2 * Jalview - A Sequence Alignment Editor and Viewer
\r
3 * Copyright (C) 2005 AM Waterhouse, J Procter, G Barton, M Clamp, S Searle
\r
5 * This program is free software; you can redistribute it and/or
\r
6 * modify it under the terms of the GNU General Public License
\r
7 * as published by the Free Software Foundation; either version 2
\r
8 * of the License, or (at your option) any later version.
\r
10 * This program is distributed in the hope that it will be useful,
\r
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
\r
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
\r
13 * GNU General Public License for more details.
\r
15 * You should have received a copy of the GNU General Public License
\r
16 * along with this program; if not, write to the Free Software
\r
17 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA
\r
19 package jalview.datamodel;
\r
21 import jalview.analysis.*;
\r
23 import jalview.util.*;
\r
27 /** Data structure to hold and manipulate a multiple sequence alignment
\r
29 public class Alignment implements AlignmentI
\r
31 protected Alignment dataset;
\r
32 protected Vector sequences;
\r
33 protected Vector groups = new Vector();
\r
34 protected char gapCharacter = '-';
\r
35 protected int type = NUCLEOTIDE;
\r
36 public static final int PROTEIN = 0;
\r
37 public static final int NUCLEOTIDE = 1;
\r
39 /** DOCUMENT ME!! */
\r
40 public AlignmentAnnotation[] annotations;
\r
42 HiddenSequences hiddenSequences = new HiddenSequences(this);
\r
45 /** Make an alignment from an array of Sequences.
\r
49 public Alignment(SequenceI[] seqs)
\r
53 if( jalview.util.Comparison.isNucleotide(seqs))
\r
58 sequences = new Vector();
\r
60 for (i = 0; i < seqs.length; i++)
\r
62 sequences.addElement(seqs[i]);
\r
69 * @return DOCUMENT ME!
\r
71 public Vector getSequences()
\r
79 * @param i DOCUMENT ME!
\r
81 * @return DOCUMENT ME!
\r
83 public SequenceI getSequenceAt(int i)
\r
85 if (i < sequences.size())
\r
87 return (SequenceI) sequences.elementAt(i);
\r
93 /** Adds a sequence to the alignment. Recalculates maxLength and size.
\r
97 public void addSequence(SequenceI snew)
\r
99 sequences.addElement(snew);
\r
105 * @param seq DOCUMENT ME!
\r
107 public void addSequence(SequenceI[] seq)
\r
109 for (int i = 0; i < seq.length; i++)
\r
111 addSequence(seq[i]);
\r
115 /** Adds a sequence to the alignment. Recalculates maxLength and size.
\r
119 public void setSequenceAt(int i, SequenceI snew)
\r
121 SequenceI oldseq = getSequenceAt(i);
\r
122 deleteSequence(oldseq);
\r
124 sequences.setElementAt(snew, i);
\r
130 * @return DOCUMENT ME!
\r
132 public Vector getGroups()
\r
137 /** Takes out columns consisting entirely of gaps (-,.," ")
\r
139 public void removeGaps()
\r
142 int iSize = getWidth();
\r
144 for (int i = 0; i < iSize; i++)
\r
146 boolean delete = true;
\r
148 for (int j = 0; j < getHeight(); j++)
\r
150 current = getSequenceAt(j);
\r
152 if (current.getLength() > i)
\r
154 /* MC Should move this to a method somewhere */
\r
155 if (!jalview.util.Comparison.isGap(current.getCharAt(i)))
\r
164 deleteColumns(i, i);
\r
171 /** Removes a range of columns (start to end inclusive).
\r
173 * @param start Start column in the alignment
\r
174 * @param end End column in the alignment
\r
176 public void deleteColumns(int start, int end)
\r
178 deleteColumns(0, getHeight() - 1, start, end);
\r
184 * @param seq1 DOCUMENT ME!
\r
185 * @param seq2 DOCUMENT ME!
\r
186 * @param start DOCUMENT ME!
\r
187 * @param end DOCUMENT ME!
\r
189 public void deleteColumns(int seq1, int seq2, int start, int end)
\r
191 for (int i = 0; i <= (end - start); i++)
\r
193 for (int j = seq1; j <= seq2; j++)
\r
195 getSequenceAt(j).deleteCharAt(start);
\r
203 * @param i DOCUMENT ME!
\r
205 public void trimLeft(int i)
\r
207 int j, jSize = getHeight();
\r
208 for (j = 0; j < jSize; j++)
\r
210 SequenceI s = getSequenceAt(j);
\r
211 int newstart = s.findPosition(i);
\r
213 if(i>s.getLength())
\r
215 sequences.removeElement(s);
\r
221 s.setStart(newstart);
\r
222 s.setSequence(s.getSequence().substring(i));
\r
230 * @param i DOCUMENT ME!
\r
232 public void trimRight(int i)
\r
234 for (int j = 0; j < getHeight(); j++)
\r
236 SequenceI s = getSequenceAt(j);
\r
237 int newend = s.findPosition(i);
\r
240 if(s.getLength()>i)
\r
241 s.setSequence(s.getSequence().substring(0, i + 1));
\r
248 * @param s DOCUMENT ME!
\r
250 public void deleteSequence(SequenceI s)
\r
252 for (int i = 0; i < getHeight(); i++)
\r
254 if (getSequenceAt(i) == s)
\r
264 * @param i DOCUMENT ME!
\r
266 public void deleteSequence(int i)
\r
268 sequences.removeElementAt(i);
\r
274 * @param threshold DOCUMENT ME!
\r
275 * @param sel DOCUMENT ME!
\r
277 * @return DOCUMENT ME!
\r
279 public Vector removeRedundancy(float threshold, Vector sel)
\r
281 Vector del = new Vector();
\r
283 for (int i = 1; i < sel.size(); i++)
\r
285 for (int j = 0; j < i; j++)
\r
287 // Only do the comparison if either have not been deleted
\r
288 if (!del.contains((SequenceI) sel.elementAt(i)) ||
\r
289 !del.contains((SequenceI) sel.elementAt(j)))
\r
291 // use PID instead of Comparison (which is really not pleasant)
\r
292 float pid = Comparison.PID((SequenceI) sel.elementAt(j),
\r
293 (SequenceI) sel.elementAt(i));
\r
295 if (pid >= threshold)
\r
297 // Delete the shortest one
\r
298 if (((SequenceI) sel.elementAt(j)).getSequence().length() > ((SequenceI) sel
\r
300 i)).getSequence().length())
\r
302 del.addElement(sel.elementAt(i));
\r
306 del.addElement(sel.elementAt(i));
\r
313 // Now delete the sequences
\r
314 for (int i = 0; i < del.size(); i++)
\r
316 deleteSequence((SequenceI) del.elementAt(i));
\r
323 public SequenceGroup findGroup(int i)
\r
325 return findGroup(getSequenceAt(i));
\r
329 public SequenceGroup findGroup(SequenceI s)
\r
331 for (int i = 0; i < this.groups.size(); i++)
\r
333 SequenceGroup sg = (SequenceGroup) groups.elementAt(i);
\r
335 if (sg.sequences.contains(s))
\r
347 * @param s DOCUMENT ME!
\r
349 * @return DOCUMENT ME!
\r
351 public SequenceGroup[] findAllGroups(SequenceI s)
\r
353 Vector temp = new Vector();
\r
355 int gSize = groups.size();
\r
356 for (int i = 0; i < gSize; i++)
\r
358 SequenceGroup sg = (SequenceGroup) groups.elementAt(i);
\r
359 if(sg==null || sg.sequences==null)
\r
361 this.deleteGroup(sg);
\r
366 if (sg.sequences.contains(s))
\r
368 temp.addElement(sg);
\r
372 SequenceGroup[] ret = new SequenceGroup[temp.size()];
\r
374 for (int i = 0; i < temp.size(); i++)
\r
376 ret[i] = (SequenceGroup) temp.elementAt(i);
\r
385 public void addGroup(SequenceGroup sg)
\r
387 if (!groups.contains(sg))
\r
389 groups.addElement(sg);
\r
396 public void deleteAllGroups()
\r
398 groups.removeAllElements();
\r
402 while (i < sequences.size())
\r
404 SequenceI s = getSequenceAt(i);
\r
405 s.setColor(java.awt.Color.white);
\r
411 public void deleteGroup(SequenceGroup g)
\r
413 if (groups.contains(g))
\r
415 groups.removeElement(g);
\r
420 public SequenceI findName(String name)
\r
424 while (i < sequences.size())
\r
426 if (getSequenceAt(i).getName().equals(name))
\r
428 return getSequenceAt(i);
\r
439 public int findIndex(SequenceI s)
\r
443 while (i < sequences.size())
\r
445 if (s == getSequenceAt(i))
\r
459 * @return DOCUMENT ME!
\r
461 public int getHeight()
\r
463 return sequences.size();
\r
469 * @return DOCUMENT ME!
\r
471 public int getWidth()
\r
473 int maxLength = -1;
\r
475 for (int i = 0; i < sequences.size(); i++)
\r
477 if (getSequenceAt(i).getLength() > maxLength)
\r
479 maxLength = getSequenceAt(i).getLength();
\r
489 * @return DOCUMENT ME!
\r
491 public int getMaxIdLength()
\r
496 while (i < sequences.size())
\r
498 SequenceI seq = getSequenceAt(i);
\r
499 String tmp = seq.getName() + "/" + seq.getStart() + "-" +
\r
502 if (tmp.length() > max)
\r
504 max = tmp.length();
\r
516 * @param gc DOCUMENT ME!
\r
518 public void setGapCharacter(char gc)
\r
522 for (int i = 0; i < sequences.size(); i++)
\r
524 Sequence seq = (Sequence) sequences.elementAt(i);
\r
525 seq.sequence = seq.sequence.replace('.', gc);
\r
526 seq.sequence = seq.sequence.replace('-', gc);
\r
527 seq.sequence = seq.sequence.replace(' ', gc);
\r
534 * @return DOCUMENT ME!
\r
536 public char getGapCharacter()
\r
538 return gapCharacter;
\r
544 * @return DOCUMENT ME!
\r
546 public Vector getAAFrequency()
\r
548 return AAFrequency.calculate(sequences, 0, getWidth());
\r
554 * @return DOCUMENT ME!
\r
556 public boolean isAligned()
\r
558 int width = getWidth();
\r
560 for (int i = 0; i < sequences.size(); i++)
\r
562 if (getSequenceAt(i).getLength() != width)
\r
574 * @param aa DOCUMENT ME!
\r
576 public void deleteAnnotation(AlignmentAnnotation aa)
\r
580 if (annotations != null)
\r
582 aSize = annotations.length;
\r
585 AlignmentAnnotation[] temp = new AlignmentAnnotation[aSize - 1];
\r
589 for (int i = 0; i < aSize; i++)
\r
591 if (annotations[i] == aa)
\r
596 temp[tIndex] = annotations[i];
\r
600 annotations = temp;
\r
604 public void adjustSequenceAnnotations()
\r
606 if(annotations!=null)
\r
608 for (int a = 0; a < annotations.length; a++)
\r
610 if (annotations[a].sequenceRef != null)
\r
612 annotations[a].adjustForAlignment();
\r
621 * @param aa DOCUMENT ME!
\r
623 public void addAnnotation(AlignmentAnnotation aa)
\r
626 if (annotations != null)
\r
628 aSize = annotations.length + 1;
\r
631 AlignmentAnnotation[] temp = new AlignmentAnnotation[aSize];
\r
633 temp[aSize-1] = aa;
\r
639 for (i = 0; i < (aSize-1); i++)
\r
641 temp[i] = annotations[i];
\r
645 annotations = temp;
\r
648 public void setAnnotationIndex(AlignmentAnnotation aa, int index)
\r
650 if(aa==null || annotations==null || annotations.length-1<index)
\r
653 int aSize = annotations.length;
\r
654 AlignmentAnnotation[] temp = new AlignmentAnnotation[aSize];
\r
658 for (int i = 0; i < aSize; i++)
\r
664 temp[i] = annotations[i];
\r
666 temp[i] = annotations[i-1];
\r
669 annotations = temp;
\r
675 * @return DOCUMENT ME!
\r
677 public AlignmentAnnotation[] getAlignmentAnnotation()
\r
679 return annotations;
\r
682 public void setNucleotide(boolean b)
\r
690 public boolean isNucleotide()
\r
692 if(type==NUCLEOTIDE)
\r
698 public void setDataset(Alignment data)
\r
700 if(dataset==null && data==null)
\r
702 // Create a new dataset for this alignment.
\r
703 // Can only be done once, if dataset is not null
\r
704 // This will not be performed
\r
705 Sequence[] seqs = new Sequence[getHeight()];
\r
706 for (int i = 0; i < getHeight(); i++)
\r
709 seqs[i] = new Sequence(getSequenceAt(i).getName(),
\r
710 AlignSeq.extractGaps(
\r
711 jalview.util.Comparison.GapChars,
\r
712 getSequenceAt(i).getSequence()
\r
714 getSequenceAt(i).getStart(),
\r
715 getSequenceAt(i).getEnd());
\r
717 getSequenceAt(i).setDatasetSequence(seqs[i]);
\r
720 dataset = new Alignment(seqs);
\r
722 else if(dataset==null && data!=null)
\r
728 public Alignment getDataset()
\r
733 public boolean padGaps() {
\r
734 boolean modified=false;
\r
736 //Remove excess gaps from the end of alignment
\r
737 int maxLength = -1;
\r
740 for (int i = 0; i < sequences.size(); i++)
\r
742 current = getSequenceAt(i);
\r
743 for (int j = current.getLength(); j > maxLength; j--)
\r
745 if (j > maxLength && !jalview.util.Comparison.isGap(
\r
746 current.getCharAt(j)))
\r
756 for (int i = 0; i < sequences.size();
\r
759 current = getSequenceAt(i);
\r
761 if (current.getLength() < maxLength)
\r
763 current.insertCharAt(maxLength - 1, gapCharacter);
\r
766 else if(current.getLength() > maxLength)
\r
768 current.deleteChars(maxLength, current.getLength());
\r
774 public HiddenSequences getHiddenSequences()
\r
776 return hiddenSequences;
\r