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 Provenance provenance;
\r
38 protected int type = NUCLEOTIDE;
\r
39 public static final int PROTEIN = 0;
\r
40 public static final int NUCLEOTIDE = 1;
\r
42 /** DOCUMENT ME!! */
\r
43 public AlignmentAnnotation[] annotations;
\r
45 /** DOCUMENT ME!! */
\r
46 public boolean featuresAdded = false;
\r
48 /** Make an alignment from an array of Sequences.
\r
52 public Alignment(SequenceI[] seqs)
\r
56 if( jalview.util.Comparison.isNucleotide(seqs))
\r
61 sequences = new Vector();
\r
63 for (i = 0; i < seqs.length; i++)
\r
65 sequences.addElement(seqs[i]);
\r
74 * @return DOCUMENT ME!
\r
76 public Vector getSequences()
\r
84 * @param i DOCUMENT ME!
\r
86 * @return DOCUMENT ME!
\r
88 public SequenceI getSequenceAt(int i)
\r
90 if (i < sequences.size())
\r
92 return (SequenceI) sequences.elementAt(i);
\r
98 /** Adds a sequence to the alignment. Recalculates maxLength and size.
\r
102 public void addSequence(SequenceI snew)
\r
104 sequences.addElement(snew);
\r
110 * @param seq DOCUMENT ME!
\r
112 public void addSequence(SequenceI[] seq)
\r
114 for (int i = 0; i < seq.length; i++)
\r
116 addSequence(seq[i]);
\r
120 /** Adds a sequence to the alignment. Recalculates maxLength and size.
\r
124 public void setSequenceAt(int i, SequenceI snew)
\r
126 SequenceI oldseq = getSequenceAt(i);
\r
127 deleteSequence(oldseq);
\r
129 sequences.setElementAt(snew, i);
\r
135 * @return DOCUMENT ME!
\r
137 public Vector getGroups()
\r
142 /** Takes out columns consisting entirely of gaps (-,.," ")
\r
144 public void removeGaps()
\r
147 int iSize = getWidth();
\r
149 for (int i = 0; i < iSize; i++)
\r
151 boolean delete = true;
\r
153 for (int j = 0; j < getHeight(); j++)
\r
155 current = getSequenceAt(j);
\r
157 if (current.getLength() > i)
\r
159 /* MC Should move this to a method somewhere */
\r
160 if (!jalview.util.Comparison.isGap(current.getCharAt(i)))
\r
169 deleteColumns(i, i);
\r
176 /** Removes a range of columns (start to end inclusive).
\r
178 * @param start Start column in the alignment
\r
179 * @param end End column in the alignment
\r
181 public void deleteColumns(int start, int end)
\r
183 deleteColumns(0, getHeight() - 1, start, end);
\r
189 * @param seq1 DOCUMENT ME!
\r
190 * @param seq2 DOCUMENT ME!
\r
191 * @param start DOCUMENT ME!
\r
192 * @param end DOCUMENT ME!
\r
194 public void deleteColumns(int seq1, int seq2, int start, int end)
\r
196 for (int i = 0; i <= (end - start); i++)
\r
198 for (int j = seq1; j <= seq2; j++)
\r
200 getSequenceAt(j).deleteCharAt(start);
\r
208 * @param i DOCUMENT ME!
\r
210 public void trimLeft(int i)
\r
212 for (int j = 0; j < getHeight(); j++)
\r
214 SequenceI s = getSequenceAt(j);
\r
215 int newstart = s.findPosition(i);
\r
217 s.setStart(newstart);
\r
218 s.setSequence(s.getSequence().substring(i));
\r
225 * @param i DOCUMENT ME!
\r
227 public void trimRight(int i)
\r
229 for (int j = 0; j < getHeight(); j++)
\r
231 SequenceI s = getSequenceAt(j);
\r
232 int newend = s.findPosition(i);
\r
235 s.setSequence(s.getSequence().substring(0, i + 1));
\r
242 * @param s DOCUMENT ME!
\r
244 public void deleteSequence(SequenceI s)
\r
246 for (int i = 0; i < getHeight(); i++)
\r
248 if (getSequenceAt(i) == s)
\r
258 * @param i DOCUMENT ME!
\r
260 public void deleteSequence(int i)
\r
262 sequences.removeElementAt(i);
\r
268 * @param threshold DOCUMENT ME!
\r
269 * @param sel DOCUMENT ME!
\r
271 * @return DOCUMENT ME!
\r
273 public Vector removeRedundancy(float threshold, Vector sel)
\r
275 Vector del = new Vector();
\r
277 for (int i = 1; i < sel.size(); i++)
\r
279 for (int j = 0; j < i; j++)
\r
281 // Only do the comparison if either have not been deleted
\r
282 if (!del.contains((SequenceI) sel.elementAt(i)) ||
\r
283 !del.contains((SequenceI) sel.elementAt(j)))
\r
285 // use PID instead of Comparison (which is really not pleasant)
\r
286 float pid = Comparison.PID((SequenceI) sel.elementAt(j),
\r
287 (SequenceI) sel.elementAt(i));
\r
289 if (pid >= threshold)
\r
291 // Delete the shortest one
\r
292 if (((SequenceI) sel.elementAt(j)).getSequence().length() > ((SequenceI) sel
\r
294 i)).getSequence().length())
\r
296 del.addElement(sel.elementAt(i));
\r
300 del.addElement(sel.elementAt(i));
\r
307 // Now delete the sequences
\r
308 for (int i = 0; i < del.size(); i++)
\r
310 deleteSequence((SequenceI) del.elementAt(i));
\r
317 public SequenceGroup findGroup(int i)
\r
319 return findGroup(getSequenceAt(i));
\r
323 public SequenceGroup findGroup(SequenceI s)
\r
325 for (int i = 0; i < this.groups.size(); i++)
\r
327 SequenceGroup sg = (SequenceGroup) groups.elementAt(i);
\r
329 if (sg.sequences.contains(s))
\r
341 * @param s DOCUMENT ME!
\r
343 * @return DOCUMENT ME!
\r
345 public SequenceGroup[] findAllGroups(SequenceI s)
\r
347 Vector temp = new Vector();
\r
349 int gSize = groups.size();
\r
350 for (int i = 0; i < gSize; i++)
\r
352 SequenceGroup sg = (SequenceGroup) groups.elementAt(i);
\r
353 if(sg==null || sg.sequences==null)
\r
355 this.deleteGroup(sg);
\r
360 if (sg.sequences.contains(s))
\r
362 temp.addElement(sg);
\r
366 SequenceGroup[] ret = new SequenceGroup[temp.size()];
\r
368 for (int i = 0; i < temp.size(); i++)
\r
370 ret[i] = (SequenceGroup) temp.elementAt(i);
\r
379 * @param sg DOCUMENT ME!
\r
381 public void addSuperGroup(SuperGroup sg)
\r
383 superGroup.addElement(sg);
\r
389 * @param sg DOCUMENT ME!
\r
391 public void removeSuperGroup(SuperGroup sg)
\r
393 superGroup.removeElement(sg);
\r
399 * @param sg DOCUMENT ME!
\r
401 * @return DOCUMENT ME!
\r
403 public SuperGroup getSuperGroup(SequenceGroup sg)
\r
405 for (int i = 0; i < this.superGroup.size(); i++)
\r
407 SuperGroup temp = (SuperGroup) superGroup.elementAt(i);
\r
409 if (temp.sequenceGroups.contains(sg))
\r
419 public void addGroup(SequenceGroup sg)
\r
421 if (!groups.contains(sg))
\r
423 groups.addElement(sg);
\r
430 public void deleteAllGroups()
\r
432 groups.removeAllElements();
\r
433 superGroup.removeAllElements();
\r
437 while (i < sequences.size())
\r
439 SequenceI s = getSequenceAt(i);
\r
440 s.setColor(java.awt.Color.white);
\r
446 public void deleteGroup(SequenceGroup g)
\r
448 if (groups.contains(g))
\r
450 groups.removeElement(g);
\r
454 public SequenceI findShortName(String name)
\r
458 while (i < sequences.size())
\r
460 if (getSequenceAt(i).getShortName().equals(name))
\r
462 return getSequenceAt(i);
\r
472 public SequenceI findName(String name)
\r
476 while (i < sequences.size())
\r
478 if (getSequenceAt(i).getName().equals(name))
\r
480 return getSequenceAt(i);
\r
491 public int findIndex(SequenceI s)
\r
495 while (i < sequences.size())
\r
497 if (s == getSequenceAt(i))
\r
511 * @return DOCUMENT ME!
\r
513 public int getHeight()
\r
515 return sequences.size();
\r
521 * @return DOCUMENT ME!
\r
523 public int getWidth()
\r
525 int maxLength = -1;
\r
527 for (int i = 0; i < sequences.size(); i++)
\r
529 if (getSequenceAt(i).getLength() > maxLength)
\r
531 maxLength = getSequenceAt(i).getLength();
\r
541 * @return DOCUMENT ME!
\r
543 public int getMaxIdLength()
\r
548 while (i < sequences.size())
\r
550 SequenceI seq = getSequenceAt(i);
\r
551 String tmp = seq.getName() + "/" + seq.getStart() + "-" +
\r
554 if (tmp.length() > max)
\r
556 max = tmp.length();
\r
568 * @param gc DOCUMENT ME!
\r
570 public void setGapCharacter(char gc)
\r
574 for (int i = 0; i < sequences.size(); i++)
\r
576 Sequence seq = (Sequence) sequences.elementAt(i);
\r
577 seq.sequence = seq.sequence.replace('.', gc);
\r
578 seq.sequence = seq.sequence.replace('-', gc);
\r
579 seq.sequence = seq.sequence.replace(' ', gc);
\r
586 * @return DOCUMENT ME!
\r
588 public char getGapCharacter()
\r
590 return gapCharacter;
\r
596 * @return DOCUMENT ME!
\r
598 public Vector getAAFrequency()
\r
600 return AAFrequency.calculate(sequences, 0, getWidth());
\r
606 * @return DOCUMENT ME!
\r
608 public boolean isAligned()
\r
610 int width = getWidth();
\r
612 for (int i = 0; i < sequences.size(); i++)
\r
614 if (getSequenceAt(i).getLength() != width)
\r
626 * @param aa DOCUMENT ME!
\r
628 public void deleteAnnotation(AlignmentAnnotation aa)
\r
632 if (annotations != null)
\r
634 aSize = annotations.length;
\r
637 AlignmentAnnotation[] temp = new AlignmentAnnotation[aSize - 1];
\r
641 for (int i = 0; i < aSize; i++)
\r
643 if (annotations[i] == aa)
\r
648 temp[tIndex] = annotations[i];
\r
652 annotations = temp;
\r
658 * @param aa DOCUMENT ME!
\r
660 public void addAnnotation(AlignmentAnnotation aa)
\r
664 if (annotations != null)
\r
666 aSize = annotations.length + 1;
\r
670 AlignmentAnnotation[] temp = new AlignmentAnnotation[aSize];
\r
675 for (i = 0; i < (aSize - 1); i++)
\r
677 temp[i] = annotations[i];
\r
683 annotations = temp;
\r
689 * @return DOCUMENT ME!
\r
691 public AlignmentAnnotation[] getAlignmentAnnotation()
\r
693 return annotations;
\r
696 public void setNucleotide(boolean b)
\r
704 public boolean isNucleotide()
\r
706 if(type==NUCLEOTIDE)
\r
712 public void setDataset(Alignment data)
\r
714 if(dataset==null && data==null)
\r
716 // Create a new dataset for this alignment.
\r
717 // Can only be done once, if dataset is not null
\r
718 // This will not be performed
\r
719 Sequence[] seqs = new Sequence[getHeight()];
\r
720 for (int i = 0; i < getHeight(); i++)
\r
723 seqs[i] = new Sequence(getSequenceAt(i).getName(),
\r
724 AlignSeq.extractGaps(
\r
725 jalview.util.Comparison.GapChars,
\r
726 getSequenceAt(i).getSequence()
\r
728 getSequenceAt(i).getStart(),
\r
729 getSequenceAt(i).getEnd());
\r
731 getSequenceAt(i).setDatasetSequence(seqs[i]);
\r
734 dataset = new Alignment(seqs);
\r
736 else if(dataset==null && data!=null)
\r
742 public Alignment getDataset()
\r
747 public void setProvenance(Provenance prov)
\r
751 public Provenance getProvenance()
\r