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
101 Sequence ds = new Sequence(snew.getName(),
\r
102 AlignSeq.extractGaps("-. ", snew.getSequence()),
\r
106 snew.setDatasetSequence(ds);
\r
107 getDataset().addSequence(ds);
\r
110 sequences.addElement(snew);
\r
114 /** Adds a sequence to the alignment. Recalculates maxLength and size.
\r
118 public void setSequenceAt(int i, SequenceI snew)
\r
120 SequenceI oldseq = getSequenceAt(i);
\r
121 deleteSequence(oldseq);
\r
123 sequences.setElementAt(snew, i);
\r
129 * @return DOCUMENT ME!
\r
131 public Vector getGroups()
\r
136 /** Takes out columns consisting entirely of gaps (-,.," ")
\r
138 public void removeGaps()
\r
141 int iSize = getWidth();
\r
143 for (int i = 0; i < iSize; i++)
\r
145 boolean delete = true;
\r
147 for (int j = 0; j < getHeight(); j++)
\r
149 current = getSequenceAt(j);
\r
151 if (current.getLength() > i)
\r
153 /* MC Should move this to a method somewhere */
\r
154 if (!jalview.util.Comparison.isGap(current.getCharAt(i)))
\r
163 deleteColumns(i, i);
\r
170 /** Removes a range of columns (start to end inclusive).
\r
172 * @param start Start column in the alignment
\r
173 * @param end End column in the alignment
\r
175 public void deleteColumns(int start, int end)
\r
177 deleteColumns(0, getHeight() - 1, start, end);
\r
183 * @param seq1 DOCUMENT ME!
\r
184 * @param seq2 DOCUMENT ME!
\r
185 * @param start DOCUMENT ME!
\r
186 * @param end DOCUMENT ME!
\r
188 public void deleteColumns(int seq1, int seq2, int start, int end)
\r
190 for (int i = 0; i <= (end - start); i++)
\r
192 for (int j = seq1; j <= seq2; j++)
\r
194 getSequenceAt(j).deleteCharAt(start);
\r
202 * @param i DOCUMENT ME!
\r
204 public void trimLeft(int i)
\r
206 int j, jSize = getHeight();
\r
207 for (j = 0; j < jSize; j++)
\r
209 SequenceI s = getSequenceAt(j);
\r
210 int newstart = s.findPosition(i);
\r
212 if(i>s.getLength())
\r
214 sequences.removeElement(s);
\r
220 s.setStart(newstart);
\r
221 s.setSequence(s.getSequence().substring(i));
\r
229 * @param i DOCUMENT ME!
\r
231 public void trimRight(int i)
\r
233 for (int j = 0; j < getHeight(); j++)
\r
235 SequenceI s = getSequenceAt(j);
\r
236 int newend = s.findPosition(i);
\r
239 if(s.getLength()>i)
\r
240 s.setSequence(s.getSequence().substring(0, i + 1));
\r
247 * @param s DOCUMENT ME!
\r
249 public void deleteSequence(SequenceI s)
\r
251 for (int i = 0; i < getHeight(); i++)
\r
253 if (getSequenceAt(i) == s)
\r
263 * @param i DOCUMENT ME!
\r
265 public void deleteSequence(int i)
\r
267 sequences.removeElementAt(i);
\r
273 * @param threshold DOCUMENT ME!
\r
274 * @param sel DOCUMENT ME!
\r
276 * @return DOCUMENT ME!
\r
278 public Vector removeRedundancy(float threshold, Vector sel)
\r
280 Vector del = new Vector();
\r
282 for (int i = 1; i < sel.size(); i++)
\r
284 for (int j = 0; j < i; j++)
\r
286 // Only do the comparison if either have not been deleted
\r
287 if (!del.contains((SequenceI) sel.elementAt(i)) ||
\r
288 !del.contains((SequenceI) sel.elementAt(j)))
\r
290 // use PID instead of Comparison (which is really not pleasant)
\r
291 float pid = Comparison.PID((SequenceI) sel.elementAt(j),
\r
292 (SequenceI) sel.elementAt(i));
\r
294 if (pid >= threshold)
\r
296 // Delete the shortest one
\r
297 if (((SequenceI) sel.elementAt(j)).getSequence().length() > ((SequenceI) sel
\r
299 i)).getSequence().length())
\r
301 del.addElement(sel.elementAt(i));
\r
305 del.addElement(sel.elementAt(i));
\r
312 // Now delete the sequences
\r
313 for (int i = 0; i < del.size(); i++)
\r
315 deleteSequence((SequenceI) del.elementAt(i));
\r
322 public SequenceGroup findGroup(int i)
\r
324 return findGroup(getSequenceAt(i));
\r
328 public SequenceGroup findGroup(SequenceI s)
\r
330 for (int i = 0; i < this.groups.size(); i++)
\r
332 SequenceGroup sg = (SequenceGroup) groups.elementAt(i);
\r
334 if (sg.sequences.contains(s))
\r
346 * @param s DOCUMENT ME!
\r
348 * @return DOCUMENT ME!
\r
350 public SequenceGroup[] findAllGroups(SequenceI s)
\r
352 Vector temp = new Vector();
\r
354 int gSize = groups.size();
\r
355 for (int i = 0; i < gSize; i++)
\r
357 SequenceGroup sg = (SequenceGroup) groups.elementAt(i);
\r
358 if(sg==null || sg.sequences==null)
\r
360 this.deleteGroup(sg);
\r
365 if (sg.sequences.contains(s))
\r
367 temp.addElement(sg);
\r
371 SequenceGroup[] ret = new SequenceGroup[temp.size()];
\r
373 for (int i = 0; i < temp.size(); i++)
\r
375 ret[i] = (SequenceGroup) temp.elementAt(i);
\r
384 public void addGroup(SequenceGroup sg)
\r
386 if (!groups.contains(sg))
\r
388 groups.addElement(sg);
\r
395 public void deleteAllGroups()
\r
397 groups.removeAllElements();
\r
401 while (i < sequences.size())
\r
403 SequenceI s = getSequenceAt(i);
\r
404 s.setColor(java.awt.Color.white);
\r
410 public void deleteGroup(SequenceGroup g)
\r
412 if (groups.contains(g))
\r
414 groups.removeElement(g);
\r
419 public SequenceI findName(String name)
\r
423 while (i < sequences.size())
\r
425 if (getSequenceAt(i).getName().equals(name))
\r
427 return getSequenceAt(i);
\r
438 public int findIndex(SequenceI s)
\r
442 while (i < sequences.size())
\r
444 if (s == getSequenceAt(i))
\r
458 * @return DOCUMENT ME!
\r
460 public int getHeight()
\r
462 return sequences.size();
\r
468 * @return DOCUMENT ME!
\r
470 public int getWidth()
\r
472 int maxLength = -1;
\r
474 for (int i = 0; i < sequences.size(); i++)
\r
476 if (getSequenceAt(i).getLength() > maxLength)
\r
478 maxLength = getSequenceAt(i).getLength();
\r
488 * @return DOCUMENT ME!
\r
490 public int getMaxIdLength()
\r
495 while (i < sequences.size())
\r
497 SequenceI seq = getSequenceAt(i);
\r
498 String tmp = seq.getName() + "/" + seq.getStart() + "-" +
\r
501 if (tmp.length() > max)
\r
503 max = tmp.length();
\r
515 * @param gc DOCUMENT ME!
\r
517 public void setGapCharacter(char gc)
\r
521 for (int i = 0; i < sequences.size(); i++)
\r
523 Sequence seq = (Sequence) sequences.elementAt(i);
\r
524 seq.sequence = seq.sequence.replace('.', gc);
\r
525 seq.sequence = seq.sequence.replace('-', gc);
\r
526 seq.sequence = seq.sequence.replace(' ', gc);
\r
533 * @return DOCUMENT ME!
\r
535 public char getGapCharacter()
\r
537 return gapCharacter;
\r
543 * @return DOCUMENT ME!
\r
545 public Vector getAAFrequency()
\r
547 return AAFrequency.calculate(sequences, 0, getWidth());
\r
553 * @return DOCUMENT ME!
\r
555 public boolean isAligned()
\r
557 int width = getWidth();
\r
559 for (int i = 0; i < sequences.size(); i++)
\r
561 if (getSequenceAt(i).getLength() != width)
\r
573 * @param aa DOCUMENT ME!
\r
575 public void deleteAnnotation(AlignmentAnnotation aa)
\r
579 if (annotations != null)
\r
581 aSize = annotations.length;
\r
584 AlignmentAnnotation[] temp = new AlignmentAnnotation[aSize - 1];
\r
588 for (int i = 0; i < aSize; i++)
\r
590 if (annotations[i] == aa)
\r
595 temp[tIndex] = annotations[i];
\r
599 annotations = temp;
\r
603 public void adjustSequenceAnnotations()
\r
605 if(annotations!=null)
\r
607 for (int a = 0; a < annotations.length; a++)
\r
609 if (annotations[a].sequenceRef != null)
\r
611 annotations[a].adjustForAlignment();
\r
620 * @param aa DOCUMENT ME!
\r
622 public void addAnnotation(AlignmentAnnotation aa)
\r
625 if (annotations != null)
\r
627 aSize = annotations.length + 1;
\r
630 AlignmentAnnotation[] temp = new AlignmentAnnotation[aSize];
\r
632 temp[aSize-1] = aa;
\r
638 for (i = 0; i < (aSize-1); i++)
\r
640 temp[i] = annotations[i];
\r
644 annotations = temp;
\r
647 public void setAnnotationIndex(AlignmentAnnotation aa, int index)
\r
649 if(aa==null || annotations==null || annotations.length-1<index)
\r
652 int aSize = annotations.length;
\r
653 AlignmentAnnotation[] temp = new AlignmentAnnotation[aSize];
\r
657 for (int i = 0; i < aSize; i++)
\r
663 temp[i] = annotations[i];
\r
665 temp[i] = annotations[i-1];
\r
668 annotations = temp;
\r
674 * @return DOCUMENT ME!
\r
676 public AlignmentAnnotation[] getAlignmentAnnotation()
\r
678 return annotations;
\r
681 public void setNucleotide(boolean b)
\r
689 public boolean isNucleotide()
\r
691 if(type==NUCLEOTIDE)
\r
697 public void setDataset(Alignment data)
\r
699 if(dataset==null && data==null)
\r
701 // Create a new dataset for this alignment.
\r
702 // Can only be done once, if dataset is not null
\r
703 // This will not be performed
\r
704 Sequence[] seqs = new Sequence[getHeight()];
\r
705 for (int i = 0; i < getHeight(); i++)
\r
708 seqs[i] = new Sequence(getSequenceAt(i).getName(),
\r
709 AlignSeq.extractGaps(
\r
710 jalview.util.Comparison.GapChars,
\r
711 getSequenceAt(i).getSequence()
\r
713 getSequenceAt(i).getStart(),
\r
714 getSequenceAt(i).getEnd());
\r
716 getSequenceAt(i).setDatasetSequence(seqs[i]);
\r
719 dataset = new Alignment(seqs);
\r
721 else if(dataset==null && data!=null)
\r
727 public Alignment getDataset()
\r
732 public boolean padGaps() {
\r
733 boolean modified=false;
\r
735 //Remove excess gaps from the end of alignment
\r
736 int maxLength = -1;
\r
739 for (int i = 0; i < sequences.size(); i++)
\r
741 current = getSequenceAt(i);
\r
742 for (int j = current.getLength(); j > maxLength; j--)
\r
744 if (j > maxLength && !jalview.util.Comparison.isGap(
\r
745 current.getCharAt(j)))
\r
755 for (int i = 0; i < sequences.size();
\r
758 current = getSequenceAt(i);
\r
760 if (current.getLength() < maxLength)
\r
762 current.insertCharAt(maxLength - 1, gapCharacter);
\r
765 else if(current.getLength() > maxLength)
\r
767 current.deleteChars(maxLength, current.getLength());
\r
773 public HiddenSequences getHiddenSequences()
\r
775 return hiddenSequences;
\r