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
28 /** Data structure to hold and manipulate a multiple sequence alignment
\r
30 public class Alignment implements AlignmentI
\r
32 protected Vector sequences;
\r
33 protected Vector groups = new Vector();
\r
34 protected Vector superGroup = new Vector();
\r
35 protected char gapCharacter = '-';
\r
36 protected boolean isNucleotide = true;
\r
39 /** DOCUMENT ME!! */
\r
40 public AlignmentAnnotation[] annotations;
\r
42 /** DOCUMENT ME!! */
\r
43 public boolean featuresAdded = false;
\r
45 /** Make an alignment from an array of Sequences.
\r
49 public Alignment(SequenceI[] seqs)
\r
51 int i=0, iSize = seqs.length, j, jSize;
\r
52 while(isNucleotide && i<iSize)
\r
54 jSize = seqs[i].getLength();
\r
55 for(j=0; j<jSize; j++)
\r
57 if(!jalview.schemes.ResidueProperties.nucleotideHash.containsKey(seqs[i].getSequence(j, j+1))
\r
58 && !jalview.util.Comparison.isGap(seqs[i].getSequence().charAt(j)))
\r
60 isNucleotide = false;
\r
67 sequences = new Vector();
\r
69 for (i = 0; i < seqs.length; i++)
\r
71 sequences.addElement(seqs[i]);
\r
80 * @return DOCUMENT ME!
\r
82 public Vector getSequences()
\r
90 * @param i DOCUMENT ME!
\r
92 * @return DOCUMENT ME!
\r
94 public SequenceI getSequenceAt(int i)
\r
96 if (i < sequences.size())
\r
98 return (SequenceI) sequences.elementAt(i);
\r
104 /** Adds a sequence to the alignment. Recalculates maxLength and size.
\r
108 public void addSequence(SequenceI snew)
\r
110 sequences.addElement(snew);
\r
116 * @param seq DOCUMENT ME!
\r
118 public void addSequence(SequenceI[] seq)
\r
120 for (int i = 0; i < seq.length; i++)
\r
122 addSequence(seq[i]);
\r
126 /** Adds a sequence to the alignment. Recalculates maxLength and size.
\r
130 public void setSequenceAt(int i, SequenceI snew)
\r
132 SequenceI oldseq = getSequenceAt(i);
\r
133 deleteSequence(oldseq);
\r
135 sequences.setElementAt(snew, i);
\r
141 * @return DOCUMENT ME!
\r
143 public Vector getGroups()
\r
148 /** Takes out columns consisting entirely of gaps (-,.," ")
\r
150 public void removeGaps()
\r
153 int iSize = getWidth();
\r
155 for (int i = 0; i < iSize; i++)
\r
157 boolean delete = true;
\r
159 for (int j = 0; j < getHeight(); j++)
\r
161 current = getSequenceAt(j);
\r
163 if (current.getLength() > i)
\r
165 /* MC Should move this to a method somewhere */
\r
166 if (!jalview.util.Comparison.isGap(current.getCharAt(i)))
\r
175 deleteColumns(i, i);
\r
182 /** Removes a range of columns (start to end inclusive).
\r
184 * @param start Start column in the alignment
\r
185 * @param end End column in the alignment
\r
187 public void deleteColumns(int start, int end)
\r
189 deleteColumns(0, getHeight() - 1, start, end);
\r
195 * @param seq1 DOCUMENT ME!
\r
196 * @param seq2 DOCUMENT ME!
\r
197 * @param start DOCUMENT ME!
\r
198 * @param end DOCUMENT ME!
\r
200 public void deleteColumns(int seq1, int seq2, int start, int end)
\r
202 for (int i = 0; i <= (end - start); i++)
\r
204 for (int j = seq1; j <= seq2; j++)
\r
206 getSequenceAt(j).deleteCharAt(start);
\r
214 * @param i DOCUMENT ME!
\r
216 public void trimLeft(int i)
\r
218 for (int j = 0; j < getHeight(); j++)
\r
220 SequenceI s = getSequenceAt(j);
\r
221 int newstart = s.findPosition(i);
\r
223 s.setStart(newstart);
\r
224 s.setSequence(s.getSequence().substring(i));
\r
231 * @param i DOCUMENT ME!
\r
233 public void trimRight(int i)
\r
235 for (int j = 0; j < getHeight(); j++)
\r
237 SequenceI s = getSequenceAt(j);
\r
238 int newend = s.findPosition(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 * @param sg DOCUMENT ME!
\r
387 public void addSuperGroup(SuperGroup sg)
\r
389 superGroup.addElement(sg);
\r
395 * @param sg DOCUMENT ME!
\r
397 public void removeSuperGroup(SuperGroup sg)
\r
399 superGroup.removeElement(sg);
\r
405 * @param sg DOCUMENT ME!
\r
407 * @return DOCUMENT ME!
\r
409 public SuperGroup getSuperGroup(SequenceGroup sg)
\r
411 for (int i = 0; i < this.superGroup.size(); i++)
\r
413 SuperGroup temp = (SuperGroup) superGroup.elementAt(i);
\r
415 if (temp.sequenceGroups.contains(sg))
\r
425 public void addGroup(SequenceGroup sg)
\r
427 if (!groups.contains(sg))
\r
429 groups.addElement(sg);
\r
436 public void deleteAllGroups()
\r
438 groups.removeAllElements();
\r
439 superGroup.removeAllElements();
\r
443 while (i < sequences.size())
\r
445 SequenceI s = getSequenceAt(i);
\r
446 s.setColor(java.awt.Color.white);
\r
452 public void deleteGroup(SequenceGroup g)
\r
454 if (groups.contains(g))
\r
456 groups.removeElement(g);
\r
461 public SequenceI findName(String name)
\r
465 while (i < sequences.size())
\r
467 SequenceI s = getSequenceAt(i);
\r
469 if (s.getName().equals(name))
\r
481 public SequenceI findbyDisplayId(String name)
\r
485 while (i < sequences.size())
\r
487 SequenceI s = getSequenceAt(i);
\r
489 if (s.getDisplayId().equals(name))
\r
501 public int findIndex(SequenceI s)
\r
505 while (i < sequences.size())
\r
507 if (s == getSequenceAt(i))
\r
521 * @return DOCUMENT ME!
\r
523 public int getHeight()
\r
525 return sequences.size();
\r
531 * @return DOCUMENT ME!
\r
533 public int getWidth()
\r
535 int maxLength = -1;
\r
537 for (int i = 0; i < sequences.size(); i++)
\r
539 if (getSequenceAt(i).getLength() > maxLength)
\r
541 maxLength = getSequenceAt(i).getLength();
\r
551 * @return DOCUMENT ME!
\r
553 public int getMaxIdLength()
\r
558 while (i < sequences.size())
\r
560 SequenceI seq = getSequenceAt(i);
\r
561 String tmp = seq.getName() + "/" + seq.getStart() + "-" +
\r
564 if (tmp.length() > max)
\r
566 max = tmp.length();
\r
578 * @param gc DOCUMENT ME!
\r
580 public void setGapCharacter(char gc)
\r
584 for (int i = 0; i < sequences.size(); i++)
\r
586 Sequence seq = (Sequence) sequences.elementAt(i);
\r
587 seq.sequence = seq.sequence.replace('.', gc);
\r
588 seq.sequence = seq.sequence.replace('-', gc);
\r
589 seq.sequence = seq.sequence.replace(' ', gc);
\r
596 * @return DOCUMENT ME!
\r
598 public char getGapCharacter()
\r
600 return gapCharacter;
\r
606 * @return DOCUMENT ME!
\r
608 public Vector getAAFrequency()
\r
610 return AAFrequency.calculate(sequences, 0, getWidth());
\r
616 * @return DOCUMENT ME!
\r
618 public boolean isAligned()
\r
620 int width = getWidth();
\r
622 for (int i = 0; i < sequences.size(); i++)
\r
624 if (getSequenceAt(i).getLength() != width)
\r
636 * @param aa DOCUMENT ME!
\r
638 public void deleteAnnotation(AlignmentAnnotation aa)
\r
642 if (annotations != null)
\r
644 aSize = annotations.length;
\r
647 AlignmentAnnotation[] temp = new AlignmentAnnotation[aSize - 1];
\r
651 for (int i = 0; i < aSize; i++)
\r
653 if (annotations[i] == aa)
\r
658 temp[tIndex] = annotations[i];
\r
662 annotations = temp;
\r
668 * @param aa DOCUMENT ME!
\r
670 public void addAnnotation(AlignmentAnnotation aa)
\r
674 if (annotations != null)
\r
676 aSize = annotations.length + 1;
\r
679 AlignmentAnnotation[] temp = new AlignmentAnnotation[aSize];
\r
684 for (i = 0; i < (aSize - 1); i++)
\r
686 temp[i] = annotations[i];
\r
692 annotations = temp;
\r
698 * @return DOCUMENT ME!
\r
700 public AlignmentAnnotation[] getAlignmentAnnotation()
\r
702 return annotations;
\r
705 public void setNucleotide(boolean b)
\r
710 public boolean isNucleotide()
\r
712 return isNucleotide;
\r