X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=src%2Fjalview%2Fdatamodel%2FAlignment.java;h=290707dc7c3091017a96bd906b12d0a6ba395c9a;hb=90add82a33c05e218a92eb1cdb2c345da05465f8;hp=f5665dbbb750e9e7a256b1e2ed0f61061944dbce;hpb=6395e8fc1fc6b33c3b305d74d9be3642dfca8417;p=jalview.git diff --git a/src/jalview/datamodel/Alignment.java b/src/jalview/datamodel/Alignment.java index f5665db..290707d 100755 --- a/src/jalview/datamodel/Alignment.java +++ b/src/jalview/datamodel/Alignment.java @@ -21,8 +21,10 @@ package jalview.datamodel; import jalview.analysis.AlignmentUtils; +import jalview.datamodel.AlignedCodonFrame.SequenceToSequenceMapping; import jalview.io.FastaFile; import jalview.util.Comparison; +import jalview.util.LinkedIdentityHashSet; import jalview.util.MessageManager; import java.util.ArrayList; @@ -52,11 +54,7 @@ public class Alignment implements AlignmentI protected char gapCharacter = '-'; - protected int type = NUCLEOTIDE; - - public static final int PROTEIN = 0; - - public static final int NUCLEOTIDE = 1; + private boolean nucleotide = true; public boolean hasRNAStructure = false; @@ -74,14 +72,7 @@ public class Alignment implements AlignmentI hiddenSequences = new HiddenSequences(this); codonFrameList = new ArrayList(); - if (Comparison.isNucleotide(seqs)) - { - type = NUCLEOTIDE; - } - else - { - type = PROTEIN; - } + nucleotide = Comparison.isNucleotide(seqs); sequences = Collections.synchronizedList(new ArrayList()); @@ -216,7 +207,9 @@ public class Alignment implements AlignmentI } /** - * Adds a sequence to the alignment. Recalculates maxLength and size. + * Adds a sequence to the alignment. Recalculates maxLength and size. Note + * this currently does not recalculate whether or not the alignment is + * nucleotide, so mixed alignments may have undefined behaviour. * * @param snew */ @@ -225,18 +218,21 @@ public class Alignment implements AlignmentI { if (dataset != null) { + // maintain dataset integrity - if (snew.getDatasetSequence() != null) - { - getDataset().addSequence(snew.getDatasetSequence()); - } - else + SequenceI dsseq = snew.getDatasetSequence(); + if (dsseq == null) { // derive new sequence SequenceI adding = snew.deriveSequence(); - getDataset().addSequence(adding.getDatasetSequence()); snew = adding; + dsseq = snew.getDatasetSequence(); + } + if (getDataset().findIndex(dsseq) == -1) + { + getDataset().addSequence(dsseq); } + } if (sequences == null) { @@ -255,18 +251,22 @@ public class Alignment implements AlignmentI } } - /** - * Adds a sequence to the alignment. Recalculates maxLength and size. - * - * @param snew - */ @Override - public void setSequenceAt(int i, SequenceI snew) + public SequenceI replaceSequenceAt(int i, SequenceI snew) { synchronized (sequences) { - deleteSequence(i); - sequences.set(i, snew); + if (sequences.size() > i) + { + return sequences.set(i, snew); + + } + else + { + sequences.add(snew); + hiddenSequences.adjustHeightSequenceAdded(); + } + return null; } } @@ -282,13 +282,23 @@ public class Alignment implements AlignmentI } @Override - public void finalize() + public void finalize() throws Throwable { if (getDataset() != null) { getDataset().removeAlignmentRef(); } + nullReferences(); + super.finalize(); + } + + /** + * Defensively nulls out references in case this object is not garbage + * collected + */ + void nullReferences() + { dataset = null; sequences = null; groups = null; @@ -297,14 +307,16 @@ public class Alignment implements AlignmentI } /** - * decrement the alignmentRefs counter by one and call finalize if it goes to - * zero. + * decrement the alignmentRefs counter by one and null references if it goes + * to zero. + * + * @throws Throwable */ - private void removeAlignmentRef() + private void removeAlignmentRef() throws Throwable { if (--alignmentRefs == 0) { - finalize(); + nullReferences(); } } @@ -640,7 +652,7 @@ public class Alignment implements AlignmentI * jalview.datamodel.AlignmentI#findIndex(jalview.datamodel.SearchResults) */ @Override - public int findIndex(SearchResults results) + public int findIndex(SearchResultsI results) { int i = 0; @@ -956,29 +968,9 @@ public class Alignment implements AlignmentI } @Override - public void setNucleotide(boolean b) - { - if (b) - { - type = NUCLEOTIDE; - } - else - { - type = PROTEIN; - } - } - - @Override public boolean isNucleotide() { - if (type == NUCLEOTIDE) - { - return true; - } - else - { - return false; - } + return nucleotide; } @Override @@ -1029,6 +1021,70 @@ public class Alignment implements AlignmentI } /** + * add dataset sequences to seq for currentSeq and any sequences it references + */ + private void resolveAndAddDatasetSeq(SequenceI currentSeq, + Set seqs, boolean createDatasetSequence) + { + SequenceI alignedSeq = currentSeq; + if (currentSeq.getDatasetSequence() != null) + { + currentSeq = currentSeq.getDatasetSequence(); + } + else + { + if (createDatasetSequence) + { + currentSeq = currentSeq.createDatasetSequence(); + } + } + if (seqs.contains(currentSeq)) + { + return; + } + List toProcess = new ArrayList(); + toProcess.add(currentSeq); + while (toProcess.size() > 0) + { + // use a queue ? + SequenceI curDs = toProcess.remove(0); + if (seqs.contains(curDs)) + { + continue; + } + seqs.add(curDs); + // iterate over database references, making sure we add forward referenced + // sequences + if (curDs.getDBRefs() != null) + { + for (DBRefEntry dbr : curDs.getDBRefs()) + { + if (dbr.getMap() != null && dbr.getMap().getTo() != null) + { + if (dbr.getMap().getTo() == alignedSeq) + { + /* + * update mapping to be to the newly created dataset sequence + */ + dbr.getMap().setTo(currentSeq); + } + if (dbr.getMap().getTo().getDatasetSequence() != null) + { + throw new Error( + "Implementation error: Map.getTo() for dbref " + dbr + + " from " + curDs.getName() + + " is not a dataset sequence."); + } + // we recurse to add all forward references to dataset sequences via + // DBRefs/etc + toProcess.add(dbr.getMap().getTo()); + } + } + } + } + } + + /** * Creates a new dataset for this alignment. Can only be done once - if * dataset is not null this will not be performed. */ @@ -1038,22 +1094,32 @@ public class Alignment implements AlignmentI { return; } - SequenceI[] seqs = new SequenceI[getHeight()]; - SequenceI currentSeq; + // try to avoid using SequenceI.equals at this stage, it will be expensive + Set seqs = new LinkedIdentityHashSet(); + for (int i = 0; i < getHeight(); i++) { - currentSeq = getSequenceAt(i); - if (currentSeq.getDatasetSequence() != null) - { - seqs[i] = currentSeq.getDatasetSequence(); - } - else + SequenceI currentSeq = getSequenceAt(i); + resolveAndAddDatasetSeq(currentSeq, seqs, true); + } + + // verify all mappings are in dataset + for (AlignedCodonFrame cf : codonFrameList) + { + for (SequenceToSequenceMapping ssm : cf.getMappings()) { - seqs[i] = currentSeq.createDatasetSequence(); + if (!seqs.contains(ssm.getFromSeq())) + { + resolveAndAddDatasetSeq(ssm.getFromSeq(), seqs, false); + } + if (!seqs.contains(ssm.getMapping().getTo())) + { + resolveAndAddDatasetSeq(ssm.getMapping().getTo(), seqs, false); + } } } - - dataset = new Alignment(seqs); + // finally construct dataset + dataset = new Alignment(seqs.toArray(new SequenceI[seqs.size()])); // move mappings to the dataset alignment dataset.codonFrameList = this.codonFrameList; this.codonFrameList = null; @@ -1373,11 +1439,7 @@ public class Alignment implements AlignmentI @Override public void append(AlignmentI toappend) { - if (toappend == this) - { - System.err.println("Self append may cause a deadlock."); - } - // TODO test this method for a future 2.5 release + // TODO JAL-1270 needs test coverage // currently tested for use in jalview.gui.SequenceFetcher boolean samegap = toappend.getGapCharacter() == getGapCharacter(); char oldc = toappend.getGapCharacter(); @@ -1388,6 +1450,8 @@ public class Alignment implements AlignmentI .getFullAlignment().getSequences() : toappend.getSequences(); if (sqs != null) { + // avoid self append deadlock by + List toappendsq = new ArrayList(); synchronized (sqs) { for (SequenceI addedsq : sqs) @@ -1403,9 +1467,13 @@ public class Alignment implements AlignmentI } } } - addSequence(addedsq); + toappendsq.add(addedsq); } } + for (SequenceI addedsq : toappendsq) + { + addSequence(addedsq); + } } AlignmentAnnotation[] alan = toappend.getAlignmentAnnotation(); for (int a = 0; alan != null && a < alan.length; a++) @@ -1704,9 +1772,11 @@ public class Alignment implements AlignmentI * Parameters control whether gaps in exon (mapped) and intron (unmapped) * regions are preserved. Gaps that connect introns to exons are treated * conservatively, i.e. only preserved if both intron and exon gaps are - * preserved. + * preserved. TODO: check caveats below where the implementation fails * * @param al + * - must have same dataset, and sequences in al must have equivalent + * dataset sequence and start/end bounds under given mapping * @param preserveMappedGaps * if true, gaps within and between mapped codons are preserved * @param preserveUnmappedGaps @@ -1717,6 +1787,7 @@ public class Alignment implements AlignmentI boolean preserveUnmappedGaps) { // TODO should this method signature be the one in the interface? + // JBPComment - yes - neither flag is used, so should be deleted. boolean thisIsNucleotide = this.isNucleotide(); boolean thatIsProtein = !al.isNucleotide(); if (!thatIsProtein && !thisIsNucleotide) @@ -1737,7 +1808,7 @@ public class Alignment implements AlignmentI @Override public String toString() { - return new FastaFile().print(getSequencesArray()); + return new FastaFile().print(getSequencesArray(), true); } /**