import java.util.ArrayList;
import java.util.Arrays;
+import java.util.BitSet;
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;
/**
*
- * Implements the SequenceI interface for a char[] based sequence object.
- *
- * @author $author$
- * @version $Revision$
+ * Implements the SequenceI interface for a char[] based sequence object
*/
public class Sequence extends ASequence implements SequenceI
{
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.
+ */
+ private int changeCount;
+
/**
* Creates a new Sequence object.
*
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)
{
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 (!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)
+ {
+ int endColumn = cursor == null ? 0 : cursor.lastColumnPosition;
+ if (residuePos == this.end)
{
- return i;
+ endColumn = column;
}
+
+ cursor = new SequenceCursor(this, residuePos, column, startColumn,
+ endColumn, this.changeCount);
}
- @Override
- public int findPosition(int i)
+ /**
+ * 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)
{
- int j = 0;
- int pos = start;
- int seqlen = sequence.length;
- while ((j < i) && (j < seqlen))
+ if (!isValidCursor(curs))
{
- if (!Comparison.isGap(sequence[j]))
+ /*
+ * 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)
{
- pos++;
+ break;
+ }
+ if (!Comparison.isGap(sequence[col]))
+ {
+ newPos += delta;
}
-
- j++;
}
- return pos;
+ col++; // convert back to base 1
+ updateCursor(pos, col, curs.firstColumnPosition);
+
+ return col;
}
/**
* {@inheritDoc}
*/
@Override
- public Range findPositions(int fromCol, int toCol)
+ public int findPosition(final int column)
{
/*
- * count residues before fromCol
+ * use a valid, hopefully nearby, cursor if available
*/
- int j = 0;
- int count = 0;
+ 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;
- while (j < fromCol && j < seqlen)
+
+ if (seqlen > 0 && !Comparison.isGap(sequence[0]))
+ {
+ lastPosFound = start;
+ lastPosFoundColumn = 0;
+ }
+
+ int j = 0;
+ int pos = start;
+
+ while (j < column && j < seqlen)
{
if (!Comparison.isGap(sequence[j]))
{
- count++;
+ 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;
+ }
+ }
/*
- * find first and last residues between fromCol and toCol
+ * update the cursor to the last residue position found (if any)
+ * (converting column position to base 1)
*/
- int firstPos = 0;
- int lastPos = 0;
- int firstPosCol = 0;
- boolean foundFirst = false;
-
- while (j <= toCol && j < seqlen)
+ if (lastPosFound != 0)
{
- if (!Comparison.isGap(sequence[j]))
+ 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)
{
- count++;
- if (!foundFirst)
+ newPos += delta;
+ lastFoundPosition = newPos;
+ lastFoundPositionColumn = column + 1;
+ if (lastFoundPosition == this.start)
{
- firstPos = count;
- firstPosCol = j;
- foundFirst = true;
+ firstResidueColumn = column + 1;
}
- lastPos = count;
}
- j++;
}
- if (firstPos == 0)
+ if (cursor == null || lastFoundPosition != cursor.residuePosition)
{
- /*
- * no residues in this range
- */
- return null;
+ updateCursor(lastFoundPosition, lastFoundPositionColumn,
+ firstResidueColumn);
}
/*
- * adjust for sequence start coordinate
+ * hack to give position to the right if on a gap
+ * or beyond the length of the sequence (see JAL-2562)
*/
- firstPos += start - 1;
- lastPos += start - 1;
+ if (delta > 0 && (gapped || column >= sequence.length))
+ {
+ newPos++;
+ }
- return new Range(firstPos, lastPos);
+ return newPos;
}
/**
}
@Override
+ public BitSet getInsertionsAsBits()
+ {
+ BitSet map = new BitSet();
+ int lastj = -1, j = 0;
+ int pos = start;
+ int seqlen = sequence.length;
+ while ((j < seqlen))
+ {
+ if (jalview.util.Comparison.isGap(sequence[j]))
+ {
+ if (lastj == -1)
+ {
+ lastj = j;
+ }
+ }
+ else
+ {
+ if (lastj != -1)
+ {
+ map.set(lastj, j);
+ lastj = -1;
+ }
+ }
+ j++;
+ }
+ if (lastj != -1)
+ {
+ map.set(lastj, j);
+ lastj = -1;
+ }
+ return map;
+ }
+
+ @Override
public void deleteChars(int i, int j)
{
int newstart = start, newend = end;
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
* {@inheritDoc}
*/
@Override
- public List<SequenceFeature> findFeatures(int from, int to,
+ public List<SequenceFeature> findFeatures(int fromColumn, int toColumn,
String... types)
{
+ int startPos = findPosition(fromColumn - 1); // convert base 1 to base 0
+ int endPos = findPosition(toColumn - 1);
+ // to trace / debug behaviour:
+ // System.out
+ // .println(String
+ // .format("%s.findFeatures columns [%d-%d] positions [%d-%d] leaves cursor %s",
+ // getName(), fromColumn, toColumn, startPos,
+ // endPos, cursor));
+ List<SequenceFeature> result = new ArrayList<>();
if (datasetSequence != null)
{
- return datasetSequence.findFeatures(from, to, types);
+ result = datasetSequence.getFeatures().findFeatures(startPos, endPos,
+ types);
+ }
+ else
+ {
+ result = sequenceFeatureStore.findFeatures(startPos, endPos, types);
+ }
+
+ /*
+ * if the start or end column is gapped, startPos or endPos may be to the
+ * left or 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[fromColumn - 1])
+ || Comparison.isGap(sequence[toColumn - 1]))
+ {
+ ListIterator<SequenceFeature> it = result.listIterator();
+ while (it.hasNext())
+ {
+ SequenceFeature sf = it.next();
+ int featureStartColumn = findIndex(sf.getBegin());
+ int featureEndColumn = findIndex(sf.getEnd());
+ boolean noOverlap = featureStartColumn > toColumn
+ || featureEndColumn < fromColumn;
+
+ /*
+ * reject an 'enclosing' feature if it is actually a contact feature
+ */
+ if (sf.isContactFeature() && featureStartColumn < fromColumn
+ && featureEndColumn > toColumn)
+ {
+ noOverlap = true;
+ }
+ if (noOverlap)
+ {
+ 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 sequenceFeatureStore.findFeatures(from, to, types);
+
+ return count;
+ }
+
+ @Override
+ public List<SequenceFeature[]> adjustFeatures(int fromColumn, int toColumn)
+ {
+ List<SequenceFeature[]> amended = new ArrayList<>();
+
+ if (toColumn < fromColumn)
+ {
+ return amended;
+ }
+
+ synchronized (sequenceFeatureStore)
+ {
+ /*
+ * get features that overlap or span the cut region
+ */
+ List<SequenceFeature> overlaps = findFeatures(fromColumn, toColumn);
+ int cutWidth = toColumn - fromColumn + 1;
+
+ /*
+ * get features that strictly follow the cut region,
+ * and shift them left by the width of the cut
+ */
+ List<SequenceFeature> follow = findFeatures(toColumn + 1,
+ Integer.MAX_VALUE);
+ follow.removeAll(overlaps);
+ for (SequenceFeature sf : follow)
+ {
+ SequenceFeature copy = new SequenceFeature(sf, sf.getBegin()
+ - cutWidth, sf.getEnd() - cutWidth, sf.getFeatureGroup(),
+ sf.getScore());
+ deleteFeature(sf);
+ addSequenceFeature(copy);
+ }
+
+ /*
+ * adjust start-end of overlapping features, and delete if enclosed by
+ * the cut, or a partially overlapping contact feature
+ */
+ for (SequenceFeature sf : overlaps)
+ {
+ // TODO recode to compute newBegin, newEnd, isDelete
+ // then perform the action
+ int sfBegin = sf.getBegin();
+ int sfEnd = sf.getEnd();
+ int startCol = findIndex(sfBegin);
+ int endCol = findIndex(sfEnd);
+ if (startCol >= fromColumn && endCol <= toColumn)
+ {
+ // within cut region - delete feature
+ deleteFeature(sf);
+ amended.add(new SequenceFeature[] { sf, null });
+ continue;
+ }
+ if (startCol < fromColumn && endCol > toColumn)
+ {
+ // feature spans cut region - shift end left
+ SequenceFeature copy = new SequenceFeature(sf, sf.getBegin(),
+ sf.getEnd() - cutWidth, sf.getFeatureGroup(),
+ sf.getScore());
+ deleteFeature(sf);
+ addSequenceFeature(copy);
+ amended.add(new SequenceFeature[] { sf, copy });
+ continue;
+ }
+ // todo partial overlap - delete if contact feature
+ }
+ }
+
+ return amended;
}
}