappings (if any). | |
| 108 | */ | |
| 109 | 4 | if (dataset == null && al.getDataset() == null) |
| 110 | { | |
| 111 | 1 | this.setCodonFrames(al.getCodonFrames()); |
| 112 | } | |
| 113 | } | |
| 114 | ||
| 115 | /** | |
| 116 | * Make an alignment from an array of Sequences. | |
| 117 | * | |
| 118 | * @param sequences | |
| 119 | */ | |
| 120 | 1205 | public Alignment(SequenceI[] seqs) |
| 121 | { | |
| 122 | 1205 | initAlignment(seqs); |
| 123 | } | |
| 124 | ||
| 125 | /** | |
| 126 | * Make a new alignment from an array of SeqCigars | |
| 127 | * | |
| 128 | * @param seqs | |
| 129 | * SeqCigar[] | |
| 130 | */ | |
| 131 | 2 | public Alignment(SeqCigar[] alseqs) |
| 132 | { | |
| 133 | 2 | SequenceI[] seqs = SeqCigar.createAlignmentSequences(alseqs, |
| 134 | gapCharacter, new HiddenColumns(), null); | |
| 135 | 2 | initAlignment(seqs); |
| 136 | } | |
| 137 | ||
| 138 | /** | |
| 139 | * Make a new alignment from an CigarArray JBPNote - can only do this when | |
| 140 | * compactAlignment does not contain hidden regions. JBPNote - must also check | |
| 141 | * that compactAlignment resolves to a set of SeqCigars - or construct them | |
| 142 | * appropriately. | |
| 143 | * | |
| 144 | * @param compactAlignment | |
| 145 | * CigarArray | |
| 146 | */ | |
| 147 | 0 | public static AlignmentI createAlignment(CigarArray compactAlignment) |
| 148 | { | |
| 149 | 0 | throw new Error(MessageManager |
| 150 | .getString("error.alignment_cigararray_not_implemented")); | |
| 151 | // this(compactAlignment.refCigars); | |
| 152 | } | |
| 153 | ||
| 154 | 64102 | @Override |
| 155 | public List<SequenceI> getSequences() | |
| 156 | { | |
| 157 | 64102 | return sequences; |
| 158 | } | |
| 159 | ||
| 160 | 12 | @Override |
| 161 | public List<SequenceI> getSequences( | |
| 162 | Map<SequenceI, SequenceCollectionI> hiddenReps) | |
| 163 | { | |
| 164 | // TODO: in jalview 2.8 we don't do anything with hiddenreps - fix design to | |
| 165 | // work on this. | |
| 166 | 12 | return sequences; |
| 167 | } | |
| 168 | ||
| 169 | 27154 | @Override |
| 170 | public SequenceI[] getSequencesArray() | |
| 171 | { | |
| 172 | 27154 | if (sequences == null) |
| 173 | { | |
| 174 | 0 | return null; |
| 175 | } | |
| 176 | 27154 | synchronized (sequences) |
| 177 | { | |
| 178 | 27154 | return sequences.toArray(new SequenceI[sequences.size()]); |
| 179 | } | |
| 180 | } | |
| 181 | ||
| 182 | /** | |
| 183 | * Returns a map of lists of sequences keyed by sequence name. | |
| 184 | * | |
| 185 | * @return | |
| 186 | */ | |
| 187 | 0 | @Override |
| 188 | public Map<String, List<SequenceI>> getSequencesByName() | |
| 189 | { | |
| 190 | 0 | return AlignmentUtils.getSequencesByName(this); |
| 191 | } | |
| 192 | ||
| 193 | 723331 | @Override |
| 194 | public SequenceI getSequenceAt(int i) | |
| 195 | { | |
| 196 | 723331 | synchronized (sequences) |
| 197 | { | |
| 198 | 723334 | if (i > -1 && i < sequences.size()) |
| 199 | { | |
| 200 | 723322 | return sequences.get(i); |
| 201 | } | |
| 202 | } | |
| 203 | ||
| 204 | 12 | return null; |
| 205 | } | |
| 206 | ||
| 207 | 0 | @Override |
| 208 | public SequenceI getSequenceAtAbsoluteIndex(int i) | |
| 209 | { | |
| 210 | 0 | SequenceI seq = null; |
| 211 | 0 | if (getHiddenSequences().getSize() > 0) |
| 212 | { | |
| 213 | 0 | seq = getHiddenSequences().getHiddenSequence(i); |
| 214 | 0 | if (seq == null) |
| 215 | { | |
| 216 | // didn't find the sequence in the hidden sequences, get it from the | |
| 217 | // alignment | |
| 218 | 0 | int index = getHiddenSequences().findIndexWithoutHiddenSeqs(i); |
| 219 | 0 | seq = getSequenceAt(index); |
| 220 | } | |
| 221 | } | |
| 222 | else | |
| 223 | { | |
| 224 | 0 | seq = getSequenceAt(i); |
| 225 | } | |
| 226 | 0 | return seq; |
| 227 | } | |
| 228 | ||
| 229 | /** | |
| 230 | * Adds a sequence to the alignment. Recalculates maxLength and size. Note | |
| 231 | * this currently does not recalculate whether or not the alignment is | |
| 232 | * nucleotide, so mixed alignments may have undefined behaviour. | |
| 233 | * | |
| 234 | * @param snew | |
| 235 | */ | |
| 236 | 240 | @Override |
| 237 | public void addSequence(SequenceI snew) | |
| 238 | { | |
| 239 | 240 | if (dataset != null) |
| 240 | { | |
| 241 | ||
| 242 | // maintain dataset integrity | |
| 243 | 12 | SequenceI dsseq = snew.getDatasetSequence(); |
| 244 | 12 | if (dsseq == null) |
| 245 | { | |
| 246 | // derive new sequence | |
| 247 | 5 | SequenceI adding = snew.deriveSequence(); |
| 248 | 5 | snew = adding; |
| 249 | 5 | dsseq = snew.getDatasetSequence(); |
| 250 | } | |
| 251 | 12 | if (getDataset().findIndex(dsseq) == -1) |
| 252 | { | |
| 253 | 11 | getDataset().addSequence(dsseq); |
| 254 | } | |
| 255 | ||
| 256 | } | |
| 257 | 240 | if (sequences == null) |
| 258 | { | |
| 259 | 0 | initAlignment(new SequenceI[] { snew }); |
| 260 | } | |
| 261 | else | |
| 262 | { | |
| 263 | 240 | synchronized (sequences) |
| 264 | { | |
| 265 | 240 | sequences.add(snew); |
| 266 | } | |
| 267 | } | |
| 268 | 240 | if (hiddenSequences != null) |
| 269 | { | |
| 270 | 240 | hiddenSequences.adjustHeightSequenceAdded(); |
| 271 | } | |
| 272 | } | |
| 273 | ||
| 274 | 0 | @Override |
| 275 | public SequenceI replaceSequenceAt(int i, SequenceI snew) | |
| 276 | { | |
| 277 | 0 | synchronized (sequences) |
| 278 | { | |
| 279 | 0 | if (sequences.size() > i) |
| 280 | { | |
| 281 | 0 | return sequences.set(i, snew); |
| 282 | ||
| 283 | } | |
| 284 | else | |
| 285 | { | |
| 286 | 0 | sequences.add(snew); |
| 287 | 0 | hiddenSequences.adjustHeightSequenceAdded(); |
| 288 | } | |
| 289 | 0 | return null; |
| 290 | } | |
| 291 | } | |
| 292 | ||
| 293 | /** | |
| 294 | * DOCUMENT ME! | |
| 295 | * | |
| 296 | * @return DOCUMENT ME! | |
| 297 | */ | |
| 298 | 4948 | @Override |
| 299 | public List<SequenceGroup> getGroups() | |
| 300 | { | |
| 301 | 4948 | return groups; |
| 302 | } | |
| 303 | ||
| 304 | 977 | @Override |
| 305 | public void finalize() throws Throwable | |
| 306 | { | |
| 307 | 977 | if (getDataset() != null) |
| 308 | { | |
| 309 | 245 | getDataset().removeAlignmentRef(); |
| 310 | } | |
| 311 | ||
| 312 | 977 | nullReferences(); |
| 313 | 977 | super.finalize(); |
| 314 | } | |
| 315 | ||
| 316 | /** | |
| 317 | * Defensively nulls out references in case this object is not garbage | |
| 318 | * collected | |
| 319 | */ | |
| 320 | 1191 | void nullReferences() |
| 321 | { | |
| 322 | 1191 | dataset = null; |
| 323 | 1191 | sequences = null; |
| 324 | 1191 | groups = null; |
| 325 | 1191 | annotations = null; |
| 326 | 1191 | hiddenSequences = null; |
| 327 | } | |
| 328 | ||
| 329 | /** | |
| 330 | * decrement the alignmentRefs counter by one and null references if it goes | |
| 331 | * to zero. | |
| 332 | * | |
| 333 | * @throws Throwable | |
| 334 | */ | |
| 335 | 245 | private void removeAlignmentRef() throws Throwable |
| 336 | { | |
| 337 | 245 | if (--alignmentRefs == 0) |
| 338 | { | |
| 339 | 214 | nullReferences(); |
| 340 | } | |
| 341 | } | |
| 342 | ||
| 343 | 11 | @Override |
| 344 | public void deleteSequence(SequenceI s) | |
| 345 | { | |
| 346 | 11 | synchronized (sequences) |
| 347 | { | |
| 348 | 11 | deleteSequence(findIndex(s)); |
| 349 | } | |
| 350 | } | |
| 351 | ||
| 352 | 16 | @Override |
| 353 | public void deleteSequence(int i) | |
| 354 | { | |
| 355 | 16 | synchronized (sequences) |
| 356 | { | |
| 357 | 16 | if (i > -1 && i < getHeight()) |
| 358 | { | |
| 359 | 14 | sequences.remove(i); |
| 360 | 14 | hiddenSequences.adjustHeightSequenceDeleted(i); |
| 361 | } | |
| 362 | } | |
| 363 | } | |
| 364 | ||
| 365 | 616 | @Override |
| 366 | public void deleteHiddenSequence(int i) | |
| 367 | { | |
| 368 | 616 | synchronized (sequences) |
| 369 | { | |
| 370 | 616 | if (i > -1 && i < getHeight()) |
| 371 | { | |
| 372 | 616 | sequences.remove(i); |
| 373 | } | |
| 374 | } | |
| 375 | } | |
| 376 | ||
| 377 | /* | |
| 378 | * (non-Javadoc) | |
| 379 | * | |
| 380 | * @see jalview.datamodel.AlignmentI#findGroup(jalview.datamodel.SequenceI) | |
| 381 | */ | |
| 382 | 19 | @Override |
| 383 | public SequenceGroup findGroup(SequenceI seq, int position) | |
| 384 | { | |
| 385 | 19 | synchronized (groups) |
| 386 | { | |
| 387 | 19 | for (SequenceGroup sg : groups) |
| 388 | { | |
| 389 | 23 | if (sg.getSequences(null).contains(seq)) |
| 390 | { | |
| 391 | 16 | if (position >= sg.getStartRes() && position <= sg.getEndRes()) |
| 392 | { | |
| 393 | 10 | return sg; |
| 394 | } | |
| 395 | } | |
| 396 | } | |
| 397 | } | |
| 398 | 9 | return null; |
| 399 | } | |
| 400 | ||
| 401 | /* | |
| 402 | * (non-Javadoc) | |
| 403 | * | |
| 404 | * @see | |
| 405 | * jalview.datamodel.AlignmentI#findAllGroups(jalview.datamodel.SequenceI) | |
| 406 | */ | |
| 407 | 9508 | @Override |
| 408 | public SequenceGroup[] findAllGroups(SequenceI s) | |
| 409 | { | |
| 410 | 9508 | ArrayList<SequenceGroup> temp = new ArrayList<>(); |
| 411 | ||
| 412 | 9508 | synchronized (groups) |
| 413 | { | |
| 414 | 9508 | int gSize = groups.size(); |
| 415 | 14824 | for (int i = 0; i < gSize; i++) |
| 416 | { | |
| 417 | 5316 | SequenceGroup sg = groups.get(i); |
| 418 | 5316 | if (sg == null || sg.getSequences() == null) |
| 419 | { | |
| 420 | 0 | this.deleteGroup(sg); |
| 421 | 0 | gSize--; |
| 422 | 0 | continue; |
| 423 | } | |
| 424 | ||
| 425 | 5316 | if (sg.getSequences().contains(s)) |
| 426 | { | |
| 427 | 4613 | temp.add(sg); |
| 428 | } | |
| 429 | } | |
| 430 | } | |
| 431 | 9508 | SequenceGroup[] ret = new SequenceGroup[temp.size()]; |
| 432 | 9508 | return temp.toArray(ret); |
| 433 | } | |
| 434 | ||
| 435 | /** */ | |
| 436 | 114 | @Override |
| 437 | public void addGroup(SequenceGroup sg) | |
| 438 | { | |
| 439 | 114 | synchronized (groups) |
| 440 | { | |
| 441 | 114 | if (!groups.contains(sg)) |
| 442 | { | |
| 443 | 105 | if (hiddenSequences.getSize() > 0) |
| 444 | { | |
| 445 | 0 | int i, iSize = sg.getSize(); |
| 446 | 0 | for (i = 0; i < iSize; i++) |
| 447 | { | |
| 448 | 0 | if (!sequences.contains(sg.getSequenceAt(i))) |
| 449 | { | |
| 450 | 0 | sg.deleteSequence(sg.getSequenceAt(i), false); |
| 451 | 0 | iSize--; |
| 452 | 0 | i--; |
| 453 | } | |
| 454 | } | |
| 455 | ||
| 456 | 0 | if (sg.getSize() < 1) |
| 457 | { | |
| 458 | 0 | return; |
| 459 | } | |
| 460 | } | |
| 461 | 105 | sg.setContext(this, true); |
| 462 | 105 | groups.add(sg); |
| 463 | } | |
| 464 | } | |
| 465 | } | |
| 466 | ||
| 467 | /** | |
| 468 | * remove any annotation that references gp | |
| 469 | * | |
| 470 | * @param gp | |
| 471 | * (if null, removes all group associated annotation) | |
| 472 | */ | |
| 473 | 0 | private void removeAnnotationForGroup(SequenceGroup gp) |
| 474 | { | |
| 475 | 0 | if (annotations == null || annotations.length == 0) |
| 476 | { | |
| 477 | 0 | return; |
| 478 | } | |
| 479 | // remove annotation very quickly | |
| 480 | 0 | AlignmentAnnotation[] t, |
| 481 | todelete = new AlignmentAnnotation[annotations.length], | |
| 482 | tokeep = new AlignmentAnnotation[annotations.length]; | |
| 483 | 0 | int i, p, k; |
| 484 | 0 | if (gp == null) |
| 485 | { | |
| 486 | 0 | for (i = 0, p = 0, k = 0; i < annotations.length; i++) |
| 487 | { | |
| 488 | 0 | if (annotations[i].groupRef != null) |
| 489 | { | |
| 490 | 0 | todelete[p++] = annotations[i]; |
| 491 | } | |
| 492 | else | |
| 493 | { | |
| 494 | 0 | tokeep[k++] = annotations[i]; |
| 495 | } | |
| 496 | } | |
| 497 | } | |
| 498 | else | |
| 499 | { | |
| 500 | 0 | for (i = 0, p = 0, k = 0; i < annotations.length; i++) |
| 501 | { | |
| 502 | 0 | if (annotations[i].groupRef == gp) |
| 503 | { | |
| 504 | 0 | todelete[p++] = annotations[i]; |
| 505 | } | |
| 506 | else | |
| 507 | { | |
| 508 | 0 | tokeep[k++] = annotations[i]; |
| 509 | } | |
| 510 | } | |
| 511 | } | |
| 512 | 0 | if (p > 0) |
| 513 | { | |
| 514 | // clear out the group associated annotation. | |
| 515 | 0 | for (i = 0; i < p; i++) |
| 516 | { | |
| 517 | 0 | unhookAnnotation(todelete[i]); |
| 518 | 0 | todelete[i] = null; |
| 519 | } | |
| 520 | 0 | t = new AlignmentAnnotation[k]; |
| 521 | 0 | for (i = 0; i < k; i++) |
| 522 | { | |
| 523 | 0 | t[i] = tokeep[i]; |
| 524 | } | |
| 525 | 0 | annotations = t; |
| 526 | } | |
| 527 | } | |
| 528 | ||
| 529 | 0 | @Override |
| 530 | public void deleteAllGroups() | |
| 531 | { | |
| 532 | 0 | synchronized (groups) |
| 533 | { | |
| 534 | 0 | if (annotations != null) |
| 535 | { | |
| 536 | 0 | removeAnnotationForGroup(null); |
| 537 | } | |
| 538 | 0 | for (SequenceGroup sg : groups) |
| 539 | { | |
| 540 | 0 | sg.setContext(null, false); |
| 541 | } | |
| 542 | 0 | groups.clear(); |
| 543 | } | |
| 544 | } | |
| 545 | ||
| 546 | /** */ | |
| 547 | 0 | @Override |
| 548 | public void deleteGroup(SequenceGroup g) | |
| 549 | { | |
| 550 | 0 | synchronized (groups) |
| 551 | { | |
| 552 | 0 | if (groups.contains(g)) |
| 553 | { | |
| 554 | 0 | removeAnnotationForGroup(g); |
| 555 | 0 | groups.remove(g); |
| 556 | 0 | g.setContext(null, false); |
| 557 | } | |
| 558 | } | |
| 559 | } | |
| 560 | ||
| 561 | /** */ | |
| 562 | 283 | @Override |
| 563 | public SequenceI findName(String name) | |
| 564 | { | |
| 565 | 283 | return findName(name, false); |
| 566 | } | |
| 567 | ||
| 568 | /* | |
| 569 | * (non-Javadoc) | |
| 570 | * | |
| 571 | * @see jalview.datamodel.AlignmentI#findName(java.lang.String, boolean) | |
| 572 | */ | |
| 573 | 969 | @Override |
| 574 | public SequenceI findName(String token, boolean b) | |
| 575 | { | |
| 576 | 969 | return findName(null, token, b); |
| 577 | } | |
| 578 | ||
| 579 | /* | |
| 580 | * (non-Javadoc) | |
| 581 | * | |
| 582 | * @see jalview.datamodel.AlignmentI#findName(SequenceI, java.lang.String, | |
| 583 | * boolean) | |
| 584 | */ | |
| 585 | 1498 | @Override |
| 586 | public SequenceI findName(SequenceI startAfter, String token, boolean b) | |
| 587 | { | |
| 588 | ||
| 589 | 1498 | int i = 0; |
| 590 | 1498 | SequenceI sq = null; |
| 591 | 1498 | String sqname = null; |
| 592 | 1498 | if (startAfter != null) |
| 593 | { | |
| 594 | // try to find the sequence in the alignment | |
| 595 | 529 | boolean matched = false; |
| 596 | 4762 | while (i < sequences.size()) |
| 597 | { | |
| 598 | 4747 | if (getSequenceAt(i++) == startAfter) |
| 599 | { | |
| 600 | 514 | matched = true; |
| 601 | 514 | break; |
| 602 | } | |
| 603 | } | |
| 604 | 529 | if (!matched) |
| 605 | { | |
| 606 | 15 | i = 0; |
| 607 | } | |
| 608 | } | |
| 609 | 12656 | while (i < sequences.size()) |
| 610 | { | |
| 611 | 12098 | sq = getSequenceAt(i); |
| 612 | 12098 | sqname = sq.getName(); |
| 613 | 12098 | if (sqname.equals(token) // exact match |
| 614 | || (b && // allow imperfect matches - case varies | |
| 615 | (sqname.equalsIgnoreCase(token)))) | |
| 616 | { | |
| 617 | 940 | return getSequenceAt(i); |
| 618 | } | |
| 619 | ||
| 620 | 11158 | i++; |
| 621 | } | |
| 622 | ||
| 623 | 558 | return null; |
| 624 | } | |
| 625 | ||
| 626 | 11 | @Override |
| 627 | public SequenceI[] findSequenceMatch(String name) | |
| 628 | { | |
| 629 | 11 | Vector matches = new Vector(); |
| 630 | 11 | int i = 0; |
| 631 | ||
| 632 | 64 | while (i < sequences.size()) |
| 633 | { | |
| 634 | 53 | if (getSequenceAt(i).getName().equals(name)) |
| 635 | { | |
| 636 | 5 | matches.addElement(getSequenceAt(i)); |
| 637 | } | |
| 638 | 53 | i++; |
| 639 | } | |
| 640 | ||
| 641 | 11 | SequenceI[] result = new SequenceI[matches.size()]; |
| 642 | 16 | for (i = 0; i < result.length; i++) |
| 643 | { | |
| 644 | 5 | result[i] = (SequenceI) matches.elementAt(i); |
| 645 | } | |
| 646 | ||
| 647 | 11 | return result; |
| 648 | ||
| 649 | } | |
| 650 | ||
| 651 | /* | |
| 652 | * (non-Javadoc) | |
| 653 | * | |
| 654 | * @see jalview.datamodel.AlignmentI#findIndex(jalview.datamodel.SequenceI) | |
| 655 | */ | |
| 656 | 1446 | @Override |
| 657 | public int findIndex(SequenceI s) | |
| 658 | { | |
| 659 | 1446 | int i = 0; |
| 660 | ||
| 661 | 65711 | while (i < sequences.size()) |
| 662 | { | |
| 663 | 65492 | if (s == getSequenceAt(i)) |
| 664 | { | |
| 665 | 1227 | return i; |
| 666 | } | |
| 667 | ||
| 668 | 64265 | i++; |
| 669 | } | |
| 670 | ||
| 671 | 219 | return -1; |
| 672 | } | |
| 673 | ||
| 674 | /* | |
| 675 | * (non-Javadoc) | |
| 676 | * | |
| 677 | * @see | |
| 678 | * jalview.datamodel.AlignmentI#findIndex(jalview.datamodel.SearchResults) | |
| 679 | */ | |
| 680 | 3 | @Override |
| 681 | public int findIndex(SearchResultsI results) | |
| 682 | { | |
| 683 | 3 | int i = 0; |
| 684 | ||
| 685 | 10 | while (i < sequences.size()) |
| 686 | { | |
| 687 | 8 | if (results.involvesSequence(getSequenceAt(i))) |
| 688 | { | |
| 689 | 1 | return i; |
| 690 | } | |
| 691 | 7 | i++; |
| 692 | } | |
| 693 | 2 | return -1; |
| 694 | } | |
| 695 | ||
| 696 | 16085 | @Override |
| 697 | public int getHeight() | |
| 698 | { | |
| 699 | 16085 | return sequences.size(); |
| 700 | } | |
| 701 | ||
| 702 | 10 | @Override |
| 703 | public int getAbsoluteHeight() | |
| 704 | { | |
| 705 | 10 | return sequences.size() + getHiddenSequences().getSize(); |
| 706 | } | |
| 707 | ||
| 708 | 8180 | @Override |
| 709 | public int getWidth() | |
| 710 | { | |
| 711 | 8180 | int maxLength = -1; |
| 712 | ||
| 713 | 588469 | for (int i = 0; i < sequences.size(); i++) |
| 714 | { | |
| 715 | 580289 | maxLength = Math.max(maxLength, getSequenceAt(i).getLength()); |
| 716 | } | |
| 717 | 8180 | return maxLength; |
| 718 | } | |
| 719 | ||
| 720 | /** | |
| 721 | * DOCUMENT ME! | |
| 722 | * | |
| 723 | * @param gc | |
| 724 | * DOCUMENT ME! | |
| 725 | */ | |
| 726 | 270 | @Override |
| 727 | public void setGapCharacter(char gc) | |
| 728 | { | |
| 729 | 270 | gapCharacter = gc; |
| 730 | 270 | synchronized (sequences) |
| 731 | { | |
| 732 | 270 | for (SequenceI seq : sequences) |
| 733 | { | |
| 734 | 2370 | seq.setSequence(seq.getSequenceAsString().replace('.', gc) |
| 735 | .replace('-', gc).replace(' ', gc)); | |
| 736 | } | |
| 737 | } | |
| 738 | } | |
| 739 | ||
| 740 | /** | |
| 741 | * DOCUMENT ME! | |
| 742 | * | |
| 743 | * @return DOCUMENT ME! | |
| 744 | */ | |
| 745 | 305 | @Override |
| 746 | public char getGapCharacter() | |
| 747 | { | |
| 748 | 305 | return gapCharacter; |
| 749 | } | |
| 750 | ||
| 751 | /* | |
| 752 | * (non-Javadoc) | |
| 753 | * | |
| 754 | * @see jalview.datamodel.AlignmentI#isAligned() | |
| 755 | */ | |
| 756 | 0 | @Override |
| 757 | public boolean isAligned() | |
| 758 | { | |
| 759 | 0 | return isAligned(false); |
| 760 | } | |
| 761 | ||
| 762 | /* | |
| 763 | * (non-Javadoc) | |
| 764 | * | |
| 765 | * @see jalview.datamodel.AlignmentI#isAligned(boolean) | |
| 766 | */ | |
| 767 | 0 | @Override |
| 768 | public boolean isAligned(boolean includeHidden) | |
| 769 | { | |
| 770 | 0 | int width = getWidth(); |
| 771 | 0 | if (hiddenSequences == null || hiddenSequences.getSize() == 0) |
| 772 | { | |
| 773 | 0 | includeHidden = true; // no hidden sequences to check against. |
| 774 | } | |
| 775 | 0 | for (int i = 0; i < sequences.size(); i++) |
| 776 | { | |
| 777 | 0 | if (includeHidden || !hiddenSequences.isHidden(getSequenceAt(i))) |
| 778 | { | |
| 779 | 0 | if (getSequenceAt(i).getLength() != width) |
| 780 | { | |
| 781 | 0 | return false; |
| 782 | } | |
| 783 | } | |
| 784 | } | |
| 785 | ||
| 786 | 0 | return true; |
| 787 | } | |
| 788 | ||
| 789 | 0 | @Override |
| 790 | public boolean isHidden(int alignmentIndex) | |
| 791 | { | |
| 792 | 0 | return (getHiddenSequences().getHiddenSequence(alignmentIndex) != null); |
| 793 | } | |
| 794 | ||
| 795 | /** | |
| 796 | * Delete all annotations, including auto-calculated if the flag is set true. | |
| 797 | * Returns true if at least one annotation was deleted, else false. | |
| 798 | * | |
| 799 | * @param includingAutoCalculated | |
| 800 | * @return | |
| 801 | */ | |
| 802 | 2 | @Override |
| 803 | public boolean deleteAllAnnotations(boolean includingAutoCalculated) | |
| 804 | { | |
| 805 | 2 | boolean result = false; |
| 806 | 2 | for (AlignmentAnnotation alan : getAlignmentAnnotation()) |
| 807 | { | |
| 808 | 8 | if (!alan.autoCalculated || includingAutoCalculated) |
| 809 | { | |
| 810 | 7 | deleteAnnotation(alan); |
| 811 | 7 | result = true; |
| 812 | } | |
| 813 | } | |
| 814 | 2 | return result; |
| 815 | } | |
| 816 | ||
| 817 | /* | |
| 818 | * (non-Javadoc) | |
| 819 | * | |
| 820 | * @seejalview.datamodel.AlignmentI#deleteAnnotation(jalview.datamodel. | |
| 821 | * AlignmentAnnotation) | |
| 822 | */ | |
| 823 | 7 | @Override |
| 824 | public boolean deleteAnnotation(AlignmentAnnotation aa) | |
| 825 | { | |
| 826 | 7 | return deleteAnnotation(aa, true); |
| 827 | } | |
| 828 | ||
| 829 | 128 | @Override |
| 830 | public boolean deleteAnnotation(AlignmentAnnotation aa, boolean unhook) | |
| 831 | { | |
| 832 | 128 | int aSize = 1; |
| 833 | ||
| 834 | 128 | if (annotations != null) |
| 835 | { | |
| 836 | 128 | aSize = annotations.length; |
| 837 | } | |
| 838 | ||
| 839 | 128 | if (aSize < 1) |
| 840 | { | |
| 841 | 0 | return false; |
| 842 | } | |
| 843 | ||
| 844 | 128 | AlignmentAnnotation[] temp = new AlignmentAnnotation[aSize - 1]; |
| 845 | ||
| 846 | 128 | boolean swap = false; |
| 847 | 128 | int tIndex = 0; |
| 848 | ||
| 849 | 1109 | for (int i = 0; i < aSize; i++) |
| 850 | { | |
| 851 | 981 | if (annotations[i] == aa) |
| 852 | { | |
| 853 | 128 | swap = true; |
| 854 | 128 | continue; |
| 855 | } | |
| 856 | 853 | if (tIndex < temp.length) |
| 857 | { | |
| 858 | 853 | temp[tIndex++] = annotations[i]; |
| 859 | } | |
| 860 | } | |
| 861 | ||
| 862 | 128 | if (swap) |
| 863 | { | |
| 864 | 128 | annotations = temp; |
| 865 | 128 | if (unhook) |
| 866 | { | |
| 867 | 7 | unhookAnnotation(aa); |
| 868 | } | |
| 869 | } | |
| 870 | 128 | return swap; |
| 871 | } | |
| 872 | ||
| 873 | /** | |
| 874 | * remove any object references associated with this annotation | |
| 875 | * | |
| 876 | * @param aa | |
| 877 | */ | |
| 878 | 7 | private void unhookAnnotation(AlignmentAnnotation aa) |
| 879 | { | |
| 880 | 7 | if (aa.sequenceRef != null) |
| 881 | { | |
| 882 | 6 | aa.sequenceRef.removeAlignmentAnnotation(aa); |
| 883 | } | |
| 884 | 7 | if (aa.groupRef != null) |
| 885 | { | |
| 886 | // probably need to do more here in the future (post 2.5.0) | |
| 887 | 0 | aa.groupRef = null; |
| 888 | } | |
| 889 | } | |
| 890 | ||
| 891 | /* | |
| 892 | * (non-Javadoc) | |
| 893 | * | |
| 894 | * @seejalview.datamodel.AlignmentI#addAnnotation(jalview.datamodel. | |
| 895 | * AlignmentAnnotation) | |
| 896 | */ | |
| 897 | 2037 | @Override |
| 898 | public void addAnnotation(AlignmentAnnotation aa) | |
| 899 | { | |
| 900 | 2038 | addAnnotation(aa, -1); |
| 901 | } | |
| 902 | ||
| 903 | /* | |
| 904 | * (non-Javadoc) | |
| 905 | * | |
| 906 | * @seejalview.datamodel.AlignmentI#addAnnotation(jalview.datamodel. | |
| 907 | * AlignmentAnnotation, int) | |
| 908 | */ | |
| 909 | 2194 | @Override |
| 910 | public void addAnnotation(AlignmentAnnotation aa, int pos) | |
| 911 | { | |
| 912 | 2195 | if (aa.getRNAStruc() != null) |
| 913 | { | |
| 914 | 398 | hasRNAStructure = true; |
| 915 | } | |
| 916 | ||
| 917 | 2195 | int aSize = 1; |
| 918 | 2195 | if (annotations != null) |
| 919 | { | |
| 920 | 1890 | aSize = annotations.length + 1; |
| 921 | } | |
| 922 | ||
| 923 | 2195 | AlignmentAnnotation[] temp = new AlignmentAnnotation[aSize]; |
| 924 | 2194 | int i = 0; |
| 925 | 2194 | if (pos == -1 || pos >= aSize) |
| 926 | { | |
| 927 | 2038 | temp[aSize - 1] = aa; |
| 928 | } | |
| 929 | else | |
| 930 | { | |
| 931 | 157 | temp[pos] = aa; |
| 932 | } | |
| 933 | 2194 | if (aSize > 1) |
| 934 | { | |
| 935 | 1884 | int p = 0; |
| 936 | 21007 | for (i = 0; i < (aSize - 1); i++, p++) |
| 937 | { | |
| 938 | 19123 | if (p == pos) |
| 939 | { | |
| 940 | 66 | p++; |
| 941 | } | |
| 942 | 19123 | if (p < temp.length) |
| 943 | { | |
| 944 | 19123 | temp[p] = annotations[i]; |
| 945 | } | |
| 946 | } | |
| 947 | } | |
| 948 | ||
| 949 | 2195 | annotations = temp; |
| 950 | } | |
| 951 | ||
| 952 | 193 | @Override |
| 953 | public void setAnnotationIndex(AlignmentAnnotation aa, int index) | |
| 954 | { | |
| 955 | 193 | if (aa == null || annotations == null || annotations.length - 1 < index) |
| 956 | { | |
| 957 | 0 | return; |
| 958 | } | |
| 959 | ||
| 960 | 193 | int aSize = annotations.length; |
| 961 | 193 | AlignmentAnnotation[] temp = new AlignmentAnnotation[aSize]; |
| 962 | ||
| 963 | 193 | temp[index] = aa; |
| 964 | ||
| 965 | 3139 | for (int i = 0; i < aSize; i++) |
| 966 | { | |
| 967 | 2946 | if (i == index) |
| 968 | { | |
| 969 | 193 | continue; |
| 970 | } | |
| 971 | ||
| 972 | 2753 | if (i < index) |
| 973 | { | |
| 974 | 2633 | temp[i] = annotations[i]; |
| 975 | } | |
| 976 | else | |
| 977 | { | |
| 978 | 120 | temp[i] = annotations[i - 1]; |
| 979 | } | |
| 980 | } | |
| 981 | ||
| 982 | 193 | annotations = temp; |
| 983 | } | |
| 984 | ||
| 985 | 30531 | @Override |
| 986 | /** | |
| 987 | * returns all annotation on the alignment | |
| 988 | */ | |
| 989 | public AlignmentAnnotation[] getAlignmentAnnotation() | |
| 990 | { | |
| 991 | 30530 | return annotations; |
| 992 | } | |
| 993 | ||
| 994 | 6225 | @Override |
| 995 | public boolean isNucleotide() | |
| 996 | { | |
| 997 | 6225 | return nucleotide; |
| 998 | } | |
| 999 | ||
| 1000 | 514 | @Override |
| 1001 | public boolean hasRNAStructure() | |
| 1002 | { | |
| 1003 | // TODO can it happen that structure is removed from alignment? | |
| 1004 | 514 | return hasRNAStructure; |
| 1005 | } | |
| 1006 | ||
| 1007 | 356 | @Override |
| 1008 | public void setDataset(AlignmentI data) | |
| 1009 | { | |
| 1010 | 356 | if (dataset == null && data == null) |
| 1011 | { | |
| 1012 | 314 | createDatasetAlignment(); |
| 1013 | } | |
| 1014 | 42 | else if (dataset == null && data != null) |
| 1015 | { | |
| 1016 | 27 | if (data == this) |
| 1017 | { | |
| 1018 | 1 | throw new IllegalArgumentException("Circular dataset reference"); |
| 1019 | } | |
| 1020 | 26 | if (!(data instanceof Alignment)) |
| 1021 | { | |
| 1022 | 0 | throw new Error( |
| 1023 | "Implementation Error: jalview.datamodel.Alignment does not yet support other implementations of AlignmentI as its dataset reference"); | |
| 1024 | } | |
| 1025 | 26 | dataset = (Alignment) data; |
| 1026 | 370 | for (int i = 0; i < getHeight(); i++) |
| 1027 | { | |
| 1028 | 344 | SequenceI currentSeq = getSequenceAt(i); |
| 1029 | 344 | SequenceI dsq = currentSeq.getDatasetSequence(); |
| 1030 | 344 | if (dsq == null) |
| 1031 | { | |
| 1032 | 0 | dsq = currentSeq.createDatasetSequence(); |
| 1033 | 0 | dataset.addSequence(dsq); |
| 1034 | } | |
| 1035 | else | |
| 1036 | { | |
| 1037 | 344 | while (dsq.getDatasetSequence() != null) |
| 1038 | { | |
| 1039 | 0 | dsq = dsq.getDatasetSequence(); |
| 1040 | } | |
| 1041 | 344 | if (dataset.findIndex(dsq) == -1) |
| 1042 | { | |
| 1043 | 174 | dataset.addSequence(dsq); |
| 1044 | } | |
| 1045 | } | |
| 1046 | } | |
| 1047 | } | |
| 1048 | 355 | dataset.addAlignmentRef(); |
| 1049 | } | |
| 1050 | ||
| 1051 | /** | |
| 1052 | * add dataset sequences to seq for currentSeq and any sequences it references | |
| 1053 | */ | |
| 1054 | 2313 | private void resolveAndAddDatasetSeq(SequenceI currentSeq, |
| 1055 | Set<SequenceI> seqs, boolean createDatasetSequence) | |
| 1056 | { | |
| 1057 | 2313 | SequenceI alignedSeq = currentSeq; |
| 1058 | 2313 | if (currentSeq.getDatasetSequence() != null) |
| 1059 | { | |
| 1060 | 1074 | currentSeq = currentSeq.getDatasetSequence(); |
| 1061 | } | |
| 1062 | else | |
| 1063 | { | |
| 1064 | 1239 | if (createDatasetSequence) |
| 1065 | { | |
| 1066 | 1239 | currentSeq = currentSeq.createDatasetSequence(); |
| 1067 | } | |
| 1068 | } | |
| 1069 | ||
| 1070 | 2313 | List<SequenceI> toProcess = new ArrayList<>(); |
| 1071 | 2313 | toProcess.add(currentSeq); |
| 1072 | 4632 | while (toProcess.size() > 0) |
| 1073 | { | |
| 1074 | // use a queue ? | |
| 1075 | 2319 | SequenceI curDs = toProcess.remove(0); |
| 1076 | ||
| 1077 | 2319 | if (!seqs.add(curDs)) |
| 1078 | { | |
| 1079 | 108 | continue; |
| 1080 | } | |
| 1081 | // iterate over database references, making sure we add forward referenced | |
| 1082 | // sequences | |
| 1083 | 2211 | if (curDs.getDBRefs() != null) |
| 1084 | { | |
| 1085 | 696 | for (DBRefEntry dbr : curDs.getDBRefs()) |
| 1086 | { | |
| 1087 | 838 | if (dbr.getMap() != null && dbr.getMap().getTo() != null) |
| 1088 | { | |
| 1089 | 6 | if (dbr.getMap().getTo() == alignedSeq) |
| 1090 | { | |
| 1091 | /* | |
| 1092 | * update mapping to be to the newly created dataset sequence | |
| 1093 | */ | |
| 1094 | 2 | dbr.getMap().setTo(currentSeq); |
| 1095 | } | |
| 1096 | 6 | if (dbr.getMap().getTo().getDatasetSequence() != null) |
| 1097 | { | |
| 1098 | 0 | throw new Error("Implementation error: Map.getTo() for dbref " |
| 1099 | + dbr + " from " + curDs.getName() | |
| 1100 | + " is not a dataset sequence."); | |
| 1101 | } | |
| 1102 | // we recurse to add all forward references to dataset sequences via | |
| 1103 | // DBRefs/etc | |
| 1104 | 6 | toProcess.add(dbr.getMap().getTo()); |
| 1105 | } | |
| 1106 | } | |
| 1107 | } | |
| 1108 | } | |
| 1109 | } | |
| 1110 | ||
| 1111 | /** | |
| 1112 | * Creates a new dataset for this alignment. Can only be done once - if | |
| 1113 | * dataset is not null this will not be performed. | |
| 1114 | */ | |
| 1115 | 328 | public void createDatasetAlignment() |
| 1116 | { | |
| 1117 | 328 | if (dataset != null) |
| 1118 | { | |
| 1119 | 0 | return; |
| 1120 | } | |
| 1121 | // try to avoid using SequenceI.equals at this stage, it will be expensive | |
| 1122 | 328 | Set<SequenceI> seqs = new LinkedIdentityHashSet<>(); |
| 1123 | ||
| 1124 | 2641 | for (int i = 0; i < getHeight(); i++) |
| 1125 | { | |
| 1126 | 2313 | SequenceI currentSeq = getSequenceAt(i); |
| 1127 | 2313 | resolveAndAddDatasetSeq(currentSeq, seqs, true); |
| 1128 | } | |
| 1129 | ||
| 1130 | // verify all mappings are in dataset | |
| 1131 | 328 | for (AlignedCodonFrame cf : codonFrameList) |
| 1132 | { | |
| 1133 | 3 | for (SequenceToSequenceMapping ssm : cf.getMappings()) |
| 1134 | { | |
| 1135 | 1 | if (!seqs.contains(ssm.getFromSeq())) |
| 1136 | { | |
| 1137 | 0 | resolveAndAddDatasetSeq(ssm.getFromSeq(), seqs, false); |
| 1138 | } | |
| 1139 | 1 | if (!seqs.contains(ssm.getMapping().getTo())) |
| 1140 | { | |
| 1141 | 0 | resolveAndAddDatasetSeq(ssm.getMapping().getTo(), seqs, false); |
| 1142 | } | |
| 1143 | } | |
| 1144 | } | |
| 1145 | // finally construct dataset | |
| 1146 | 328 | dataset = new Alignment(seqs.toArray(new SequenceI[seqs.size()])); |
| 1147 | // move mappings to the dataset alignment | |
| 1148 | 328 | dataset.codonFrameList = this.codonFrameList; |
| 1149 | 328 | this.codonFrameList = null; |
| 1150 | } | |
| 1151 | ||
| 1152 | /** | |
| 1153 | * reference count for number of alignments referencing this one. | |
| 1154 | */ | |
| 1155 | int alignmentRefs = 0; | |
| 1156 | ||
| 1157 | /** | |
| 1158 | * increase reference count to this alignment. | |
| 1159 | */ | |
| 1160 | 355 | private void addAlignmentRef() |
| 1161 | { | |
| 1162 | 355 | alignmentRefs++; |
| 1163 | } | |
| 1164 | ||
| 1165 | 2757 | @Override |
| 1166 | public Alignment getDataset() | |
| 1167 | { | |
| 1168 | 2757 | return dataset; |
| 1169 | } | |
| 1170 | ||
| 1171 | 368 | @Override |
| 1172 | public boolean padGaps() | |
| 1173 | { | |
| 1174 | 368 | boolean modified = false; |
| 1175 | ||
| 1176 | // Remove excess gaps from the end of alignment | |
| 1177 | 368 | int maxLength = -1; |
| 1178 | ||
| 1179 | 368 | SequenceI current; |
| 1180 | 4818 | for (int i = 0; i < sequences.size(); i++) |
| 1181 | { | |
| 1182 | 4450 | current = getSequenceAt(i); |
| 1183 | 11562 | for (int j = current.getLength(); j > maxLength; j--) |
| 1184 | { | |
| 1185 | 7756 | if (j > maxLength |
| 1186 | && !jalview.util.Comparison.isGap(current.getCharAt(j))) | |
| 1187 | { | |
| 1188 | 644 | maxLength = j; |
| 1189 | 644 | break; |
| 1190 | } | |
| 1191 | } | |
| 1192 | } | |
| 1193 | ||
| 1194 | 368 | maxLength++; |
| 1195 | ||
| 1196 | 368 | int cLength; |
| 1197 | 4818 | for (int i = 0; i < sequences.size(); i++) |
| 1198 | { | |
| 1199 | 4450 | current = getSequenceAt(i); |
| 1200 | 4450 | cLength = current.getLength(); |
| 1201 | ||
| 1202 | 4450 | if (cLength < maxLength) |
| 1203 | { | |
| 1204 | 115 | current.insertCharAt(cLength, maxLength - cLength, gapCharacter); |
| 1205 | 115 | modified = true; |
| 1206 | } | |
| 1207 | 4335 | else if (current.getLength() > maxLength) |
| 1208 | { | |
| 1209 | 94 | current.deleteChars(maxLength, current.getLength()); |
| 1210 | } | |
| 1211 | } | |
| 1212 | 368 | return modified; |
| 1213 | } | |
| 1214 | ||
| 1215 | /** | |
| 1216 | * Justify the sequences to the left or right by deleting and inserting gaps | |
| 1217 | * before the initial residue or after the terminal residue | |
| 1218 | * | |
| 1219 | * @param right | |
| 1220 | * true if alignment padded to right, false to justify to left | |
| 1221 | * @return true if alignment was changed | |
| 1222 | */ | |
| 1223 | 0 | @Override |
| 1224 | public boolean justify(boolean right) | |
| 1225 | { | |
| 1226 | 0 | boolean modified = false; |
| 1227 | ||
| 1228 | // Remove excess gaps from the end of alignment | |
| 1229 | 0 | int maxLength = -1; |
| 1230 | 0 | int ends[] = new int[sequences.size() * 2]; |
| 1231 | 0 | SequenceI current; |
| 1232 | 0 | for (int i = 0; i < sequences.size(); i++) |
| 1233 | { | |
| 1234 | 0 | current = getSequenceAt(i); |
| 1235 | // This should really be a sequence method | |
| 1236 | 0 | ends[i * 2] = current.findIndex(current.getStart()); |
| 1237 | 0 | ends[i * 2 + 1] = current |
| 1238 | .findIndex(current.getStart() + current.getLength()); | |
| 1239 | 0 | boolean hitres = false; |
| 1240 | 0 | for (int j = 0, rs = 0, ssiz = current.getLength(); j < ssiz; j++) |
| 1241 | { | |
| 1242 | 0 | if (!jalview.util.Comparison.isGap(current.getCharAt(j))) |
| 1243 | { | |
| 1244 | 0 | if (!hitres) |
| 1245 | { | |
| 1246 | 0 | ends[i * 2] = j; |
| 1247 | 0 | hitres = true; |
| 1248 | } | |
| 1249 | else | |
| 1250 | { | |
| 1251 | 0 | ends[i * 2 + 1] = j; |
| 1252 | 0 | if (j - ends[i * 2] > maxLength) |
| 1253 | { | |
| 1254 | 0 | maxLength = j - ends[i * 2]; |
| 1255 | } | |
| 1256 | } | |
| 1257 | } | |
| 1258 | } | |
| 1259 | } | |
| 1260 | ||
| 1261 | 0 | maxLength++; |
| 1262 | // now edit the flanking gaps to justify to either left or right | |
| 1263 | 0 | int cLength, extent, diff; |
| 1264 | 0 | for (int i = 0; i < sequences.size(); i++) |
| 1265 | { | |
| 1266 | 0 | current = getSequenceAt(i); |
| 1267 | ||
| 1268 | 0 | cLength = 1 + ends[i * 2 + 1] - ends[i * 2]; |
| 1269 | 0 | diff = maxLength - cLength; // number of gaps to indent |
| 1270 | 0 | extent = current.getLength(); |
| 1271 | 0 | if (right) |
| 1272 | { | |
| 1273 | // right justify | |
| 1274 | 0 | if (extent > ends[i * 2 + 1]) |
| 1275 | { | |
| 1276 | 0 | current.deleteChars(ends[i * 2 + 1] + 1, extent); |
| 1277 | 0 | modified = true; |
| 1278 | } | |
| 1279 | 0 | if (ends[i * 2] > diff) |
| 1280 | { | |
| 1281 | 0 | current.deleteChars(0, ends[i * 2] - diff); |
| 1282 | 0 | modified = true; |
| 1283 | } | |
| 1284 | else | |
| 1285 | { | |
| 1286 | 0 | if (ends[i * 2] < diff) |
| 1287 | { | |
| 1288 | 0 | current.insertCharAt(0, diff - ends[i * 2], gapCharacter); |
| 1289 | 0 | modified = true; |
| 1290 | } | |
| 1291 | } | |
| 1292 | } | |
| 1293 | else | |
| 1294 | { | |
| 1295 | // left justify | |
| 1296 | 0 | if (ends[i * 2] > 0) |
| 1297 | { | |
| 1298 | 0 | current.deleteChars(0, ends[i * 2]); |
| 1299 | 0 | modified = true; |
| 1300 | 0 | ends[i * 2 + 1] -= ends[i * 2]; |
| 1301 | 0 | extent -= ends[i * 2]; |
| 1302 | } | |
| 1303 | 0 | if (extent > maxLength) |
| 1304 | { | |
| 1305 | 0 | current.deleteChars(maxLength + 1, extent); |
| 1306 | 0 | modified = true; |
| 1307 | } | |
| 1308 | else | |
| 1309 | { | |
| 1310 | 0 | if (extent < maxLength) |
| 1311 | { | |
| 1312 | 0 | current.insertCharAt(extent, maxLength - extent, gapCharacter); |
| 1313 | 0 | modified = true; |
| 1314 | } | |
| 1315 | } | |
| 1316 | } | |
| 1317 | } | |
| 1318 | 0 | return modified; |
| 1319 | } | |
| 1320 | ||
| 1321 | 9833 | @Override |
| 1322 | public HiddenSequences getHiddenSequences() | |
| 1323 | { | |
| 1324 | 9833 | return hiddenSequences; |
| 1325 | } | |
| 1326 | ||
| 1327 | 17485 | @Override |
| 1328 | public HiddenColumns getHiddenColumns() | |
| 1329 | { | |
| 1330 | 17485 | return hiddenCols; |
| 1331 | } | |
| 1332 | ||
| 1333 | 0 | @Override |
| 1334 | public CigarArray getCompactAlignment() | |
| 1335 | { | |
| 1336 | 0 | synchronized (sequences) |
| 1337 | { | |
| 1338 | 0 | SeqCigar alseqs[] = new SeqCigar[sequences.size()]; |
| 1339 | 0 | int i = 0; |
| 1340 | 0 | for (SequenceI seq : sequences) |
| 1341 | { | |
| 1342 | 0 | alseqs[i++] = new SeqCigar(seq); |
| 1343 | } | |
| 1344 | 0 | CigarArray cal = new CigarArray(alseqs); |
| 1345 | 0 | cal.addOperation(CigarArray.M, getWidth()); |
| 1346 | 0 | return cal; |
| 1347 | } | |
| 1348 | } | |
| 1349 | ||
| 1350 | 131 | @Override |
| 1351 | public void setProperty(Object key, Object value) | |
| 1352 | { | |
| 1353 | 131 | if (alignmentProperties == null) |
| 1354 | { | |
| 1355 | 9 | alignmentProperties = new Hashtable(); |
| 1356 | } | |
| 1357 | ||
| 1358 | 131 | alignmentProperties.put(key, value); |
| 1359 | } | |
| 1360 | ||
| 1361 | 0 | @Override |
| 1362 | public Object getProperty(Object key) | |
| 1363 | { | |
| 1364 | 0 | if (alignmentProperties != null) |
| 1365 | { | |
| 1366 | 0 | return alignmentProperties.get(key); |
| 1367 | } | |
| 1368 | else | |
| 1369 | { | |
| 1370 | 0 | return null; |
| 1371 | } | |
| 1372 | } | |
| 1373 | ||
| 1374 | 350 | @Override |
| 1375 | public Hashtable getProperties() | |
| 1376 | { | |
| 1377 | 350 | return alignmentProperties; |
| 1378 | } | |
| 1379 | ||
| 1380 | /** | |
| 1381 | * Adds the given mapping to the stored set. Note this may be held on the | |
| 1382 | * dataset alignment. | |
| 1383 | */ | |
| 1384 | 80 | @Override |
| 1385 | public void addCodonFrame(AlignedCodonFrame codons) | |
| 1386 | { | |
| 1387 | 80 | List<AlignedCodonFrame> acfs = getCodonFrames(); |
| 1388 | 80 | if (codons != null && acfs != null && !acfs.contains(codons)) |
| 1389 | { | |
| 1390 | 77 | acfs.add(codons); |
| 1391 | } | |
| 1392 | } | |
| 1393 | ||
| 1394 | /* | |
| 1395 | * (non-Javadoc) | |
| 1396 | * | |
| 1397 | * @see | |
| 1398 | * jalview.datamodel.AlignmentI#getCodonFrame(jalview.datamodel.SequenceI) | |
| 1399 | */ | |
| 1400 | 49 | @Override |
| 1401 | public List<AlignedCodonFrame> getCodonFrame(SequenceI seq) | |
| 1402 | { | |
| 1403 | 49 | if (seq == null) |
| 1404 | { | |
| 1405 | 0 | return null; |
| 1406 | } | |
| 1407 | 49 | List<AlignedCodonFrame> cframes = new ArrayList<>(); |
| 1408 | 49 | for (AlignedCodonFrame acf : getCodonFrames()) |
| 1409 | { | |
| 1410 | 103 | if (acf.involvesSequence(seq)) |
| 1411 | { | |
| 1412 | 44 | cframes.add(acf); |
| 1413 | } | |
| 1414 | } | |
| 1415 | 49 | return cframes; |
| 1416 | } | |
| 1417 | ||
| 1418 | /** | |
| 1419 | * Sets the codon frame mappings (replacing any existing mappings). Note the | |
| 1420 | * mappings are set on the dataset alignment instead if there is one. | |
| 1421 | * | |
| 1422 | * @see jalview.datamodel.AlignmentI#setCodonFrames() | |
| 1423 | */ | |
| 1424 | 37 | @Override |
| 1425 | public void setCodonFrames(List<AlignedCodonFrame> acfs) | |
| 1426 | { | |
| 1427 | 37 | if (dataset != null) |
| 1428 | { | |
| 1429 | 17 | dataset.setCodonFrames(acfs); |
| 1430 | } | |
| 1431 | else | |
| 1432 | { | |
| 1433 | 20 | this.codonFrameList = acfs; |
| 1434 | } | |
| 1435 | } | |
| 1436 | ||
| 1437 | /** | |
| 1438 | * Returns the set of codon frame mappings. Any changes to the returned set | |
| 1439 | * will affect the alignment. The mappings are held on (and read from) the | |
| 1440 | * dataset alignment if there is one. | |
| 1441 | * | |
| 1442 | * @see jalview.datamodel.AlignmentI#getCodonFrames() | |
| 1443 | */ | |
| 1444 | 3404 | @Override |
| 1445 | public List<AlignedCodonFrame> getCodonFrames() | |
| 1446 | { | |
| 1447 | // TODO: Fix this method to fix failing AlignedCodonFrame tests | |
| 1448 | // this behaviour is currently incorrect. method should return codon frames | |
| 1449 | // for just the alignment, | |
| 1450 | // selected from dataset | |
| 1451 | 3404 | return dataset != null ? dataset.getCodonFrames() : codonFrameList; |
| 1452 | } | |
| 1453 | ||
| 1454 | /** | |
| 1455 | * Removes the given mapping from the stored set. Note that the mappings are | |
| 1456 | * held on the dataset alignment if there is one. | |
| 1457 | */ | |
| 1458 | 0 | @Override |
| 1459 | public boolean removeCodonFrame(AlignedCodonFrame codons) | |
| 1460 | { | |
| 1461 | 0 | List<AlignedCodonFrame> acfs = getCodonFrames(); |
| 1462 | 0 | if (codons == null || acfs == null) |
| 1463 | { | |
| 1464 | 0 | return false; |
| 1465 | } | |
| 1466 | 0 | return acfs.remove(codons); |
| 1467 | } | |
| 1468 | ||
| 1469 | 1 | @Override |
| 1470 | public void append(AlignmentI toappend) | |
| 1471 | { | |
| 1472 | // TODO JAL-1270 needs test coverage | |
| 1473 | // currently tested for use in jalview.gui.SequenceFetcher | |
| 1474 | 1 | char oldc = toappend.getGapCharacter(); |
| 1475 | 1 | boolean samegap = oldc == getGapCharacter(); |
| 1476 | 1 | boolean hashidden = toappend.getHiddenSequences() != null |
| 1477 | && toappend.getHiddenSequences().hiddenSequences != null; | |
| 1478 | // get all sequences including any hidden ones | |
| 1479 | 1 | List<SequenceI> sqs = (hashidden) |
| 1480 | ? toappend.getHiddenSequences().getFullAlignment() | |
| 1481 | .getSequences() | |
| 1482 | : toappend.getSequences(); | |
| 1483 | 1 | if (sqs != null) |
| 1484 | { | |
| 1485 | // avoid self append deadlock by | |
| 1486 | 1 | List<SequenceI> toappendsq = new ArrayList<>(); |
| 1487 | 1 | synchronized (sqs) |
| 1488 | { | |
| 1489 | 1 | for (SequenceI addedsq : sqs) |
| 1490 | { | |
| 1491 | 1 | if (!samegap) |
| 1492 | { | |
| 1493 | 1 | addedsq.replace(oldc, gapCharacter); |
| 1494 | } | |
| 1495 | 1 | toappendsq.add(addedsq); |
| 1496 | } | |
| 1497 | } | |
| 1498 | 1 | for (SequenceI addedsq : toappendsq) |
| 1499 | { | |
| 1500 | 1 | addSequence(addedsq); |
| 1501 | } | |
| 1502 | } | |
| 1503 | 1 | AlignmentAnnotation[] alan = toappend.getAlignmentAnnotation(); |
| 1504 | 1 | for (int a = 0; alan != null && a < alan.length; a++) |
| 1505 | { | |
| 1506 | 0 | addAnnotation(alan[a]); |
| 1507 | } | |
| 1508 | ||
| 1509 | // use add method | |
| 1510 | 1 | getCodonFrames().addAll(toappend.getCodonFrames()); |
| 1511 | ||
| 1512 | 1 | List<SequenceGroup> sg = toappend.getGroups(); |
| 1513 | 1 | if (sg != null) |
| 1514 | { | |
| 1515 | 1 | for (SequenceGroup _sg : sg) |
| 1516 | { | |
| 1517 | 0 | addGroup(_sg); |
| 1518 | } | |
| 1519 | } | |
| 1520 | 1 | if (toappend.getHiddenSequences() != null) |
| 1521 | { | |
| 1522 | 1 | HiddenSequences hs = toappend.getHiddenSequences(); |
| 1523 | 1 | if (hiddenSequences == null) |
| 1524 | { | |
| 1525 | 0 | hiddenSequences = new HiddenSequences(this); |
| 1526 | } | |
| 1527 | 1 | if (hs.hiddenSequences != null) |
| 1528 | { | |
| 1529 | 0 | for (int s = 0; s < hs.hiddenSequences.length; s++) |
| 1530 | { | |
| 1531 | // hide the newly appended sequence in the alignment | |
| 1532 | 0 | if (hs.hiddenSequences[s] != null) |
| 1533 | { | |
| 1534 | 0 | hiddenSequences.hideSequence(hs.hiddenSequences[s]); |
| 1535 | } | |
| 1536 | } | |
| 1537 | } | |
| 1538 | } | |
| 1539 | 1 | if (toappend.getProperties() != null) |
| 1540 | { | |
| 1541 | // we really can't do very much here - just try to concatenate strings | |
| 1542 | // where property collisions occur. | |
| 1543 | 0 | Enumeration key = toappend.getProperties().keys(); |
| 1544 | 0 | while (key.hasMoreElements()) |
| 1545 | { | |
| 1546 | 0 | Object k = key.nextElement(); |
| 1547 | 0 | Object ourval = this.getProperty(k); |
| 1548 | 0 | Object toapprop = toappend.getProperty(k); |
| 1549 | 0 | if (ourval != null) |
| 1550 | { | |
| 1551 | 0 | if (ourval.getClass().equals(toapprop.getClass()) |
| 1552 | && !ourval.equals(toapprop)) | |
| 1553 | { | |
| 1554 | 0 | if (ourval instanceof String) |
| 1555 | { | |
| 1556 | // append strings | |
| 1557 | 0 | this.setProperty(k, |
| 1558 | ((String) ourval) + "; " + ((String) toapprop)); | |
| 1559 | } | |
| 1560 | else | |
| 1561 | { | |
| 1562 | 0 | if (ourval instanceof Vector) |
| 1563 | { | |
| 1564 | // append vectors | |
| 1565 | 0 | Enumeration theirv = ((Vector) toapprop).elements(); |
| 1566 | 0 | while (theirv.hasMoreElements()) |
| 1567 | { | |
| 1568 | 0 | ((Vector) ourval).addElement(theirv); |
| 1569 | } | |
| 1570 | } | |
| 1571 | } | |
| 1572 | } | |
| 1573 | } | |
| 1574 | else | |
| 1575 | { | |
| 1576 | // just add new property directly | |
| 1577 | 0 | setProperty(k, toapprop); |
| 1578 | } | |
| 1579 | ||
| 1580 | } | |
| 1581 | } | |
| 1582 | } | |
| 1583 | ||
| 1584 | 19 | @Override |
| 1585 | public AlignmentAnnotation findOrCreateAnnotation(String name, | |
| 1586 | String calcId, boolean autoCalc, SequenceI seqRef, | |
| 1587 | SequenceGroup groupRef) | |
| 1588 | { | |
| 1589 | 19 | if (annotations != null) |
| 1590 | { | |
| 1591 | 18 | for (AlignmentAnnotation annot : getAlignmentAnnotation()) |
| 1592 | { | |
| 1593 | 192 | if (annot.autoCalculated == autoCalc && (name.equals(annot.label)) |
| 1594 | && (calcId == null || annot.getCalcId().equals(calcId)) | |
| 1595 | && annot.sequenceRef == seqRef | |
| 1596 | && annot.groupRef == groupRef) | |
| 1597 | { | |
| 1598 | 0 | return annot; |
| 1599 | } | |
| 1600 | } | |
| 1601 | } | |
| 1602 | 19 | AlignmentAnnotation annot = new AlignmentAnnotation(name, name, |
| 1603 | new Annotation[1], 0f, 0f, AlignmentAnnotation.BAR_GRAPH); | |
| 1604 | 19 | annot.hasText = false; |
| 1605 | 19 | if (calcId != null) |
| 1606 | { | |
| 1607 | 18 | annot.setCalcId(new String(calcId)); |
| 1608 | } | |
| 1609 | 19 | annot.autoCalculated = autoCalc; |
| 1610 | 19 | if (seqRef != null) |
| 1611 | { | |
| 1612 | 16 | annot.setSequenceRef(seqRef); |
| 1613 | } | |
| 1614 | 19 | annot.groupRef = groupRef; |
| 1615 | 19 | addAnnotation(annot); |
| 1616 | ||
| 1617 | 18 | return annot; |
| 1618 | } | |
| 1619 | ||
| 1620 | 230 | @Override |
| 1621 | public Iterable<AlignmentAnnotation> findAnnotation(String calcId) | |
| 1622 | { | |
| 1623 | 230 | AlignmentAnnotation[] alignmentAnnotation = getAlignmentAnnotation(); |
| 1624 | 230 | if (alignmentAnnotation != null) |
| 1625 | { | |
| 1626 | 223 | return AlignmentAnnotation.findAnnotation( |
| 1627 | Arrays.asList(getAlignmentAnnotation()), calcId); | |
| 1628 | } | |
| 1629 | 7 | return Arrays.asList(new AlignmentAnnotation[] {}); |
| 1630 | } | |
| 1631 | ||
| 1632 | 17 | @Override |
| 1633 | public Iterable<AlignmentAnnotation> findAnnotations(SequenceI seq, | |
| 1634 | String calcId, String label) | |
| 1635 | { | |
| 1636 | 17 | return AlignmentAnnotation.findAnnotations( |
| 1637 | Arrays.asList(getAlignmentAnnotation()), seq, calcId, label); | |
| 1638 | } | |
| 1639 | ||
| 1640 | 0 | @Override |
| 1641 | public void moveSelectedSequencesByOne(SequenceGroup sg, | |
| 1642 | Map<SequenceI, SequenceCollectionI> map, boolean up) | |
| 1643 | { | |
| 1644 | 0 | synchronized (sequences) |
| 1645 | { | |
| 1646 | 0 | if (up) |
| 1647 | { | |
| 1648 | ||
| 1649 | 0 | for (int i = 1, iSize = sequences.size(); i < iSize; i++) |
| 1650 | { | |
| 1651 | 0 | SequenceI seq = sequences.get(i); |
| 1652 | 0 | if (!sg.getSequences(map).contains(seq)) |
| 1653 | { | |
| 1654 | 0 | continue; |
| 1655 | } | |
| 1656 | ||
| 1657 | 0 | SequenceI temp = sequences.get(i - 1); |
| 1658 | 0 | if (sg.getSequences(null).contains(temp)) |
| 1659 | { | |
| 1660 | 0 | continue; |
| 1661 | } | |
| 1662 | ||
| 1663 | 0 | sequences.set(i, temp); |
| 1664 | 0 | sequences.set(i - 1, seq); |
| 1665 | } | |
| 1666 | } | |
| 1667 | else | |
| 1668 | { | |
| 1669 | 0 | for (int i = sequences.size() - 2; i > -1; i--) |
| 1670 | { | |
| 1671 | 0 | SequenceI seq = sequences.get(i); |
| 1672 | 0 | if (!sg.getSequences(map).contains(seq)) |
| 1673 | { | |
| 1674 | 0 | continue; |
| 1675 | } | |
| 1676 | ||
| 1677 | 0 | SequenceI temp = sequences.get(i + 1); |
| 1678 | 0 | if (sg.getSequences(map).contains(temp)) |
| 1679 | { | |
| 1680 | 0 | continue; |
| 1681 | } | |
| 1682 | ||
| 1683 | 0 | sequences.set(i, temp); |
| 1684 | 0 | sequences.set(i + 1, seq); |
| 1685 | } | |
| 1686 | } | |
| 1687 | ||
| 1688 | } | |
| 1689 | } | |
| 1690 | ||
| 1691 | 0 | @Override |
| 1692 | public void validateAnnotation(AlignmentAnnotation alignmentAnnotation) | |
| 1693 | { | |
| 1694 | 0 | alignmentAnnotation.validateRangeAndDisplay(); |
| 1695 | 0 | if (isNucleotide() && alignmentAnnotation.isValidStruc()) |
| 1696 | { | |
| 1697 | 0 | hasRNAStructure = true; |
| 1698 | } | |
| 1699 | } | |
| 1700 | ||
| 1701 | private SequenceI seqrep = null; | |
| 1702 | ||
| 1703 | /** | |
| 1704 | * | |
| 1705 | * @return the representative sequence for this group | |
| 1706 | */ | |
| 1707 | 13447 | @Override |
| 1708 | public SequenceI getSeqrep() | |
| 1709 | { | |
| 1710 | 13447 | return seqrep; |
| 1711 | } | |
| 1712 | ||
| 1713 | /** | |
| 1714 | * set the representative sequence for this group. Note - this affects the | |
| 1715 | * interpretation of the Hidereps attribute. | |
| 1716 | * | |
| 1717 | * @param seqrep | |
| 1718 | * the seqrep to set (null means no sequence representative) | |
| 1719 | */ | |
| 1720 | 280 | @Override |
| 1721 | public void setSeqrep(SequenceI seqrep) | |
| 1722 | { | |
| 1723 | 280 | this.seqrep = seqrep; |
| 1724 | } | |
| 1725 | ||
| 1726 | /** | |
| 1727 | * | |
| 1728 | * @return true if group has a sequence representative | |
| 1729 | */ | |
| 1730 | 7380 | @Override |
| 1731 | public boolean hasSeqrep() | |
| 1732 | { | |
| 1733 | 7380 | return seqrep != null; |
| 1734 | } | |
| 1735 | ||
| 1736 | 19 | @Override |
| 1737 | public int getEndRes() | |
| 1738 | { | |
| 1739 | 19 | return getWidth() - 1; |
| 1740 | } | |
| 1741 | ||
| 1742 | 19 | @Override |
| 1743 | public int getStartRes() | |
| 1744 | { | |
| 1745 | 19 | return 0; |
| 1746 | } | |
| 1747 | ||
| 1748 | /* | |
| 1749 | * In the case of AlignmentI - returns the dataset for the alignment, if set | |
| 1750 | * (non-Javadoc) | |
| 1751 | * | |
| 1752 | * @see jalview.datamodel.AnnotatedCollectionI#getContext() | |
| 1753 | */ | |
| 1754 | 468 | @Override |
| 1755 | public AnnotatedCollectionI getContext() | |
| 1756 | { | |
| 1757 | 468 | return dataset; |
| 1758 | } | |
| 1759 | ||
| 1760 | /** | |
| 1761 | * Align this alignment like the given (mapped) one. | |
| 1762 | */ | |
| 1763 | 0 | @Override |
| 1764 | public int alignAs(AlignmentI al) | |
| 1765 | { | |
| 1766 | /* | |
| 1767 | * Currently retains unmapped gaps (in introns), regaps mapped regions | |
| 1768 | * (exons) | |
| 1769 | */ | |
| 1770 | 0 | return alignAs(al, false, true); |
| 1771 | } | |
| 1772 | ||
| 1773 | /** | |
| 1774 | * Align this alignment 'the same as' the given one. Mapped sequences only are | |
| 1775 | * realigned. If both of the same type (nucleotide/protein) then align both | |
| 1776 | * identically. If this is nucleotide and the other is protein, make 3 gaps | |
| 1777 | * for each gap in the protein sequences. If this is protein and the other is | |
| 1778 | * nucleotide, insert a gap for each 3 gaps (or part thereof) between | |
| 1779 | * nucleotide bases. If this is protein and the other is nucleotide, gaps | |
| 1780 | * protein to match the relative ordering of codons in the nucleotide. | |
| 1781 | * | |
| 1782 | * Parameters control whether gaps in exon (mapped) and intron (unmapped) | |
| 1783 | * regions are preserved. Gaps that connect introns to exons are treated | |
| 1784 | * conservatively, i.e. only preserved if both intron and exon gaps are | |
| 1785 | * preserved. TODO: check caveats below where the implementation fails | |
| 1786 | * | |
| 1787 | * @param al | |
| 1788 | * - must have same dataset, and sequences in al must have equivalent | |
| 1789 | * dataset sequence and start/end bounds under given mapping | |
| 1790 | * @param preserveMappedGaps | |
| 1791 | * if true, gaps within and between mapped codons are preserved | |
| 1792 | * @param preserveUnmappedGaps | |
| 1793 | * if true, gaps within and between unmapped codons are preserved | |
| 1794 | */ | |
| 1795 | // @Override | |
| 1796 | 6 | public int alignAs(AlignmentI al, boolean preserveMappedGaps, |
| 1797 | boolean preserveUnmappedGaps) | |
| 1798 | { | |
| 1799 | // TODO should this method signature be the one in the interface? | |
| 1800 | // JBPComment - yes - neither flag is used, so should be deleted. | |
| 1801 | 6 | boolean thisIsNucleotide = this.isNucleotide(); |
| 1802 | 6 | boolean thatIsProtein = !al.isNucleotide(); |
| 1803 | 6 | if (!thatIsProtein && !thisIsNucleotide) |
| 1804 | { | |
| 1805 | 1 | return AlignmentUtils.alignProteinAsDna(this, al); |
| 1806 | } | |
| 1807 | 5 | else if (thatIsProtein && thisIsNucleotide) |
| 1808 | { | |
| 1809 | 4 | return AlignmentUtils.alignCdsAsProtein(this, al); |
| 1810 | } | |
| 1811 | 1 | return AlignmentUtils.alignAs(this, al); |
| 1812 | } | |
| 1813 | ||
| 1814 | /** | |
| 1815 | * Returns the alignment in Fasta format. Behaviour of this method is not | |
| 1816 | * guaranteed between versions. | |
| 1817 | */ | |
| 1818 | 2 | @Override |
| 1819 | public String toString() | |
| 1820 | { | |
| 1821 | 2 | return new FastaFile().print(getSequencesArray(), true); |
| 1822 | } | |
| 1823 | ||
| 1824 | /** | |
| 1825 | * Returns the set of distinct sequence names. No ordering is guaranteed. | |
| 1826 | */ | |
| 1827 | 0 | @Override |
| 1828 | public Set<String> getSequenceNames() | |
| 1829 | { | |
| 1830 | 0 | Set<String> names = new HashSet<>(); |
| 1831 | 0 | for (SequenceI seq : getSequences()) |
| 1832 | { | |
| 1833 | 0 | names.add(seq.getName()); |
| 1834 | } | |
| 1835 | 0 | return names; |
| 1836 | } | |
| 1837 | ||
| 1838 | 71 | @Override |
| 1839 | public boolean hasValidSequence() | |
| 1840 | { | |
| 1841 | 71 | boolean hasValidSeq = false; |
| 1842 | 71 | for (SequenceI seq : getSequences()) |
| 1843 | { | |
| 1844 | 71 | if ((seq.getEnd() - seq.getStart()) > 0) |
| 1845 | { | |
| 1846 | 71 | hasValidSeq = true; |
| 1847 | 71 | break; |
| 1848 | } | |
| 1849 | } | |
| 1850 | 71 | return hasValidSeq; |
| 1851 | } | |
| 1852 | ||
| 1853 | /** | |
| 1854 | * Update any mappings to 'virtual' sequences to compatible real ones, if | |
| 1855 | * present in the added sequences. Returns a count of mappings updated. | |
| 1856 | * | |
| 1857 | * @param seqs | |
| 1858 | * @return | |
| 1859 | */ | |
| 1860 | 1 | @Override |
| 1861 | public int realiseMappings(List<SequenceI> seqs) | |
| 1862 | { | |
| 1863 | 1 | int count = 0; |
| 1864 | 1 | for (SequenceI seq : seqs) |
| 1865 | { | |
| 1866 | 4 | for (AlignedCodonFrame mapping : getCodonFrames()) |
| 1867 | { | |
| 1868 | 0 | count += mapping.realiseWith(seq); |
| 1869 | } | |
| 1870 | } | |
| 1871 | 1 | return count; |
| 1872 | } | |
| 1873 | ||
| 1874 | /** | |
| 1875 | * Returns the first AlignedCodonFrame that has a mapping between the given | |
| 1876 | * dataset sequences | |
| 1877 | * | |
| 1878 | * @param mapFrom | |
| 1879 | * @param mapTo | |
| 1880 | * @return | |
| 1881 | */ | |
| 1882 | 16 | @Override |
| 1883 | public AlignedCodonFrame getMapping(SequenceI mapFrom, SequenceI mapTo) | |
| 1884 | { | |
| 1885 | 16 | for (AlignedCodonFrame acf : getCodonFrames()) |
| 1886 | { | |
| 1887 | 2 | if (acf.getAaForDnaSeq(mapFrom) == mapTo) |
| 1888 | { | |
| 1889 | 1 | return acf; |
| 1890 | } | |
| 1891 | } | |
| 1892 | 15 | return null; |
| 1893 | } | |
| 1894 | ||
| 1895 | 30 | @Override |
| 1896 | public void setHiddenColumns(HiddenColumns cols) | |
| 1897 | { | |
| 1898 | 30 | hiddenCols = cols; |
| 1899 | } | |
| 1900 | ||
| 1901 | 0 | @Override |
| 1902 | public void setupJPredAlignment() | |
| 1903 | { | |
| 1904 | 0 | SequenceI repseq = getSequenceAt(0); |
| 1905 | 0 | setSeqrep(repseq); |
| 1906 | 0 | HiddenColumns cs = new HiddenColumns(); |
| 1907 | 0 | cs.hideList(repseq.getInsertions()); |
| 1908 | 0 | setHiddenColumns(cs); |
| 1909 | } | |
| 1910 | ||
| 1911 | 2 | @Override |
| 1912 | public HiddenColumns propagateInsertions(SequenceI profileseq, | |
| 1913 | AlignmentView input) | |
| 1914 | { | |
| 1915 | 2 | int profsqpos = 0; |
| 1916 | ||
| 1917 | 2 | char gc = getGapCharacter(); |
| 1918 | 2 | Object[] alandhidden = input.getAlignmentAndHiddenColumns(gc); |
| 1919 | 2 | HiddenColumns nview = (HiddenColumns) alandhidden[1]; |
| 1920 | 2 | SequenceI origseq = ((SequenceI[]) alandhidden[0])[profsqpos]; |
| 1921 | 2 | return propagateInsertions(profileseq, origseq, nview); |
| 1922 | } | |
| 1923 | ||
| 1924 | /** | |
| 1925 | * | |
| 1926 | * @param profileseq | |
| 1927 | * sequence in al which corresponds to origseq | |
| 1928 | * @param al | |
| 1929 | * alignment which is to have gaps inserted into it | |
| 1930 | * @param origseq | |
| 1931 | * sequence corresponding to profileseq which defines gap map for | |
| 1932 | * modifying al | |
| 1933 | */ | |
| 1934 | 2 | private HiddenColumns propagateInsertions(SequenceI profileseq, |
| 1935 | SequenceI origseq, HiddenColumns hc) | |
| 1936 | { | |
| 1937 | // take the set of hidden columns, and the set of gaps in origseq, | |
| 1938 | // and remove all the hidden gaps from hiddenColumns | |
| 1939 | ||
| 1940 | // first get the gaps as a Bitset | |
| 1941 | // then calculate hidden ^ not(gap) | |
| 1942 | 2 | BitSet gaps = origseq.gapBitset(); |
| 1943 | 2 | hc.andNot(gaps); |
| 1944 | ||
| 1945 | // for each sequence in the alignment, except the profile sequence, | |
| 1946 | // insert gaps corresponding to each hidden region but where each hidden | |
| 1947 | // column region is shifted backwards by the number of preceding visible | |
| 1948 | // gaps update hidden columns at the same time | |
| 1949 | 2 | HiddenColumns newhidden = new HiddenColumns(); |
| 1950 | ||
| 1951 | 2 | int numGapsBefore = 0; |
| 1952 | 2 | int gapPosition = 0; |
| 1953 | 2 | Iterator<int[]> it = hc.iterator(); |
| 1954 | 4 | while (it.hasNext()) |
| 1955 | { | |
| 1956 | 2 | int[] region = it.next(); |
| 1957 | ||
| 1958 | // get region coordinates accounting for gaps | |
| 1959 | // we can rely on gaps not being *in* hidden regions because we already | |
| 1960 | // removed those | |
| 1961 | 26 | while (gapPosition < region[0]) |
| 1962 | { | |
| 1963 | 24 | gapPosition++; |
| 1964 | 24 | if (gaps.get(gapPosition)) |
| 1965 | { | |
| 1966 | 8 | numGapsBefore++; |
| 1967 | } | |
| 1968 | } | |
| 1969 | ||
| 1970 | 2 | int left = region[0] - numGapsBefore; |
| 1971 | 2 | int right = region[1] - numGapsBefore; |
| 1972 | ||
| 1973 | 2 | newhidden.hideColumns(left, right); |
| 1974 | 2 | padGaps(left, right, profileseq); |
| 1975 | } | |
| 1976 | 2 | return newhidden; |
| 1977 | } | |
| 1978 | ||
| 1979 | /** | |
| 1980 | * Pad gaps in all sequences in alignment except profileseq | |
| 1981 | * | |
| 1982 | * @param left | |
| 1983 | * position of first gap to insert | |
| 1984 | * @param right | |
| 1985 | * position of last gap to insert | |
| 1986 | * @param profileseq | |
| 1987 | * sequence not to pad | |
| 1988 | */ | |
| 1989 | 2 | private void padGaps(int left, int right, SequenceI profileseq) |
| 1990 | { | |
| 1991 | 2 | char gc = getGapCharacter(); |
| 1992 | ||
| 1993 | // make a string with number of gaps = length of hidden region | |
| 1994 | 2 | StringBuilder sb = new StringBuilder(); |
| 1995 | 7 | for (int g = 0; g < right - left + 1; g++) |
| 1996 | { | |
| 1997 | 5 | sb.append(gc); |
| 1998 | } | |
| 1999 | ||
| 2000 | // loop over the sequences and pad with gaps where required | |
| 2001 | 22 | for (int s = 0, ns = getHeight(); s < ns; s++) |
| 2002 | { | |
| 2003 | 20 | SequenceI sqobj = getSequenceAt(s); |
| 2004 | 20 | if ((sqobj != profileseq) && (sqobj.getLength() >= left)) |
| 2005 | { | |
| 2006 | 15 | String sq = sqobj.getSequenceAsString(); |
| 2007 | 15 | sqobj.setSequence( |
| 2008 | sq.substring(0, left) + sb.toString() + sq.substring(left)); | |
| 2009 | } | |
| 2010 | } | |
| 2011 | } | |
| 2012 | ||
| 2013 | } |