import jalview.analysis.AlignSeq;
import jalview.api.DBRefEntryI;
+import jalview.datamodel.features.SequenceFeatures;
+import jalview.datamodel.features.SequenceFeaturesI;
import jalview.util.Comparison;
import jalview.util.DBRefUtils;
import jalview.util.MapList;
import java.util.Collections;
import java.util.Enumeration;
import java.util.List;
+import java.util.ListIterator;
import java.util.Vector;
+import com.stevesoft.pat.Regex;
+
import fr.orsay.lri.varna.models.rna.RNA;
/**
*/
public class Sequence extends ASequence implements SequenceI
{
+ private static final Regex limitrx = new Regex(
+ "[/][0-9]{1,}[-][0-9]{1,}$");
+
+ private static final Regex endrx = new Regex("[0-9]{1,}$");
+
SequenceI datasetSequence;
String name;
*/
int index = -1;
- /**
- * array of sequence features - may not be null for a valid sequence object
+ private SequenceFeatures sequenceFeatureStore;
+
+ /*
+ * A cursor holding the approximate current view position to the sequence,
+ * as determined by findIndex or findPosition or findPositions.
+ * Using a cursor as a hint allows these methods to be more performant for
+ * large sequences.
+ */
+ private SequenceCursor cursor;
+
+ /*
+ * A number that should be incremented whenever the sequence is edited.
+ * If the value matches the cursor token, then we can trust the cursor,
+ * if not then it should be recomputed.
*/
- public SequenceFeature[] sequenceFeatures;
+ private int changeCount;
/**
* Creates a new Sequence object.
*/
public Sequence(String name, String sequence, int start, int end)
{
+ this();
initSeqAndName(name, sequence.toCharArray(), start, end);
}
public Sequence(String name, char[] sequence, int start, int end)
{
+ this();
initSeqAndName(name, sequence, start, end);
}
checkValidRange();
}
- com.stevesoft.pat.Regex limitrx = new com.stevesoft.pat.Regex(
- "[/][0-9]{1,}[-][0-9]{1,}$");
-
- com.stevesoft.pat.Regex endrx = new com.stevesoft.pat.Regex("[0-9]{1,}$");
-
void parseId()
{
if (name == null)
}
/**
+ * default constructor
+ */
+ private Sequence()
+ {
+ sequenceFeatureStore = new SequenceFeatures();
+ }
+
+ /**
* Creates a new Sequence object.
*
* @param name
*/
public Sequence(SequenceI seq, AlignmentAnnotation[] alAnnotation)
{
+ this();
initSeqFrom(seq, alAnnotation);
-
}
/**
protected void initSeqFrom(SequenceI seq,
AlignmentAnnotation[] alAnnotation)
{
- {
- char[] oseq = seq.getSequence();
- initSeqAndName(seq.getName(), Arrays.copyOf(oseq, oseq.length),
- seq.getStart(), seq.getEnd());
- }
+ char[] oseq = seq.getSequence(); // returns a copy of the array
+ initSeqAndName(seq.getName(), oseq, seq.getStart(), seq.getEnd());
+
description = seq.getDescription();
if (seq != datasetSequence)
{
setDatasetSequence(seq.getDatasetSequence());
}
- if (datasetSequence == null && seq.getDBRefs() != null)
+
+ /*
+ * only copy DBRefs and seqfeatures if we really are a dataset sequence
+ */
+ if (datasetSequence == null)
{
- // only copy DBRefs and seqfeatures if we really are a dataset sequence
- DBRefEntry[] dbr = seq.getDBRefs();
- for (int i = 0; i < dbr.length; i++)
- {
- addDBRef(new DBRefEntry(dbr[i]));
- }
- if (seq.getSequenceFeatures() != null)
+ if (seq.getDBRefs() != null)
{
- SequenceFeature[] sf = seq.getSequenceFeatures();
- for (int i = 0; i < sf.length; i++)
+ DBRefEntry[] dbr = seq.getDBRefs();
+ for (int i = 0; i < dbr.length; i++)
{
- addSequenceFeature(new SequenceFeature(sf[i]));
+ addDBRef(new DBRefEntry(dbr[i]));
}
}
+
+ /*
+ * make copies of any sequence features
+ */
+ for (SequenceFeature sf : seq.getSequenceFeatures())
+ {
+ addSequenceFeature(new SequenceFeature(sf));
+ }
}
+
if (seq.getAnnotation() != null)
{
AlignmentAnnotation[] sqann = seq.getAnnotation();
}
@Override
- public void setSequenceFeatures(SequenceFeature[] features)
+ public void setSequenceFeatures(List<SequenceFeature> features)
{
- if (datasetSequence == null)
- {
- sequenceFeatures = features;
- }
- else
+ if (datasetSequence != null)
{
- if (datasetSequence.getSequenceFeatures() != features
- && datasetSequence.getSequenceFeatures() != null
- && datasetSequence.getSequenceFeatures().length > 0)
- {
- new Exception(
- "Warning: JAL-2046 side effect ? Possible implementation error: overwriting dataset sequence features by setting sequence features on alignment")
- .printStackTrace();
- }
datasetSequence.setSequenceFeatures(features);
+ return;
}
+ sequenceFeatureStore = new SequenceFeatures(features);
}
@Override
public synchronized boolean addSequenceFeature(SequenceFeature sf)
{
- if (sequenceFeatures == null && datasetSequence != null)
- {
- return datasetSequence.addSequenceFeature(sf);
- }
- if (sequenceFeatures == null)
+ if (sf.getType() == null)
{
- sequenceFeatures = new SequenceFeature[0];
+ System.err.println("SequenceFeature type may not be null: "
+ + sf.toString());
+ return false;
}
- for (int i = 0; i < sequenceFeatures.length; i++)
+ if (datasetSequence != null)
{
- if (sequenceFeatures[i].equals(sf))
- {
- return false;
- }
+ return datasetSequence.addSequenceFeature(sf);
}
- SequenceFeature[] temp = new SequenceFeature[sequenceFeatures.length
- + 1];
- System.arraycopy(sequenceFeatures, 0, temp, 0, sequenceFeatures.length);
- temp[sequenceFeatures.length] = sf;
-
- sequenceFeatures = temp;
- return true;
+ return sequenceFeatureStore.add(sf);
}
@Override
public void deleteFeature(SequenceFeature sf)
{
- if (sequenceFeatures == null)
- {
- if (datasetSequence != null)
- {
- datasetSequence.deleteFeature(sf);
- }
- return;
- }
-
- int index = 0;
- for (index = 0; index < sequenceFeatures.length; index++)
- {
- if (sequenceFeatures[index].equals(sf))
- {
- break;
- }
- }
-
- if (index == sequenceFeatures.length)
- {
- return;
- }
-
- int sfLength = sequenceFeatures.length;
- if (sfLength < 2)
+ if (datasetSequence != null)
{
- sequenceFeatures = null;
+ datasetSequence.deleteFeature(sf);
}
else
{
- SequenceFeature[] temp = new SequenceFeature[sfLength - 1];
- System.arraycopy(sequenceFeatures, 0, temp, 0, index);
-
- if (index < sfLength)
- {
- System.arraycopy(sequenceFeatures, index + 1, temp, index,
- sequenceFeatures.length - index - 1);
- }
-
- sequenceFeatures = temp;
+ sequenceFeatureStore.delete(sf);
}
}
/**
- * Returns the sequence features (if any), looking first on the sequence, then
- * on its dataset sequence, and so on until a non-null value is found (or
- * none). This supports retrieval of sequence features stored on the sequence
- * (as in the applet) or on the dataset sequence (as in the Desktop version).
+ * {@inheritDoc}
*
* @return
*/
@Override
- public SequenceFeature[] getSequenceFeatures()
+ public List<SequenceFeature> getSequenceFeatures()
{
- SequenceFeature[] features = sequenceFeatures;
-
- SequenceI seq = this;
- int count = 0; // failsafe against loop in sequence.datasetsequence...
- while (features == null && seq.getDatasetSequence() != null
- && count++ < 10)
+ if (datasetSequence != null)
{
- seq = seq.getDatasetSequence();
- features = ((Sequence) seq).sequenceFeatures;
+ return datasetSequence.getSequenceFeatures();
}
- return features;
+ return sequenceFeatureStore.getAllFeatures();
+ }
+
+ @Override
+ public SequenceFeaturesI getFeatures()
+ {
+ return datasetSequence != null ? datasetSequence.getFeatures()
+ : sequenceFeatureStore;
}
@Override
{
this.sequence = seq.toCharArray();
checkValidRange();
+ sequenceChanged();
}
@Override
@Override
public char[] getSequence()
{
- return sequence;
+ // return sequence;
+ return sequence == null ? null : Arrays.copyOf(sequence,
+ sequence.length);
}
/*
return this.description;
}
- /*
- * (non-Javadoc)
- *
- * @see jalview.datamodel.SequenceI#findIndex(int)
+ /**
+ * {@inheritDoc}
*/
@Override
public int findIndex(int pos)
{
- // returns the alignment position for a residue
+ /*
+ * use a valid, hopefully nearby, cursor if available
+ */
+ if (isValidCursor(cursor))
+ {
+ return findIndex(pos, cursor);
+ }
+
int j = start;
int i = 0;
- // Rely on end being at least as long as the length of the sequence.
+ int startColumn = 0;
+
+ /*
+ * traverse sequence from the start counting gaps; make a note of
+ * the column of the first residue to save in the cursor
+ */
while ((i < sequence.length) && (j <= end) && (j <= pos))
{
- if (!jalview.util.Comparison.isGap(sequence[i]))
+ if (!Comparison.isGap(sequence[i]))
{
+ if (j == start)
+ {
+ startColumn = i;
+ }
j++;
}
-
i++;
}
- if ((j == end) && (j < pos))
+ if (j == end && j < pos)
{
return end + 1;
}
- else
+
+ updateCursor(pos, i, startColumn);
+ return i;
+ }
+
+ /**
+ * Updates the cursor to the latest found residue and column position
+ *
+ * @param residuePos
+ * (start..)
+ * @param column
+ * (1..)
+ * @param startColumn
+ * column position of the first sequence residue
+ */
+ protected void updateCursor(int residuePos, int column, int startColumn)
+ {
+ /*
+ * preserve end residue column provided cursor was valid
+ */
+ int endColumn = isValidCursor(cursor) ? cursor.lastColumnPosition : 0;
+ if (residuePos == this.end)
{
- return i;
+ endColumn = column;
}
+
+ cursor = new SequenceCursor(this, residuePos, column, startColumn,
+ endColumn, this.changeCount);
}
+ /**
+ * Answers the aligned column position (1..) for the given residue position
+ * (start..) given a 'hint' of a residue/column location in the neighbourhood.
+ * The hint may be left of, at, or to the right of the required position.
+ *
+ * @param pos
+ * @param curs
+ * @return
+ */
+ protected int findIndex(int pos, SequenceCursor curs)
+ {
+ if (!isValidCursor(curs))
+ {
+ /*
+ * wrong or invalidated cursor, compute de novo
+ */
+ return findIndex(pos);
+ }
+
+ if (curs.residuePosition == pos)
+ {
+ return curs.columnPosition;
+ }
+
+ /*
+ * move left or right to find pos from hint.position
+ */
+ int col = curs.columnPosition - 1; // convert from base 1 to 0-based array
+ // index
+ int newPos = curs.residuePosition;
+ int delta = newPos > pos ? -1 : 1;
+
+ while (newPos != pos)
+ {
+ col += delta; // shift one column left or right
+ if (col < 0 || col == sequence.length)
+ {
+ break;
+ }
+ if (!Comparison.isGap(sequence[col]))
+ {
+ newPos += delta;
+ }
+ }
+
+ col++; // convert back to base 1
+ updateCursor(pos, col, curs.firstColumnPosition);
+
+ return col;
+ }
+
+ /**
+ * {@inheritDoc}
+ */
@Override
- public int findPosition(int i)
+ public int findPosition(final int column)
{
+ /*
+ * use a valid, hopefully nearby, cursor if available
+ */
+ if (isValidCursor(cursor))
+ {
+ return findPosition(column + 1, cursor);
+ }
+
+ // TODO recode this more naturally i.e. count residues only
+ // as they are found, not 'in anticipation'
+
+ /*
+ * traverse the sequence counting gaps; note the column position
+ * of the first residue, to save in the cursor
+ */
+ int firstResidueColumn = 0;
+ int lastPosFound = 0;
+ int lastPosFoundColumn = 0;
+ int seqlen = sequence.length;
+
+ if (seqlen > 0 && !Comparison.isGap(sequence[0]))
+ {
+ lastPosFound = start;
+ lastPosFoundColumn = 0;
+ }
+
int j = 0;
int pos = start;
- int seqlen = sequence.length;
- while ((j < i) && (j < seqlen))
+
+ while (j < column && j < seqlen)
{
- if (!jalview.util.Comparison.isGap(sequence[j]))
+ if (!Comparison.isGap(sequence[j]))
{
+ lastPosFound = pos;
+ lastPosFoundColumn = j;
+ if (pos == this.start)
+ {
+ firstResidueColumn = j;
+ }
pos++;
}
-
j++;
}
+ if (j < seqlen && !Comparison.isGap(sequence[j]))
+ {
+ lastPosFound = pos;
+ lastPosFoundColumn = j;
+ if (pos == this.start)
+ {
+ firstResidueColumn = j;
+ }
+ }
+
+ /*
+ * update the cursor to the last residue position found (if any)
+ * (converting column position to base 1)
+ */
+ if (lastPosFound != 0)
+ {
+ updateCursor(lastPosFound, lastPosFoundColumn + 1,
+ firstResidueColumn + 1);
+ }
return pos;
}
/**
+ * Answers true if the given cursor is not null, is for this sequence object,
+ * and has a token value that matches this object's changeCount, else false.
+ * This allows us to ignore a cursor as 'stale' if the sequence has been
+ * modified since the cursor was created.
+ *
+ * @param curs
+ * @return
+ */
+ protected boolean isValidCursor(SequenceCursor curs)
+ {
+ if (curs == null || curs.sequence != this || curs.token != changeCount)
+ {
+ return false;
+ }
+ /*
+ * sanity check against range
+ */
+ if (curs.columnPosition < 0 || curs.columnPosition > sequence.length)
+ {
+ return false;
+ }
+ if (curs.residuePosition < start || curs.residuePosition > end)
+ {
+ return false;
+ }
+ return true;
+ }
+
+ /**
+ * Answers the sequence position (start..) for the given aligned column
+ * position (1..), given a hint of a cursor in the neighbourhood. The cursor
+ * may lie left of, at, or to the right of the column position.
+ *
+ * @param col
+ * @param curs
+ * @return
+ */
+ protected int findPosition(final int col, SequenceCursor curs)
+ {
+ if (!isValidCursor(curs))
+ {
+ /*
+ * wrong or invalidated cursor, compute de novo
+ */
+ return findPosition(col - 1);// ugh back to base 0
+ }
+
+ if (curs.columnPosition == col)
+ {
+ cursor = curs; // in case this method becomes public
+ return curs.residuePosition; // easy case :-)
+ }
+
+ if (curs.lastColumnPosition > 0 && curs.lastColumnPosition < col)
+ {
+ /*
+ * sequence lies entirely to the left of col
+ * - return last residue + 1
+ */
+ return end + 1;
+ }
+
+ if (curs.firstColumnPosition > 0 && curs.firstColumnPosition > col)
+ {
+ /*
+ * sequence lies entirely to the right of col
+ * - return first residue
+ */
+ return start;
+ }
+
+ // todo could choose closest to col out of column,
+ // firstColumnPosition, lastColumnPosition as a start point
+
+ /*
+ * move left or right to find pos from cursor position
+ */
+ int firstResidueColumn = curs.firstColumnPosition;
+ int column = curs.columnPosition - 1; // to base 0
+ int newPos = curs.residuePosition;
+ int delta = curs.columnPosition > col ? -1 : 1;
+ boolean gapped = false;
+ int lastFoundPosition = curs.residuePosition;
+ int lastFoundPositionColumn = curs.columnPosition;
+
+ while (column != col - 1)
+ {
+ column += delta; // shift one column left or right
+ if (column < 0 || column == sequence.length)
+ {
+ break;
+ }
+ gapped = Comparison.isGap(sequence[column]);
+ if (!gapped)
+ {
+ newPos += delta;
+ lastFoundPosition = newPos;
+ lastFoundPositionColumn = column + 1;
+ if (lastFoundPosition == this.start)
+ {
+ firstResidueColumn = column + 1;
+ }
+ }
+ }
+
+ if (cursor == null || lastFoundPosition != cursor.residuePosition)
+ {
+ updateCursor(lastFoundPosition, lastFoundPositionColumn,
+ firstResidueColumn);
+ }
+
+ /*
+ * hack to give position to the right if on a gap
+ * or beyond the length of the sequence (see JAL-2562)
+ */
+ if (delta > 0 && (gapped || column >= sequence.length))
+ {
+ newPos++;
+ }
+
+ return newPos;
+ }
+
+ /**
+ * {@inheritDoc}
+ */
+ @Override
+ public Range findPositions(int fromColumn, int toColumn)
+ {
+ if (toColumn < fromColumn || fromColumn < 1)
+ {
+ return null;
+ }
+
+ /*
+ * find the first non-gapped position, if any
+ */
+ int firstPosition = 0;
+ int col = fromColumn - 1;
+ int length = sequence.length;
+ while (col < length && col < toColumn)
+ {
+ if (!Comparison.isGap(sequence[col]))
+ {
+ firstPosition = findPosition(col++);
+ break;
+ }
+ col++;
+ }
+
+ if (firstPosition == 0)
+ {
+ return null;
+ }
+
+ /*
+ * find the last non-gapped position
+ */
+ int lastPosition = firstPosition;
+ while (col < length && col < toColumn)
+ {
+ if (!Comparison.isGap(sequence[col++]))
+ {
+ lastPosition++;
+ }
+ }
+
+ return new Range(firstPosition, lastPosition);
+ }
+
+ /**
* Returns an int array where indices correspond to each residue in the
* sequence and the element value gives its position in the alignment
*
start = newstart;
end = newend;
sequence = tmp;
+ sequenceChanged();
}
@Override
}
sequence = tmp;
+ sequenceChanged();
}
@Override
private boolean _isNa;
- private long _seqhash = 0;
+ private int _seqhash = 0;
/**
* Answers false if the sequence is more than 85% nucleotide (ACGTU), else
dsseq.setDescription(description);
// move features and database references onto dataset sequence
- dsseq.sequenceFeatures = sequenceFeatures;
- sequenceFeatures = null;
+ dsseq.sequenceFeatureStore = sequenceFeatureStore;
+ sequenceFeatureStore = null;
dsseq.dbrefs = dbrefs;
dbrefs = null;
// TODO: search and replace any references to this sequence with
return null;
}
- Vector subset = new Vector();
- Enumeration e = annotation.elements();
+ Vector<AlignmentAnnotation> subset = new Vector<AlignmentAnnotation>();
+ Enumeration<AlignmentAnnotation> e = annotation.elements();
while (e.hasMoreElements())
{
- AlignmentAnnotation ann = (AlignmentAnnotation) e.nextElement();
+ AlignmentAnnotation ann = e.nextElement();
if (ann.label != null && ann.label.equals(label))
{
subset.addElement(ann);
e = subset.elements();
while (e.hasMoreElements())
{
- anns[i++] = (AlignmentAnnotation) e.nextElement();
+ anns[i++] = e.nextElement();
}
subset.removeAllElements();
return anns;
if (entry.getSequenceFeatures() != null)
{
- SequenceFeature[] sfs = entry.getSequenceFeatures();
- for (int si = 0; si < sfs.length; si++)
+ List<SequenceFeature> sfs = entry.getSequenceFeatures();
+ for (SequenceFeature feature : sfs)
{
- SequenceFeature sf[] = (mp != null) ? mp.locateFeature(sfs[si])
- : new SequenceFeature[]
- { new SequenceFeature(sfs[si]) };
- if (sf != null && sf.length > 0)
+ SequenceFeature sf[] = (mp != null) ? mp.locateFeature(feature)
+ : new SequenceFeature[] { new SequenceFeature(feature) };
+ if (sf != null)
{
for (int sfi = 0; sfi < sf.length; sfi++)
{
// transfer PDB entries
if (entry.getAllPDBEntries() != null)
{
- Enumeration e = entry.getAllPDBEntries().elements();
+ Enumeration<PDBEntry> e = entry.getAllPDBEntries().elements();
while (e.hasMoreElements())
{
- PDBEntry pdb = (PDBEntry) e.nextElement();
+ PDBEntry pdb = e.nextElement();
addPDBId(pdb);
}
}
}
}
+ /**
+ * {@inheritDoc}
+ */
+ @Override
+ public List<SequenceFeature> findFeatures(int fromColumn, int toColumn,
+ String... types)
+ {
+ int startPos = findPosition(fromColumn - 1); // convert base 1 to base 0
+ int endPos = fromColumn == toColumn ? startPos
+ : findPosition(toColumn - 1);
+
+ List<SequenceFeature> result = getFeatures().findFeatures(startPos,
+ endPos, types);
+
+ /*
+ * if end column is gapped, endPos may be to the right,
+ * and we may have included adjacent or enclosing features;
+ * remove any that are not enclosing, non-contact features
+ */
+ if (endPos > this.end || Comparison.isGap(sequence[toColumn - 1]))
+ {
+ ListIterator<SequenceFeature> it = result.listIterator();
+ while (it.hasNext())
+ {
+ SequenceFeature sf = it.next();
+ int sfBegin = sf.getBegin();
+ int sfEnd = sf.getEnd();
+ int featureStartColumn = findIndex(sfBegin);
+ if (featureStartColumn > toColumn)
+ {
+ it.remove();
+ }
+ else if (featureStartColumn < fromColumn)
+ {
+ int featureEndColumn = sfEnd == sfBegin ? featureStartColumn
+ : findIndex(sfEnd);
+ if (featureEndColumn < fromColumn)
+ {
+ it.remove();
+ }
+ else if (featureEndColumn > toColumn && sf.isContactFeature())
+ {
+ /*
+ * remove an enclosing feature if it is a contact feature
+ */
+ it.remove();
+ }
+ }
+ }
+ }
+
+ return result;
+ }
+
+ /**
+ * Invalidates any stale cursors (forcing recalculation) by incrementing the
+ * token that has to match the one presented by the cursor
+ */
+ @Override
+ public void sequenceChanged()
+ {
+ changeCount++;
+ }
+
+ /**
+ * {@inheritDoc}
+ */
+ @Override
+ public int replace(char c1, char c2)
+ {
+ if (c1 == c2)
+ {
+ return 0;
+ }
+ int count = 0;
+ synchronized (sequence)
+ {
+ for (int c = 0; c < sequence.length; c++)
+ {
+ if (sequence[c] == c1)
+ {
+ sequence[c] = c2;
+ count++;
+ }
+ }
+ }
+ if (count > 0)
+ {
+ sequenceChanged();
+ }
+
+ return count;
+ }
}
git://source.jalview.org / jalview.git / blobdiff