\r
import java.util.*;\r
\r
-\r
/** Data structure to hold and manipulate a multiple sequence alignment\r
*/\r
public class Alignment implements AlignmentI\r
{\r
+ protected Alignment dataset;\r
protected Vector sequences;\r
protected Vector groups = new Vector();\r
- protected Vector superGroup = new Vector();\r
protected char gapCharacter = '-';\r
+ protected int type = NUCLEOTIDE;\r
+ public static final int PROTEIN = 0;\r
+ public static final int NUCLEOTIDE = 1;\r
\r
/** DOCUMENT ME!! */\r
public AlignmentAnnotation[] annotations;\r
\r
- /** DOCUMENT ME!! */\r
- public boolean featuresAdded = false;\r
+ HiddenSequences hiddenSequences = new HiddenSequences(this);\r
+\r
\r
/** Make an alignment from an array of Sequences.\r
*\r
*/\r
public Alignment(SequenceI[] seqs)\r
{\r
+ int i=0;\r
+\r
+ if( jalview.util.Comparison.isNucleotide(seqs))\r
+ type = NUCLEOTIDE;\r
+ else\r
+ type = PROTEIN;\r
+\r
sequences = new Vector();\r
\r
- for (int i = 0; i < seqs.length; i++)\r
+ for (i = 0; i < seqs.length; i++)\r
{\r
sequences.addElement(seqs[i]);\r
}\r
-\r
- getWidth();\r
}\r
\r
/**\r
return sequences;\r
}\r
\r
+ public SequenceI [] getSequencesArray()\r
+ {\r
+ SequenceI [] reply = new SequenceI[sequences.size()];\r
+ for(int i=0; i<sequences.size(); i++)\r
+ {\r
+ reply[i] = (SequenceI)sequences.elementAt(i);\r
+ }\r
+ return reply;\r
+ }\r
+\r
/**\r
* DOCUMENT ME!\r
*\r
*/\r
public void addSequence(SequenceI snew)\r
{\r
- sequences.addElement(snew);\r
- }\r
-\r
- /**\r
- * DOCUMENT ME!\r
- *\r
- * @param seq DOCUMENT ME!\r
- */\r
- public void addSequence(SequenceI[] seq)\r
- {\r
- for (int i = 0; i < seq.length; i++)\r
+ if(dataset!=null)\r
+ {\r
+ if(snew.getDatasetSequence()!=null)\r
+ {\r
+ System.out.println(snew.getName());\r
+ getDataset().addSequence(snew.getDatasetSequence());\r
+ }\r
+ else\r
{\r
- addSequence(seq[i]);\r
+ Sequence ds = new Sequence(snew.getName(),\r
+ AlignSeq.extractGaps("-. ",\r
+ snew.getSequence()),\r
+ snew.getStart(),\r
+ snew.getEnd());\r
+\r
+ snew.setDatasetSequence(ds);\r
+ getDataset().addSequence(ds);\r
}\r
+ }\r
+\r
+ sequences.addElement(snew);\r
}\r
\r
+\r
/** Adds a sequence to the alignment. Recalculates maxLength and size.\r
*\r
* @param snew\r
*/\r
public void removeGaps()\r
{\r
+ SequenceI[] seqs = getVisibleAndRepresentedSeqs();\r
+ int j, jSize = seqs.length;\r
+\r
SequenceI current;\r
- int iSize = getWidth();\r
+ int width = 0;\r
+ for (int i = 0; i < jSize; i++)\r
+ {\r
+ if (seqs[i].getLength() > width)\r
+ {\r
+ width = seqs[i].getLength();\r
+ }\r
+ }\r
\r
- for (int i = 0; i < iSize; i++)\r
+ int startCol = -1, endCol = -1;\r
+ boolean delete = true;\r
+ for (int i = 0; i < width; i++)\r
{\r
- boolean delete = true;\r
+ delete = true;\r
\r
- for (int j = 0; j < getHeight(); j++)\r
+ for (j = 0; j < jSize; j++)\r
{\r
current = getSequenceAt(j);\r
\r
if (current.getLength() > i)\r
{\r
- /* MC Should move this to a method somewhere */\r
if (!jalview.util.Comparison.isGap(current.getCharAt(i)))\r
{\r
+ if(delete)\r
+ endCol = i;\r
+\r
delete = false;\r
+ break;\r
}\r
}\r
+\r
+ }\r
+\r
+ if(delete && startCol==-1)\r
+ {\r
+ startCol = i;\r
}\r
\r
- if (delete)\r
+\r
+ if (!delete && startCol > -1)\r
{\r
- deleteColumns(i, i);\r
- iSize--;\r
- i--;\r
+ deleteColumns(seqs, startCol, endCol);\r
+ width -= (endCol - startCol);\r
+ i -= (endCol - startCol);\r
+ startCol = -1;\r
+ endCol = -1;\r
}\r
}\r
+\r
+ if (delete && startCol > -1)\r
+ {\r
+ deleteColumns(seqs, startCol, endCol);\r
+ }\r
+\r
}\r
\r
/** Removes a range of columns (start to end inclusive).\r
*\r
+ * @param seqs Sequences to remove columns from\r
* @param start Start column in the alignment\r
* @param end End column in the alignment\r
*/\r
- public void deleteColumns(int start, int end)\r
+ public void deleteColumns(SequenceI [] seqs, int start, int end)\r
{\r
- deleteColumns(0, getHeight() - 1, start, end);\r
+ for(int i=0; i<seqs.length; i++)\r
+ seqs[i].deleteChars(start, end);\r
}\r
\r
- /**\r
- * DOCUMENT ME!\r
- *\r
- * @param seq1 DOCUMENT ME!\r
- * @param seq2 DOCUMENT ME!\r
- * @param start DOCUMENT ME!\r
- * @param end DOCUMENT ME!\r
- */\r
- public void deleteColumns(int seq1, int seq2, int start, int end)\r
- {\r
- for (int i = 0; i <= (end - start); i++)\r
- {\r
- for (int j = seq1; j <= seq2; j++)\r
- {\r
- getSequenceAt(j).deleteCharAt(start);\r
- }\r
- }\r
- }\r
\r
/**\r
* DOCUMENT ME!\r
*/\r
public void trimLeft(int i)\r
{\r
- for (int j = 0; j < getHeight(); j++)\r
+ SequenceI[] seqs = getVisibleAndRepresentedSeqs();\r
+ int j, jSize = seqs.length;\r
+ for (j = 0; j < jSize; j++)\r
{\r
- SequenceI s = getSequenceAt(j);\r
- int newstart = s.findPosition(i);\r
+ int newstart = seqs[j].findPosition(i);\r
\r
- s.setStart(newstart);\r
- s.setSequence(s.getSequence().substring(i));\r
+ if(i>seqs[j].getLength())\r
+ {\r
+ sequences.removeElement(seqs[j]);\r
+ j--;\r
+ jSize--;\r
+ }\r
+ else\r
+ {\r
+ seqs[j].setStart(newstart);\r
+ seqs[j].setSequence(seqs[j].getSequence().substring(i));\r
+ }\r
}\r
}\r
\r
*/\r
public void trimRight(int i)\r
{\r
- for (int j = 0; j < getHeight(); j++)\r
+ SequenceI[] seqs = getVisibleAndRepresentedSeqs();\r
+ int j, jSize = seqs.length;\r
+ for (j = 0; j < jSize; j++)\r
{\r
- SequenceI s = getSequenceAt(j);\r
- int newend = s.findPosition(i);\r
+ int newend = seqs[j].findPosition(i);\r
\r
- s.setEnd(newend);\r
- s.setSequence(s.getSequence().substring(0, i + 1));\r
+ seqs[j].setEnd(newend);\r
+ if(seqs[j].getLength()>i)\r
+ seqs[j].setSequence(seqs[j].getSequence().substring(0, i + 1));\r
}\r
}\r
\r
sequences.removeElementAt(i);\r
}\r
\r
- /**\r
- * DOCUMENT ME!\r
- *\r
- * @param threshold DOCUMENT ME!\r
- * @param sel DOCUMENT ME!\r
- *\r
- * @return DOCUMENT ME!\r
- */\r
- public Vector removeRedundancy(float threshold, Vector sel)\r
- {\r
- Vector del = new Vector();\r
-\r
- for (int i = 1; i < sel.size(); i++)\r
- {\r
- for (int j = 0; j < i; j++)\r
- {\r
- // Only do the comparison if either have not been deleted\r
- if (!del.contains((SequenceI) sel.elementAt(i)) ||\r
- !del.contains((SequenceI) sel.elementAt(j)))\r
- {\r
- // use PID instead of Comparison (which is really not pleasant)\r
- float pid = Comparison.PID((SequenceI) sel.elementAt(j),\r
- (SequenceI) sel.elementAt(i));\r
-\r
- if (pid >= threshold)\r
- {\r
- // Delete the shortest one\r
- if (((SequenceI) sel.elementAt(j)).getSequence().length() > ((SequenceI) sel\r
- .elementAt(\r
- i)).getSequence().length())\r
- {\r
- del.addElement(sel.elementAt(i));\r
- }\r
- else\r
- {\r
- del.addElement(sel.elementAt(i));\r
- }\r
- }\r
- }\r
- }\r
- }\r
-\r
- // Now delete the sequences\r
- for (int i = 0; i < del.size(); i++)\r
- {\r
- deleteSequence((SequenceI) del.elementAt(i));\r
- }\r
-\r
- return del;\r
- }\r
-\r
- /** */\r
- public SequenceGroup findGroup(int i)\r
- {\r
- return findGroup(getSequenceAt(i));\r
- }\r
\r
/** */\r
public SequenceGroup findGroup(SequenceI s)\r
{\r
SequenceGroup sg = (SequenceGroup) groups.elementAt(i);\r
\r
- if (sg.sequences.contains(s))\r
+ if (sg.getSequences(false).contains(s))\r
{\r
return sg;\r
}\r
for (int i = 0; i < gSize; i++)\r
{\r
SequenceGroup sg = (SequenceGroup) groups.elementAt(i);\r
- if(sg==null || sg.sequences==null)\r
+ if(sg==null || sg.getSequences(false)==null)\r
{\r
this.deleteGroup(sg);\r
gSize--;\r
continue;\r
}\r
\r
- if (sg.sequences.contains(s))\r
+ if (sg.getSequences(false).contains(s))\r
{\r
temp.addElement(sg);\r
}\r
return ret;\r
}\r
\r
- /**\r
- * DOCUMENT ME!\r
- *\r
- * @param sg DOCUMENT ME!\r
- */\r
- public void addSuperGroup(SuperGroup sg)\r
- {\r
- superGroup.addElement(sg);\r
- }\r
-\r
- /**\r
- * DOCUMENT ME!\r
- *\r
- * @param sg DOCUMENT ME!\r
- */\r
- public void removeSuperGroup(SuperGroup sg)\r
- {\r
- superGroup.removeElement(sg);\r
- }\r
-\r
- /**\r
- * DOCUMENT ME!\r
- *\r
- * @param sg DOCUMENT ME!\r
- *\r
- * @return DOCUMENT ME!\r
- */\r
- public SuperGroup getSuperGroup(SequenceGroup sg)\r
- {\r
- for (int i = 0; i < this.superGroup.size(); i++)\r
- {\r
- SuperGroup temp = (SuperGroup) superGroup.elementAt(i);\r
-\r
- if (temp.sequenceGroups.contains(sg))\r
- {\r
- return temp;\r
- }\r
- }\r
\r
- return null;\r
- }\r
\r
/** */\r
public void addGroup(SequenceGroup sg)\r
public void deleteAllGroups()\r
{\r
groups.removeAllElements();\r
- superGroup.removeAllElements();\r
\r
int i = 0;\r
\r
\r
while (i < sequences.size())\r
{\r
- SequenceI s = getSequenceAt(i);\r
-\r
- if (s.getName().equals(name))\r
+ if (getSequenceAt(i).getName().equals(name))\r
{\r
- return s;\r
+ return getSequenceAt(i);\r
}\r
\r
i++;\r
return null;\r
}\r
\r
- /** */\r
- public SequenceI findbyDisplayId(String name)\r
- {\r
- int i = 0;\r
-\r
- while (i < sequences.size())\r
- {\r
- SequenceI s = getSequenceAt(i);\r
-\r
- if (s.getDisplayId().equals(name))\r
- {\r
- return s;\r
- }\r
-\r
- i++;\r
- }\r
-\r
- return null;\r
- }\r
\r
/** */\r
public int findIndex(SequenceI s)\r
for (int i = 0; i < sequences.size(); i++)\r
{\r
Sequence seq = (Sequence) sequences.elementAt(i);\r
- seq.sequence = seq.sequence.replace('.', gc);\r
- seq.sequence = seq.sequence.replace('-', gc);\r
- seq.sequence = seq.sequence.replace(' ', gc);\r
+ seq.setSequence( seq.getSequence().replace('.', gc) );\r
+ seq.setSequence( seq.getSequence().replace('-', gc) );\r
+ seq.setSequence( seq.getSequence().replace(' ', gc) );\r
}\r
}\r
\r
annotations = temp;\r
}\r
\r
+\r
+ public void adjustSequenceAnnotations()\r
+ {\r
+ if(annotations!=null)\r
+ {\r
+ for (int a = 0; a < annotations.length; a++)\r
+ {\r
+ if (annotations[a].sequenceRef != null)\r
+ {\r
+ annotations[a].adjustForAlignment();\r
+ }\r
+ }\r
+ }\r
+ }\r
+\r
/**\r
* DOCUMENT ME!\r
*\r
public void addAnnotation(AlignmentAnnotation aa)\r
{\r
int aSize = 1;\r
-\r
if (annotations != null)\r
{\r
aSize = annotations.length + 1;\r
}\r
\r
AlignmentAnnotation[] temp = new AlignmentAnnotation[aSize];\r
+\r
+ temp[aSize-1] = aa;\r
+\r
int i = 0;\r
\r
if (aSize > 1)\r
{\r
- for (i = 0; i < (aSize - 1); i++)\r
+ for (i = 0; i < (aSize-1); i++)\r
{\r
temp[i] = annotations[i];\r
}\r
}\r
\r
- temp[i] = aa;\r
+ annotations = temp;\r
+ }\r
+\r
+ public void setAnnotationIndex(AlignmentAnnotation aa, int index)\r
+ {\r
+ if(aa==null || annotations==null || annotations.length-1<index)\r
+ return;\r
+\r
+ int aSize = annotations.length;\r
+ AlignmentAnnotation[] temp = new AlignmentAnnotation[aSize];\r
+\r
+ temp[index] = aa;\r
+\r
+ for (int i = 0; i < aSize; i++)\r
+ {\r
+ if(i==index)\r
+ continue;\r
+\r
+ if(i<index)\r
+ temp[i] = annotations[i];\r
+ else\r
+ temp[i] = annotations[i-1];\r
+ }\r
\r
annotations = temp;\r
}\r
{\r
return annotations;\r
}\r
+\r
+ public void setNucleotide(boolean b)\r
+ {\r
+ if(b)\r
+ type = NUCLEOTIDE;\r
+ else\r
+ type = PROTEIN;\r
+ }\r
+\r
+ public boolean isNucleotide()\r
+ {\r
+ if(type==NUCLEOTIDE)\r
+ return true;\r
+ else\r
+ return false;\r
+ }\r
+\r
+ public void setDataset(Alignment data)\r
+ {\r
+ if(dataset==null && data==null)\r
+ {\r
+ // Create a new dataset for this alignment.\r
+ // Can only be done once, if dataset is not null\r
+ // This will not be performed\r
+ Sequence[] seqs = new Sequence[getHeight()];\r
+ for (int i = 0; i < getHeight(); i++)\r
+ {\r
+ if(getSequenceAt(i).getDatasetSequence()!=null)\r
+ {\r
+ seqs[i] = (Sequence)getSequenceAt(i).getDatasetSequence();\r
+ }\r
+ else\r
+ {\r
+ seqs[i] = new Sequence(getSequenceAt(i).getName(),\r
+ AlignSeq.extractGaps(\r
+ jalview.util.Comparison.GapChars,\r
+ getSequenceAt(i).getSequence()\r
+ ),\r
+ getSequenceAt(i).getStart(),\r
+ getSequenceAt(i).getEnd());\r
+\r
+ getSequenceAt(i).setDatasetSequence(seqs[i]);\r
+ }\r
+ }\r
+\r
+ dataset = new Alignment(seqs);\r
+ }\r
+ else if(dataset==null && data!=null)\r
+ {\r
+ dataset = data;\r
+ }\r
+ }\r
+\r
+ public Alignment getDataset()\r
+ {\r
+ return dataset;\r
+ }\r
+\r
+ public boolean padGaps() {\r
+ boolean modified=false;\r
+\r
+ //Remove excess gaps from the end of alignment\r
+ int maxLength = -1;\r
+\r
+ SequenceI current;\r
+ for (int i = 0; i < sequences.size(); i++)\r
+ {\r
+ current = getSequenceAt(i);\r
+ for (int j = current.getLength(); j > maxLength; j--)\r
+ {\r
+ if (j > maxLength && !jalview.util.Comparison.isGap(\r
+ current.getCharAt(j)))\r
+ {\r
+ maxLength = j;\r
+ break;\r
+ }\r
+ }\r
+ }\r
+\r
+ maxLength++;\r
+\r
+ for (int i = 0; i < sequences.size();\r
+ i++)\r
+ {\r
+ current = getSequenceAt(i);\r
+\r
+ if (current.getLength() < maxLength)\r
+ {\r
+ current.insertCharAt(maxLength - 1, gapCharacter);\r
+ modified=true;\r
+ }\r
+ else if(current.getLength() > maxLength)\r
+ {\r
+ current.deleteChars(maxLength, current.getLength());\r
+ }\r
+ }\r
+ return modified;\r
+ }\r
+\r
+ public HiddenSequences getHiddenSequences()\r
+ {\r
+ return hiddenSequences;\r
+ }\r
+\r
+ SequenceI [] getVisibleAndRepresentedSeqs()\r
+ {\r
+ if(hiddenSequences==null || hiddenSequences.getSize()<1)\r
+ return getSequencesArray();\r
+\r
+ Vector seqs = new Vector();\r
+ SequenceI seq;\r
+ SequenceGroup hidden;\r
+ for (int i = 0; i < sequences.size(); i++)\r
+ {\r
+ seq = (SequenceI) sequences.elementAt(i);\r
+ seqs.addElement(seq);\r
+ hidden = seq.getHiddenSequences();\r
+ if(hidden!=null)\r
+ {\r
+ for(int j=0; j<hidden.getSize(false); j++)\r
+ {\r
+ seqs.addElement(hidden.getSequenceAt(j));\r
+ }\r
+ }\r
+ }\r
+ SequenceI [] result = new SequenceI[seqs.size()];\r
+ for(int i=0; i<seqs.size(); i++)\r
+ result[i] = (SequenceI)seqs.elementAt(i);\r
+\r
+ return result;\r
+\r
+ }\r
+\r
}\r
git://source.jalview.org / jalview.git / blobdiff