1 package jalview.datamodel;
\r
3 import jalview.analysis.*;
\r
4 import jalview.util.*;
\r
7 /** Data structure to hold and manipulate a multiple sequence alignment
\r
9 public class Alignment implements AlignmentI
\r
12 protected Vector sequences;
\r
13 protected Vector groups = new Vector();
\r
14 protected ArrayList superGroup = new ArrayList();
\r
15 protected char gapCharacter = '-';
\r
17 /** Make an alignment from an array of Sequences.
\r
21 public Alignment(SequenceI[] seqs) {
\r
22 sequences = new Vector();
\r
24 for (int i=0; i < seqs.length; i++)
\r
25 sequences.addElement(seqs[i]);
\r
30 public Vector getSequences() {
\r
34 public SequenceI getSequenceAt(int i) {
\r
35 if (i < sequences.size()) {
\r
36 return (SequenceI)sequences.elementAt(i);
\r
42 /** Adds a sequence to the alignment. Recalculates maxLength and size.
\r
46 public void addSequence(SequenceI snew) {
\r
47 sequences.addElement(snew);
\r
50 public void addSequence(SequenceI[] seq) {
\r
51 for (int i=0; i < seq.length; i++) {
\r
52 addSequence(seq[i]);
\r
56 /** Adds a sequence to the alignment. Recalculates maxLength and size.
\r
60 public void setSequenceAt(int i,SequenceI snew) {
\r
61 SequenceI oldseq = getSequenceAt(i);
\r
62 deleteSequence(oldseq);
\r
64 sequences.setElementAt(snew,i);
\r
67 public Vector getGroups() {
\r
71 /** Sorts the sequences by sequence group size - largest to smallest.
\r
74 public void sortGroups() {
\r
75 float[] arr = new float [groups.size()];
\r
76 Object[] s = new Object[groups.size()];
\r
78 for (int i=0; i < groups.size(); i++) {
\r
79 arr[i] = ((SequenceGroup)groups.elementAt(i)).sequences.size();
\r
80 s[i] = groups.elementAt(i);
\r
83 QuickSort.sort(arr,s);
\r
85 Vector newg = new Vector(groups.size());
\r
87 for (int i=groups.size()-1; i >= 0; i--) {
\r
88 newg.addElement(s[i]);
\r
94 /** Takes out columns consisting entirely of gaps (-,.," ")
\r
96 public void removeGaps()
\r
100 int iSize = getWidth();
\r
101 for (int i=0; i < iSize; i++)
\r
103 boolean delete = true;
\r
104 for (int j=0; j < getHeight(); j++)
\r
106 current = getSequenceAt(j);
\r
107 if (current.getLength() > i)
\r
109 /* MC Should move this to a method somewhere */
\r
110 if ( !jalview.util.Comparison.isGap(current.getCharAt(i)))
\r
118 deleteColumns(i,i);
\r
127 /** Returns an array of Sequences containing columns
\r
128 * start to end (inclusive) only.
\r
130 * @param start start column to fetch
\r
131 * @param end end column to fetch
\r
132 * @return Array of Sequences, ready to put into a new Alignment
\r
134 public SequenceI[] getColumns(int start, int end) {
\r
135 return getColumns(0,getHeight()-1,start,end);
\r
138 /** Removes a range of columns (start to end inclusive).
\r
140 * @param start Start column in the alignment
\r
141 * @param end End column in the alignment
\r
143 public void deleteColumns(int start, int end) {
\r
144 deleteColumns(0,getHeight()-1,start,end);
\r
147 public void deleteColumns(int seq1, int seq2, int start, int end) {
\r
149 for (int i=0; i <= (end-start); i++) {
\r
150 for (int j=seq1; j <= seq2; j++) {
\r
151 getSequenceAt(j).deleteCharAt(start);
\r
156 public void insertColumns(SequenceI[] seqs, int pos) {
\r
157 if (seqs.length == getHeight()) {
\r
158 for (int i=0; i < getHeight();i++) {
\r
159 String tmp = new String(getSequenceAt(i).getSequence());
\r
160 getSequenceAt(i).setSequence(tmp.substring(0,pos) + seqs[i].getSequence() + tmp.substring(pos));
\r
166 public SequenceI[] getColumns(int seq1, int seq2, int start, int end) {
\r
167 SequenceI[] seqs = new Sequence[(seq2-seq1)+1];
\r
168 for (int i=seq1; i<= seq2; i++ ) {
\r
169 seqs[i] = new Sequence(getSequenceAt(i).getName(),
\r
170 getSequenceAt(i).getSequence().substring(start,end),
\r
171 getSequenceAt(i).findPosition(start),
\r
172 getSequenceAt(i).findPosition(end));
\r
177 public void trimLeft(int i) {
\r
178 for (int j = 0;j< getHeight();j++) {
\r
180 SequenceI s = getSequenceAt(j);
\r
181 int newstart = s.findPosition(i);
\r
183 s.setStart(newstart);
\r
184 s.setSequence(s.getSequence().substring(i));
\r
189 public void trimRight(int i) {
\r
190 for (int j = 0;j< getHeight();j++) {
\r
191 SequenceI s = getSequenceAt(j);
\r
192 int newend = s.findPosition(i);
\r
195 s.setSequence(s.getSequence().substring(0,i+1));
\r
199 public void deleteSequence(SequenceI s)
\r
201 for (int i=0; i < getHeight(); i++)
\r
202 if (getSequenceAt(i) == s)
\r
206 public void deleteSequence(int i)
\r
208 sequences.removeElementAt(i);
\r
212 public Vector removeRedundancy(float threshold, Vector sel) {
\r
213 Vector del = new Vector();
\r
215 for (int i = 1; i < sel.size(); i++)
\r
217 for (int j = 0; j < i; j++)
\r
219 // Only do the comparison if either have not been deleted
\r
220 if (!del.contains( (SequenceI) sel.elementAt(i)) ||
\r
221 !del.contains( (SequenceI) sel.elementAt(j)))
\r
224 float pid = Comparison.compare( (SequenceI) sel.elementAt(j),
\r
225 (SequenceI) sel.elementAt(i));
\r
227 if (pid >= threshold)
\r
229 // Delete the shortest one
\r
230 if ( ( (SequenceI) sel.elementAt(j)).getSequence().length() >
\r
231 ( (SequenceI) sel.elementAt(i)).getSequence().length())
\r
232 del.addElement(sel.elementAt(i));
\r
234 del.addElement(sel.elementAt(i));
\r
240 // Now delete the sequences
\r
241 for (int i=0; i < del.size(); i++)
\r
242 deleteSequence((SequenceI)del.elementAt(i));
\r
247 public void sortByPID(SequenceI s) {
\r
249 float scores[] = new float[getHeight()];
\r
250 SequenceI seqs[] = new SequenceI[getHeight()];
\r
252 for (int i = 0; i < getHeight(); i++) {
\r
253 scores[i] = Comparison.compare(getSequenceAt(i),s);
\r
254 seqs[i] = getSequenceAt(i);
\r
257 QuickSort.sort(scores,0,scores.length-1,seqs);
\r
261 if (getHeight()%2 == 0) {
\r
262 len = getHeight()/2;
\r
264 len = (getHeight()+1)/2;
\r
267 for (int i = 0; i < len; i++) {
\r
268 SequenceI tmp = seqs[i];
\r
269 sequences.setElementAt(seqs[getHeight()-i-1],i);
\r
270 sequences.setElementAt(tmp,getHeight()-i-1);
\r
274 public void sortByID() {
\r
275 String ids[] = new String[getHeight()];
\r
276 SequenceI seqs[] = new SequenceI[getHeight()];
\r
278 for (int i = 0; i < getHeight(); i++) {
\r
279 ids[i] = getSequenceAt(i).getName();
\r
280 seqs[i] = getSequenceAt(i);
\r
283 QuickSort.sort(ids,seqs);
\r
287 if (getHeight()%2 == 0) {
\r
288 len = getHeight()/2;
\r
290 len = (getHeight()+1)/2;
\r
291 System.out.println("Sort len is odd = " + len);
\r
293 for (int i = 0; i < len; i++) {
\r
294 System.out.println("Swapping " + seqs[i].getName() + " and " + seqs[getHeight()-i-1].getName());
\r
295 SequenceI tmp = seqs[i];
\r
296 sequences.setElementAt(seqs[getHeight()-i-1],i);
\r
297 sequences.setElementAt(tmp,getHeight()-i-1);
\r
302 public SequenceGroup findGroup(int i) {
\r
303 return findGroup(getSequenceAt(i));
\r
307 public SequenceGroup findGroup(SequenceI s) {
\r
308 for (int i = 0; i < this.groups.size();i++)
\r
310 SequenceGroup sg = (SequenceGroup)groups.elementAt(i);
\r
311 if (sg.sequences.contains(s))
\r
318 public SequenceGroup [] findAllGroups(SequenceI s)
\r
321 ArrayList temp = new ArrayList();
\r
323 for (int i = 0; i < this.groups.size();i++)
\r
325 SequenceGroup sg = (SequenceGroup)groups.elementAt(i);
\r
327 if (sg.sequences.contains(s))
\r
331 SequenceGroup [] ret = new SequenceGroup[temp.size()];
\r
332 temp.toArray( ret );
\r
338 public void addToGroup(SequenceGroup g, SequenceI s) {
\r
339 if (!(g.sequences.contains(s))) {
\r
340 g.sequences.addElement(s);
\r
344 public void removeFromGroup(SequenceGroup g,SequenceI s) {
\r
345 if (g != null && g.sequences != null) {
\r
346 if (g.sequences.contains(s)) {
\r
347 g.sequences.removeElement(s);
\r
348 if (g.sequences.size() == 0) {
\r
349 groups.removeElement(g);
\r
355 public void addSuperGroup(SuperGroup sg)
\r
357 superGroup.add(sg);
\r
360 public void removeSuperGroup(SuperGroup sg)
\r
362 superGroup.remove(sg);
\r
365 public SuperGroup getSuperGroup(SequenceGroup sg)
\r
367 for (int i = 0; i < this.superGroup.size(); i++)
\r
369 SuperGroup temp = (SuperGroup) superGroup.get(i);
\r
370 if (temp.sequenceGroups.contains(sg))
\r
377 public void addGroup(SequenceGroup sg) {
\r
378 if(!groups.contains(sg))
\r
379 groups.addElement(sg);
\r
382 public void deleteAllGroups()
\r
385 superGroup.clear();
\r
387 while (i < sequences.size()) {
\r
388 SequenceI s = getSequenceAt(i);
\r
389 s.setColor(java.awt.Color.white);
\r
397 public void deleteGroup(SequenceGroup g) {
\r
398 if (groups.contains(g)) {
\r
399 groups.removeElement(g);
\r
404 public SequenceI findName(String name) {
\r
406 while (i < sequences.size()) {
\r
407 SequenceI s = getSequenceAt(i);
\r
408 if (s.getName().equals(name))
\r
417 public SequenceI findbyDisplayId(String name) {
\r
419 while (i < sequences.size()) {
\r
420 SequenceI s = getSequenceAt(i);
\r
421 if (s.getDisplayId().equals(name))
\r
430 public int findIndex(SequenceI s)
\r
433 while (i < sequences.size())
\r
435 if (s == getSequenceAt(i))
\r
443 public int getHeight() {
\r
444 return sequences.size();
\r
448 public int getWidth()
\r
450 int maxLength = -1;
\r
451 for (int i = 0; i < sequences.size(); i++)
\r
453 if (getSequenceAt(i).getLength() > maxLength)
\r
454 maxLength = getSequenceAt(i).getLength();
\r
461 public int getMaxIdLength() {
\r
465 while (i < sequences.size()) {
\r
466 SequenceI seq = getSequenceAt(i);
\r
467 String tmp = seq.getName() + "/" + seq.getStart() + "-" + seq.getEnd();
\r
469 if (tmp.length() > max) {
\r
470 max = tmp.length();
\r
478 public void setGapCharacter(char gc)
\r
480 char old = getGapCharacter();
\r
482 for (int i=0; i < sequences.size(); i++)
\r
484 Sequence seq = (Sequence)sequences.elementAt(i);
\r
485 seq.sequence = seq.sequence.replace(old, gc);
\r
489 public char getGapCharacter() {
\r
490 return gapCharacter;
\r
493 public Vector getAAFrequency()
\r
495 return AAFrequency.calculate(sequences, 0, getWidth());
\r
498 public boolean isAligned()
\r
500 int width = getWidth();
\r
501 for (int i = 0; i < sequences.size(); i++)
\r
502 if (getSequenceAt(i).getLength() != width)
\r