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 Alignment dataset;
\r
33 protected Vector sequences;
\r
34 protected Vector groups = new Vector();
\r
35 protected Vector superGroup = new Vector();
\r
36 protected char gapCharacter = '-';
\r
37 protected int type = NUCLEOTIDE;
\r
38 public static final int PROTEIN = 0;
\r
39 public static final int NUCLEOTIDE = 1;
\r
41 /** DOCUMENT ME!! */
\r
42 public AlignmentAnnotation[] annotations;
\r
44 /** Make an alignment from an array of Sequences.
\r
48 public Alignment(SequenceI[] seqs)
\r
52 if( jalview.util.Comparison.isNucleotide(seqs))
\r
57 sequences = new Vector();
\r
59 for (i = 0; i < seqs.length; i++)
\r
61 sequences.addElement(seqs[i]);
\r
70 * @return DOCUMENT ME!
\r
72 public Vector getSequences()
\r
80 * @param i DOCUMENT ME!
\r
82 * @return DOCUMENT ME!
\r
84 public SequenceI getSequenceAt(int i)
\r
86 if (i < sequences.size())
\r
88 return (SequenceI) sequences.elementAt(i);
\r
94 /** Adds a sequence to the alignment. Recalculates maxLength and size.
\r
98 public void addSequence(SequenceI snew)
\r
100 sequences.addElement(snew);
\r
106 * @param seq DOCUMENT ME!
\r
108 public void addSequence(SequenceI[] seq)
\r
110 for (int i = 0; i < seq.length; i++)
\r
112 addSequence(seq[i]);
\r
116 /** Adds a sequence to the alignment. Recalculates maxLength and size.
\r
120 public void setSequenceAt(int i, SequenceI snew)
\r
122 SequenceI oldseq = getSequenceAt(i);
\r
123 deleteSequence(oldseq);
\r
125 sequences.setElementAt(snew, i);
\r
131 * @return DOCUMENT ME!
\r
133 public Vector getGroups()
\r
138 /** Takes out columns consisting entirely of gaps (-,.," ")
\r
140 public void removeGaps()
\r
143 int iSize = getWidth();
\r
145 for (int i = 0; i < iSize; i++)
\r
147 boolean delete = true;
\r
149 for (int j = 0; j < getHeight(); j++)
\r
151 current = getSequenceAt(j);
\r
153 if (current.getLength() > i)
\r
155 /* MC Should move this to a method somewhere */
\r
156 if (!jalview.util.Comparison.isGap(current.getCharAt(i)))
\r
165 deleteColumns(i, i);
\r
172 /** Removes a range of columns (start to end inclusive).
\r
174 * @param start Start column in the alignment
\r
175 * @param end End column in the alignment
\r
177 public void deleteColumns(int start, int end)
\r
179 deleteColumns(0, getHeight() - 1, start, end);
\r
185 * @param seq1 DOCUMENT ME!
\r
186 * @param seq2 DOCUMENT ME!
\r
187 * @param start DOCUMENT ME!
\r
188 * @param end DOCUMENT ME!
\r
190 public void deleteColumns(int seq1, int seq2, int start, int end)
\r
192 for (int i = 0; i <= (end - start); i++)
\r
194 for (int j = seq1; j <= seq2; j++)
\r
196 getSequenceAt(j).deleteCharAt(start);
\r
204 * @param i DOCUMENT ME!
\r
206 public void trimLeft(int i)
\r
208 int j, jSize = getHeight();
\r
209 for (j = 0; j < jSize; j++)
\r
211 SequenceI s = getSequenceAt(j);
\r
212 int newstart = s.findPosition(i);
\r
214 if(i>s.getLength())
\r
216 sequences.removeElement(s);
\r
222 s.setStart(newstart);
\r
223 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 if(s.getLength()>i)
\r
242 s.setSequence(s.getSequence().substring(0, i + 1));
\r
249 * @param s DOCUMENT ME!
\r
251 public void deleteSequence(SequenceI s)
\r
253 for (int i = 0; i < getHeight(); i++)
\r
255 if (getSequenceAt(i) == s)
\r
265 * @param i DOCUMENT ME!
\r
267 public void deleteSequence(int i)
\r
269 sequences.removeElementAt(i);
\r
275 * @param threshold DOCUMENT ME!
\r
276 * @param sel DOCUMENT ME!
\r
278 * @return DOCUMENT ME!
\r
280 public Vector removeRedundancy(float threshold, Vector sel)
\r
282 Vector del = new Vector();
\r
284 for (int i = 1; i < sel.size(); i++)
\r
286 for (int j = 0; j < i; j++)
\r
288 // Only do the comparison if either have not been deleted
\r
289 if (!del.contains((SequenceI) sel.elementAt(i)) ||
\r
290 !del.contains((SequenceI) sel.elementAt(j)))
\r
292 // use PID instead of Comparison (which is really not pleasant)
\r
293 float pid = Comparison.PID((SequenceI) sel.elementAt(j),
\r
294 (SequenceI) sel.elementAt(i));
\r
296 if (pid >= threshold)
\r
298 // Delete the shortest one
\r
299 if (((SequenceI) sel.elementAt(j)).getSequence().length() > ((SequenceI) sel
\r
301 i)).getSequence().length())
\r
303 del.addElement(sel.elementAt(i));
\r
307 del.addElement(sel.elementAt(i));
\r
314 // Now delete the sequences
\r
315 for (int i = 0; i < del.size(); i++)
\r
317 deleteSequence((SequenceI) del.elementAt(i));
\r
324 public SequenceGroup findGroup(int i)
\r
326 return findGroup(getSequenceAt(i));
\r
330 public SequenceGroup findGroup(SequenceI s)
\r
332 for (int i = 0; i < this.groups.size(); i++)
\r
334 SequenceGroup sg = (SequenceGroup) groups.elementAt(i);
\r
336 if (sg.sequences.contains(s))
\r
348 * @param s DOCUMENT ME!
\r
350 * @return DOCUMENT ME!
\r
352 public SequenceGroup[] findAllGroups(SequenceI s)
\r
354 Vector temp = new Vector();
\r
356 int gSize = groups.size();
\r
357 for (int i = 0; i < gSize; i++)
\r
359 SequenceGroup sg = (SequenceGroup) groups.elementAt(i);
\r
360 if(sg==null || sg.sequences==null)
\r
362 this.deleteGroup(sg);
\r
367 if (sg.sequences.contains(s))
\r
369 temp.addElement(sg);
\r
373 SequenceGroup[] ret = new SequenceGroup[temp.size()];
\r
375 for (int i = 0; i < temp.size(); i++)
\r
377 ret[i] = (SequenceGroup) temp.elementAt(i);
\r
386 * @param sg DOCUMENT ME!
\r
388 public void addSuperGroup(SuperGroup sg)
\r
390 superGroup.addElement(sg);
\r
396 * @param sg DOCUMENT ME!
\r
398 public void removeSuperGroup(SuperGroup sg)
\r
400 superGroup.removeElement(sg);
\r
406 * @param sg DOCUMENT ME!
\r
408 * @return DOCUMENT ME!
\r
410 public SuperGroup getSuperGroup(SequenceGroup sg)
\r
412 for (int i = 0; i < this.superGroup.size(); i++)
\r
414 SuperGroup temp = (SuperGroup) superGroup.elementAt(i);
\r
416 if (temp.sequenceGroups.contains(sg))
\r
426 public void addGroup(SequenceGroup sg)
\r
428 if (!groups.contains(sg))
\r
430 groups.addElement(sg);
\r
437 public void deleteAllGroups()
\r
439 groups.removeAllElements();
\r
440 superGroup.removeAllElements();
\r
444 while (i < sequences.size())
\r
446 SequenceI s = getSequenceAt(i);
\r
447 s.setColor(java.awt.Color.white);
\r
453 public void deleteGroup(SequenceGroup g)
\r
455 if (groups.contains(g))
\r
457 groups.removeElement(g);
\r
462 public SequenceI findName(String name)
\r
466 while (i < sequences.size())
\r
468 if (getSequenceAt(i).getName().equals(name))
\r
470 return getSequenceAt(i);
\r
481 public int findIndex(SequenceI s)
\r
485 while (i < sequences.size())
\r
487 if (s == getSequenceAt(i))
\r
501 * @return DOCUMENT ME!
\r
503 public int getHeight()
\r
505 return sequences.size();
\r
511 * @return DOCUMENT ME!
\r
513 public int getWidth()
\r
515 int maxLength = -1;
\r
517 for (int i = 0; i < sequences.size(); i++)
\r
519 if (getSequenceAt(i).getLength() > maxLength)
\r
521 maxLength = getSequenceAt(i).getLength();
\r
531 * @return DOCUMENT ME!
\r
533 public int getMaxIdLength()
\r
538 while (i < sequences.size())
\r
540 SequenceI seq = getSequenceAt(i);
\r
541 String tmp = seq.getName() + "/" + seq.getStart() + "-" +
\r
544 if (tmp.length() > max)
\r
546 max = tmp.length();
\r
558 * @param gc DOCUMENT ME!
\r
560 public void setGapCharacter(char gc)
\r
564 for (int i = 0; i < sequences.size(); i++)
\r
566 Sequence seq = (Sequence) sequences.elementAt(i);
\r
567 seq.sequence = seq.sequence.replace('.', gc);
\r
568 seq.sequence = seq.sequence.replace('-', gc);
\r
569 seq.sequence = seq.sequence.replace(' ', gc);
\r
576 * @return DOCUMENT ME!
\r
578 public char getGapCharacter()
\r
580 return gapCharacter;
\r
586 * @return DOCUMENT ME!
\r
588 public Vector getAAFrequency()
\r
590 return AAFrequency.calculate(sequences, 0, getWidth());
\r
596 * @return DOCUMENT ME!
\r
598 public boolean isAligned()
\r
600 int width = getWidth();
\r
602 for (int i = 0; i < sequences.size(); i++)
\r
604 if (getSequenceAt(i).getLength() != width)
\r
616 * @param aa DOCUMENT ME!
\r
618 public void deleteAnnotation(AlignmentAnnotation aa)
\r
622 if (annotations != null)
\r
624 aSize = annotations.length;
\r
627 AlignmentAnnotation[] temp = new AlignmentAnnotation[aSize - 1];
\r
631 for (int i = 0; i < aSize; i++)
\r
633 if (annotations[i] == aa)
\r
638 temp[tIndex] = annotations[i];
\r
642 annotations = temp;
\r
648 * @param aa DOCUMENT ME!
\r
650 public void addAnnotation(AlignmentAnnotation aa)
\r
654 if (annotations != null)
\r
656 aSize = annotations.length + 1;
\r
660 AlignmentAnnotation[] temp = new AlignmentAnnotation[aSize];
\r
665 for (i = 0; i < (aSize - 1); i++)
\r
667 temp[i] = annotations[i];
\r
673 annotations = temp;
\r
679 * @return DOCUMENT ME!
\r
681 public AlignmentAnnotation[] getAlignmentAnnotation()
\r
683 return annotations;
\r
686 public void setNucleotide(boolean b)
\r
694 public boolean isNucleotide()
\r
696 if(type==NUCLEOTIDE)
\r
702 public void setDataset(Alignment data)
\r
704 if(dataset==null && data==null)
\r
706 // Create a new dataset for this alignment.
\r
707 // Can only be done once, if dataset is not null
\r
708 // This will not be performed
\r
709 Sequence[] seqs = new Sequence[getHeight()];
\r
710 for (int i = 0; i < getHeight(); i++)
\r
713 seqs[i] = new Sequence(getSequenceAt(i).getName(),
\r
714 AlignSeq.extractGaps(
\r
715 jalview.util.Comparison.GapChars,
\r
716 getSequenceAt(i).getSequence()
\r
718 getSequenceAt(i).getStart(),
\r
719 getSequenceAt(i).getEnd());
\r
721 getSequenceAt(i).setDatasetSequence(seqs[i]);
\r
724 dataset = new Alignment(seqs);
\r
726 else if(dataset==null && data!=null)
\r
732 public Alignment getDataset()
\r