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
76 public SequenceI [] getSequencesArray()
\r
78 SequenceI [] reply = new SequenceI[sequences.size()];
\r
79 for(int i=0; i<sequences.size(); i++)
\r
81 reply[i] = (SequenceI)sequences.elementAt(i);
\r
89 * @param i DOCUMENT ME!
\r
91 * @return DOCUMENT ME!
\r
93 public SequenceI getSequenceAt(int i)
\r
95 if (i < sequences.size())
\r
97 return (SequenceI) sequences.elementAt(i);
\r
103 /** Adds a sequence to the alignment. Recalculates maxLength and size.
\r
107 public void addSequence(SequenceI snew)
\r
111 Sequence ds = new Sequence(snew.getName(),
\r
112 AlignSeq.extractGaps("-. ", snew.getSequence()),
\r
116 snew.setDatasetSequence(ds);
\r
117 getDataset().addSequence(ds);
\r
120 sequences.addElement(snew);
\r
124 /** Adds a sequence to the alignment. Recalculates maxLength and size.
\r
128 public void setSequenceAt(int i, SequenceI snew)
\r
130 SequenceI oldseq = getSequenceAt(i);
\r
131 deleteSequence(oldseq);
\r
133 sequences.setElementAt(snew, i);
\r
139 * @return DOCUMENT ME!
\r
141 public Vector getGroups()
\r
146 /** Takes out columns consisting entirely of gaps (-,.," ")
\r
148 public void removeGaps()
\r
151 int iSize = getWidth();
\r
153 for (int i = 0; i < iSize; i++)
\r
155 boolean delete = true;
\r
157 for (int j = 0; j < getHeight(); j++)
\r
159 current = getSequenceAt(j);
\r
161 if (current.getLength() > i)
\r
163 if (!jalview.util.Comparison.isGap(current.getCharAt(i)))
\r
172 deleteColumns(i, i);
\r
179 /** Removes a range of columns (start to end inclusive).
\r
181 * @param start Start column in the alignment
\r
182 * @param end End column in the alignment
\r
184 public void deleteColumns(int start, int end)
\r
186 deleteColumns(0, getHeight() - 1, start, end);
\r
192 * @param seq1 DOCUMENT ME!
\r
193 * @param seq2 DOCUMENT ME!
\r
194 * @param start DOCUMENT ME!
\r
195 * @param end DOCUMENT ME!
\r
197 public void deleteColumns(int seq1, int seq2, int start, int end)
\r
199 for (int i = 0; i <= (end - start); i++)
\r
201 for (int j = seq1; j <= seq2; j++)
\r
203 getSequenceAt(j).deleteCharAt(start);
\r
211 * @param i DOCUMENT ME!
\r
213 public void trimLeft(int i)
\r
215 int j, jSize = getHeight();
\r
216 for (j = 0; j < jSize; j++)
\r
218 SequenceI s = getSequenceAt(j);
\r
219 int newstart = s.findPosition(i);
\r
221 if(i>s.getLength())
\r
223 sequences.removeElement(s);
\r
229 s.setStart(newstart);
\r
230 s.setSequence(s.getSequence().substring(i));
\r
238 * @param i DOCUMENT ME!
\r
240 public void trimRight(int i)
\r
242 for (int j = 0; j < getHeight(); j++)
\r
244 SequenceI s = getSequenceAt(j);
\r
245 int newend = s.findPosition(i);
\r
248 if(s.getLength()>i)
\r
249 s.setSequence(s.getSequence().substring(0, i + 1));
\r
256 * @param s DOCUMENT ME!
\r
258 public void deleteSequence(SequenceI s)
\r
260 for (int i = 0; i < getHeight(); i++)
\r
262 if (getSequenceAt(i) == s)
\r
272 * @param i DOCUMENT ME!
\r
274 public void deleteSequence(int i)
\r
276 sequences.removeElementAt(i);
\r
281 public SequenceGroup findGroup(SequenceI s)
\r
283 for (int i = 0; i < this.groups.size(); i++)
\r
285 SequenceGroup sg = (SequenceGroup) groups.elementAt(i);
\r
287 if (sg.getSequences(false).contains(s))
\r
299 * @param s DOCUMENT ME!
\r
301 * @return DOCUMENT ME!
\r
303 public SequenceGroup[] findAllGroups(SequenceI s)
\r
305 Vector temp = new Vector();
\r
307 int gSize = groups.size();
\r
308 for (int i = 0; i < gSize; i++)
\r
310 SequenceGroup sg = (SequenceGroup) groups.elementAt(i);
\r
311 if(sg==null || sg.getSequences(false)==null)
\r
313 this.deleteGroup(sg);
\r
318 if (sg.getSequences(false).contains(s))
\r
320 temp.addElement(sg);
\r
324 SequenceGroup[] ret = new SequenceGroup[temp.size()];
\r
326 for (int i = 0; i < temp.size(); i++)
\r
328 ret[i] = (SequenceGroup) temp.elementAt(i);
\r
337 public void addGroup(SequenceGroup sg)
\r
339 if (!groups.contains(sg))
\r
341 groups.addElement(sg);
\r
348 public void deleteAllGroups()
\r
350 groups.removeAllElements();
\r
354 while (i < sequences.size())
\r
356 SequenceI s = getSequenceAt(i);
\r
357 s.setColor(java.awt.Color.white);
\r
363 public void deleteGroup(SequenceGroup g)
\r
365 if (groups.contains(g))
\r
367 groups.removeElement(g);
\r
372 public SequenceI findName(String name)
\r
376 while (i < sequences.size())
\r
378 if (getSequenceAt(i).getName().equals(name))
\r
380 return getSequenceAt(i);
\r
391 public int findIndex(SequenceI s)
\r
395 while (i < sequences.size())
\r
397 if (s == getSequenceAt(i))
\r
411 * @return DOCUMENT ME!
\r
413 public int getHeight()
\r
415 return sequences.size();
\r
421 * @return DOCUMENT ME!
\r
423 public int getWidth()
\r
425 int maxLength = -1;
\r
427 for (int i = 0; i < sequences.size(); i++)
\r
429 if (getSequenceAt(i).getLength() > maxLength)
\r
431 maxLength = getSequenceAt(i).getLength();
\r
441 * @return DOCUMENT ME!
\r
443 public int getMaxIdLength()
\r
448 while (i < sequences.size())
\r
450 SequenceI seq = getSequenceAt(i);
\r
451 String tmp = seq.getName() + "/" + seq.getStart() + "-" +
\r
454 if (tmp.length() > max)
\r
456 max = tmp.length();
\r
468 * @param gc DOCUMENT ME!
\r
470 public void setGapCharacter(char gc)
\r
474 for (int i = 0; i < sequences.size(); i++)
\r
476 Sequence seq = (Sequence) sequences.elementAt(i);
\r
477 seq.setSequence( seq.getSequence().replace('.', gc) );
\r
478 seq.setSequence( seq.getSequence().replace('-', gc) );
\r
479 seq.setSequence( seq.getSequence().replace(' ', gc) );
\r
486 * @return DOCUMENT ME!
\r
488 public char getGapCharacter()
\r
490 return gapCharacter;
\r
496 * @return DOCUMENT ME!
\r
498 public Vector getAAFrequency()
\r
500 return AAFrequency.calculate(sequences, 0, getWidth());
\r
506 * @return DOCUMENT ME!
\r
508 public boolean isAligned()
\r
510 int width = getWidth();
\r
512 for (int i = 0; i < sequences.size(); i++)
\r
514 if (getSequenceAt(i).getLength() != width)
\r
526 * @param aa DOCUMENT ME!
\r
528 public void deleteAnnotation(AlignmentAnnotation aa)
\r
532 if (annotations != null)
\r
534 aSize = annotations.length;
\r
537 AlignmentAnnotation[] temp = new AlignmentAnnotation[aSize - 1];
\r
541 for (int i = 0; i < aSize; i++)
\r
543 if (annotations[i] == aa)
\r
548 temp[tIndex] = annotations[i];
\r
552 annotations = temp;
\r
556 public void adjustSequenceAnnotations()
\r
558 if(annotations!=null)
\r
560 for (int a = 0; a < annotations.length; a++)
\r
562 if (annotations[a].sequenceRef != null)
\r
564 annotations[a].adjustForAlignment();
\r
573 * @param aa DOCUMENT ME!
\r
575 public void addAnnotation(AlignmentAnnotation aa)
\r
578 if (annotations != null)
\r
580 aSize = annotations.length + 1;
\r
583 AlignmentAnnotation[] temp = new AlignmentAnnotation[aSize];
\r
585 temp[aSize-1] = aa;
\r
591 for (i = 0; i < (aSize-1); i++)
\r
593 temp[i] = annotations[i];
\r
597 annotations = temp;
\r
600 public void setAnnotationIndex(AlignmentAnnotation aa, int index)
\r
602 if(aa==null || annotations==null || annotations.length-1<index)
\r
605 int aSize = annotations.length;
\r
606 AlignmentAnnotation[] temp = new AlignmentAnnotation[aSize];
\r
610 for (int i = 0; i < aSize; i++)
\r
616 temp[i] = annotations[i];
\r
618 temp[i] = annotations[i-1];
\r
621 annotations = temp;
\r
627 * @return DOCUMENT ME!
\r
629 public AlignmentAnnotation[] getAlignmentAnnotation()
\r
631 return annotations;
\r
634 public void setNucleotide(boolean b)
\r
642 public boolean isNucleotide()
\r
644 if(type==NUCLEOTIDE)
\r
650 public void setDataset(Alignment data)
\r
652 if(dataset==null && data==null)
\r
654 // Create a new dataset for this alignment.
\r
655 // Can only be done once, if dataset is not null
\r
656 // This will not be performed
\r
657 Sequence[] seqs = new Sequence[getHeight()];
\r
658 for (int i = 0; i < getHeight(); i++)
\r
661 seqs[i] = new Sequence(getSequenceAt(i).getName(),
\r
662 AlignSeq.extractGaps(
\r
663 jalview.util.Comparison.GapChars,
\r
664 getSequenceAt(i).getSequence()
\r
666 getSequenceAt(i).getStart(),
\r
667 getSequenceAt(i).getEnd());
\r
669 getSequenceAt(i).setDatasetSequence(seqs[i]);
\r
672 dataset = new Alignment(seqs);
\r
674 else if(dataset==null && data!=null)
\r
680 public Alignment getDataset()
\r
685 public boolean padGaps() {
\r
686 boolean modified=false;
\r
688 //Remove excess gaps from the end of alignment
\r
689 int maxLength = -1;
\r
692 for (int i = 0; i < sequences.size(); i++)
\r
694 current = getSequenceAt(i);
\r
695 for (int j = current.getLength(); j > maxLength; j--)
\r
697 if (j > maxLength && !jalview.util.Comparison.isGap(
\r
698 current.getCharAt(j)))
\r
708 for (int i = 0; i < sequences.size();
\r
711 current = getSequenceAt(i);
\r
713 if (current.getLength() < maxLength)
\r
715 current.insertCharAt(maxLength - 1, gapCharacter);
\r
718 else if(current.getLength() > maxLength)
\r
720 current.deleteChars(maxLength, current.getLength());
\r
726 public HiddenSequences getHiddenSequences()
\r
728 return hiddenSequences;
\r