X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=src%2Fjalview%2Fdatamodel%2FSeqCigar.java;h=46989bfb92168d578bbed482c93f60a33f602e92;hb=797df64fa2a0a30773d0f48f5494d4155e5a8be3;hp=536e4eaed3480678db86ac1de18b64d0a79be74e;hpb=60508bc218cee42c6fa3405db19f7790acafabab;p=jalview.git
diff --git a/src/jalview/datamodel/SeqCigar.java b/src/jalview/datamodel/SeqCigar.java
index 536e4ea..46989bf 100644
--- a/src/jalview/datamodel/SeqCigar.java
+++ b/src/jalview/datamodel/SeqCigar.java
@@ -1,299 +1,770 @@
+/*
+ * Jalview - A Sequence Alignment Editor and Viewer (Version 2.7)
+ * Copyright (C) 2011 J Procter, AM Waterhouse, J Engelhardt, LM Lui, G Barton, M Clamp, S Searle
+ *
+ * This file is part of Jalview.
+ *
+ * Jalview is free software: you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.
+ *
+ * Jalview is distributed in the hope that it will be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR
+ * PURPOSE. See the GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License along with Jalview. If not, see .
+ */
package jalview.datamodel;
-import jalview.analysis.AlignSeq;
+import java.util.Enumeration;
+import java.util.Hashtable;
+import java.util.Vector;
-public class SeqCigar
- extends CigarSimple
+import jalview.analysis.*;
+import jalview.util.*;
+
+public class SeqCigar extends CigarSimple
{
+ /**
+ * start(inclusive) and end(exclusive) of subsequence on refseq
+ */
+ private int start, end;
+
+ private SequenceI refseq = null;
+
+ private Hashtable seqProps;
- private SequenceI refseq=null;
/**
* Reference dataset sequence for the cigar string
+ *
* @return SequenceI
*/
- public SequenceI getRefSeq() {
+ public SequenceI getRefSeq()
+ {
return refseq;
}
+
+ /**
+ *
+ * @return int start index of cigar ops on refSeq
+ */
+ public int getStart()
+ {
+ return start;
+ }
+
+ /**
+ *
+ * @return int end index (exclusive) of cigar ops on refSeq
+ */
+ public int getEnd()
+ {
+ return end;
+ }
+
/**
* Returns sequence as a string with cigar operations applied to it
+ *
* @return String
*/
public String getSequenceString(char GapChar)
{
- return (length==0) ? "" : (String) getSequenceAndDeletions(refseq.getSequence(), GapChar)[0];
+ return (length == 0) ? "" : (String) getSequenceAndDeletions(
+ refseq.getSequenceAsString(start, end), GapChar)[0];
}
/**
- * recreates a gapped and edited version of RefSeq or null for an empty cigar string
+ * recreates a gapped and edited version of RefSeq or null for an empty cigar
+ * string
+ *
* @return SequenceI
*/
- public SequenceI getSeq(char GapChar) {
+ public SequenceI getSeq(char GapChar)
+ {
Sequence seq;
- if (refseq==null || length==0)
+ if (refseq == null || length == 0)
+ {
return null;
- Object[] edit_result=getSequenceAndDeletions(refseq.getSequence(), GapChar);
- if (edit_result==null)
- throw new Error("Implementation Error - unexpected null from getSequenceAndDeletions");
-
- seq = new Sequence(refseq.getName(), (String) edit_result[0], refseq.getStart()+((int[]) edit_result[1])[0], refseq.getStart()+((int[]) edit_result[1])[2]);
+ }
+ Object[] edit_result = getSequenceAndDeletions(
+ refseq.getSequenceAsString(start, end), GapChar);
+ if (edit_result == null)
+ {
+ throw new Error(
+ "Implementation Error - unexpected null from getSequenceAndDeletions");
+ }
+ int bounds[] = (int[]) edit_result[1];
+ seq = new Sequence(refseq.getName(), (String) edit_result[0],
+ refseq.getStart() + start + bounds[0], refseq.getStart()
+ + start + ((bounds[2] == 0) ? -1 : bounds[2]));
+ seq.setDescription(refseq.getDescription());
+ int sstart = seq.getStart(), send = seq.getEnd();
+ // seq.checkValidRange(); probably not needed
+ // recover local properties if present
+ if (seqProps != null)
+ {
+ // this recovers dataset sequence reference as well as local features,
+ // names, start/end settings.
+ SeqsetUtils.SeqCharacterUnhash(seq, seqProps);
+ }
+ // ensure dataset sequence is up to date from local reference
seq.setDatasetSequence(refseq);
+ seq.setStart(sstart);
+ seq.setEnd(send);
return seq;
}
+
/*
- We don't allow this - refseq is given at construction time only
- public void setSeq(SequenceI seq) {
- this.seq = seq;
- }
- */
+ * We don't allow this - refseq is given at construction time only public void
+ * setSeq(SequenceI seq) { this.seq = seq; }
+ */
/**
- * internal constructor - sets seq to a gapless sequence derived from seq
- * and prepends any 'D' operations needed to get to the first residue of seq.
- * @param seq SequenceI
+ * internal constructor - sets seq to a gapless sequence derived from seq and
+ * prepends any 'D' operations needed to get to the first residue of seq.
+ *
+ * @param seq
+ * SequenceI
+ * @param initialDeletion
+ * true to mark initial dataset sequence residues as deleted in
+ * subsequence
+ * @param _s
+ * index of first position in seq
+ * @param _e
+ * index after last position in (possibly gapped) seq
* @return true if gaps are present in seq
*/
- private boolean _setSeq(SequenceI seq) {
- boolean hasgaps=false;
-
- if (seq==null)
- throw new Error("Implementation Error - _setSeq(null)");
-
- // Find correct sequence to reference and add initial hidden offset
+ private boolean _setSeq(SequenceI seq, boolean initialDeletion, int _s,
+ int _e)
+ {
+ boolean hasgaps = false;
+ if (seq == null)
+ {
+ throw new Error("Implementation Error - _setSeq(null,...)");
+ }
+ if (_s < 0)
+ {
+ throw new Error("Implementation Error: _s=" + _s);
+ }
+ String seq_string = seq.getSequenceAsString();
+ if (_e == 0 || _e < _s || _e > seq_string.length())
+ {
+ _e = seq_string.length();
+ }
+ // resolve start and end positions relative to ungapped reference sequence
+ start = seq.findPosition(_s) - seq.getStart();
+ end = seq.findPosition(_e) - seq.getStart();
+ int l_ungapped = end - start;
+ // Find correct sequence to reference and correct start and end - if
+ // necessary
SequenceI ds = seq.getDatasetSequence();
- if (ds==null) {
- ds = new Sequence(seq.getName(),
- AlignSeq.extractGaps(jalview.util.Comparison.GapChars, new String(seq.getSequence())),
- seq.getStart(),
- seq.getEnd());
- }
- // check that we haven't just duplicated an ungapped sequence.
- if (ds.getLength()==seq.getLength()) {
+ if (ds == null)
+ {
+ // make a new dataset sequence
+ String ungapped = AlignSeq.extractGaps(
+ jalview.util.Comparison.GapChars, new String(seq_string));
+ l_ungapped = ungapped.length();
+ // check that we haven't just duplicated an ungapped sequence.
+ if (l_ungapped == seq.getLength())
+ {
ds = seq;
- } else {
- hasgaps = true;
}
- this.refseq = ds;
- // Adjust offset
- if (ds.getStart() ds.getLength())
+ {
+ throw new Error(
+ "SeqCigar: Possible implementation error: sequence is longer than dataset sequence");
+ // end = ds.getLength();
+ }
+
return hasgaps;
}
+
/**
- * directly initialise a cigar object with a sequence of range, operation pairs and a sequence to apply it to.
- * operation and range should be relative to the seq.getStart()'th residue of the dataset seq resolved from seq.
- * @param seq SequenceI
- * @param operation char[]
- * @param range int[]
+ * directly initialise a cigar object with a sequence of range, operation
+ * pairs and a sequence to apply it to. operation and range should be relative
+ * to the seq.getStart()'th residue of the dataset seq resolved from seq.
+ *
+ * @param seq
+ * SequenceI
+ * @param operation
+ * char[]
+ * @param range
+ * int[]
*/
- public SeqCigar(SequenceI seq, char operation[], int range[]) {
+ public SeqCigar(SequenceI seq, char operation[], int range[])
+ {
super();
- if (seq==null)
+ if (seq == null)
+ {
throw new Error("Implementation Bug. Null seq !");
- if (operation.length!=range.length) {
- throw new Error("Implementation Bug. Cigar Operation list!= range list");
+ }
+ if (operation.length != range.length)
+ {
+ throw new Error(
+ "Implementation Bug. Cigar Operation list!= range list");
}
- if (operation!=null) {
- this.operation = new char[operation.length+_inc_length];
- this.range = new int[operation.length+_inc_length];
+ if (operation != null)
+ {
+ this.operation = new char[operation.length + _inc_length];
+ this.range = new int[operation.length + _inc_length];
- if (_setSeq(seq)) {
- throw new Error("NOT YET Implemented: Constructing a Cigar object from a cigar string and a gapped sequence.");
+ if (_setSeq(seq, false, 0, 0))
+ {
+ throw new Error(
+ "NOT YET Implemented: Constructing a Cigar object from a cigar string and a gapped sequence.");
}
- for (int i = this.length, j=0; j < operation.length; i++,j++)
+ for (int i = this.length, j = 0; j < operation.length; i++, j++)
{
char op = operation[j];
if (op != M && op != I && op != D)
{
- throw new Error(
- "Implementation Bug. Cigar Operation '"+j+"' '"+op+"' not one of '"+M+"', '"+I+"', or '"+D+"'.");
+ throw new Error("Implementation Bug. Cigar Operation '" + j
+ + "' '" + op + "' not one of '" + M + "', '" + I
+ + "', or '" + D + "'.");
}
this.operation[i] = op;
this.range[i] = range[j];
}
- this.length+=operation.length;
- } else {
+ this.length += operation.length;
+ }
+ else
+ {
this.operation = null;
this.range = null;
- this.length=0;
- if (_setSeq(seq)) {
- throw new Error("NOT YET Implemented: Constructing a Cigar object from a cigar string and a gapped sequence.");
+ this.length = 0;
+ if (_setSeq(seq, false, 0, 0))
+ {
+ throw new Error(
+ "NOT YET Implemented: Constructing a Cigar object from a cigar string and a gapped sequence.");
}
}
}
+
/**
* add range matched residues to cigar string
- * @param range int
+ *
+ * @param range
+ * int
*/
- public void addMatch(int range) {
+ public void addMatch(int range)
+ {
this.addOperation(M, range);
}
+
/**
- * Deleted regions mean that there will be discontinuous sequence numbering in the
- * sequence returned by getSeq(char).
- * @return true if there are non-terminal deletions
+ * Adds insertion and match operations based on seq to the cigar up to the
+ * endpos column of seq.
+ *
+ * @param cigar
+ * CigarBase
+ * @param seq
+ * SequenceI
+ * @param startpos
+ * int
+ * @param endpos
+ * int
+ * @param initialDeletions
+ * if true then initial deletions will be added from start of seq to
+ * startpos
*/
- public boolean hasDeletedRegions() {
- for (int i=1, l=length-1; i 0 && op != I)
+ if (isGap)
{
- cigar.addOperation(op, range);
- range = 0;
+ if (range > 0 && op != I)
+ {
+ cigar.addOperation(op, range);
+ range = 0;
+ }
+ op = I;
+ range++;
+ }
+ else
+ {
+ if (range > 0 && op != M)
+ {
+ cigar.addOperation(op, range);
+ range = 0;
+ }
+ op = M;
+ range++;
}
- op = I;
- range++;
}
else
{
- if (range > 0 && op != M)
+ if (!isGap)
{
- cigar.addOperation(op, range);
- range = 0;
+ if (range > 0 && op != D)
+ {
+ cigar.addOperation(op, range);
+ range = 0;
+ }
+ op = D;
+ range++;
}
- op = M;
- range++;
- }
- }
- else
- {
- if (!isGap)
- {
- if (range > 0 && op != D)
+ else
{
- cigar.addOperation(op, range);
- range = 0;
+ // do nothing - insertions are not made in flanking regions
}
- op = D;
- range++;
- }
- else
- {
- // do nothing - insertions are not recorded in flanking regions.
}
+ p++;
+ }
+ if (range > 0)
+ {
+ cigar.addOperation(op, range);
}
}
- if (range > 0)
- {
- cigar.addOperation(op, range);
- }
- }
+
/**
* create a cigar string for given sequence
- * @param seq SequenceI
+ *
+ * @param seq
+ * SequenceI
*/
- public SeqCigar(SequenceI seq) {
+ public SeqCigar(SequenceI seq)
+ {
super();
if (seq == null)
- throw new Error("Implementation error for new Cigar(SequenceI)");
- if (_setSeq(seq))
{
- // there is still work to do
- addSequenceOps(this, seq, 0, seq.getLength());
+ throw new Error("Implementation error for new Cigar(SequenceI)");
}
+ _setSeq(seq, false, 0, 0);
+ // there is still work to do
+ addSequenceOps(this, seq, 0, seq.getLength() - 1, false);
}
- public SeqCigar(SequenceI seq, int start, int end) {
+
+ /**
+ * Create Cigar from a range of gaps and residues on a sequence object
+ *
+ * @param seq
+ * SequenceI
+ * @param start
+ * int - first column in range
+ * @param end
+ * int - last column in range
+ */
+ public SeqCigar(SequenceI seq, int start, int end)
+ {
super();
if (seq == null)
- throw new Error("Implementation error for new Cigar(SequenceI)");
- if (_setSeq(seq))
{
- // there is still work to do
- addSequenceOps(this, seq, start, end);
+ throw new Error("Implementation error for new Cigar(SequenceI)");
}
+ _setSeq(seq, false, start, end + 1);
+ // there is still work to do
+ addSequenceOps(this, seq, start, end, false);
}
/**
- * Create a cigar object from a cigar string like '[]+'
- * Will fail if the given seq already contains gaps (JBPNote: future implementation will fix)
- * @param seq SequenceI object resolvable to a dataset sequence
- * @param cigarString String
+ * Create a cigar object from a cigar string like '[]+' Will
+ * fail if the given seq already contains gaps (JBPNote: future implementation
+ * will fix)
+ *
+ * @param seq
+ * SequenceI object resolvable to a dataset sequence
+ * @param cigarString
+ * String
* @return Cigar
*/
public static SeqCigar parseCigar(SequenceI seq, String cigarString)
- throws Exception
+ throws Exception
{
Object[] opsandrange = parseCigarString(cigarString);
- return new SeqCigar(seq, (char[]) opsandrange[0], (int[]) opsandrange[1]);
+ return new SeqCigar(seq, (char[]) opsandrange[0],
+ (int[]) opsandrange[1]);
+ }
+
+ /**
+ * create an alignment from the given array of cigar sequences and gap
+ * character, and marking the given segments as visible in the given
+ * columselection.
+ *
+ * @param alseqs
+ * @param gapCharacter
+ * @param colsel
+ * - columnSelection where hidden regions are marked
+ * @param segments
+ * - visible regions of alignment
+ * @return SequenceI[]
+ */
+ public static SequenceI[] createAlignmentSequences(SeqCigar[] alseqs,
+ char gapCharacter, ColumnSelection colsel, int[] segments)
+ {
+ SequenceI[] seqs = new SequenceI[alseqs.length];
+ StringBuffer[] g_seqs = new StringBuffer[alseqs.length];
+ String[] alseqs_string = new String[alseqs.length];
+ Object[] gs_regions = new Object[alseqs.length];
+ for (int i = 0; i < alseqs.length; i++)
+ {
+ alseqs_string[i] = alseqs[i].getRefSeq().getSequenceAsString(
+ alseqs[i].start, alseqs[i].end);
+ gs_regions[i] = alseqs[i].getSequenceAndDeletions(alseqs_string[i],
+ gapCharacter); // gapped sequence, {start, start col, end.
+ // endcol}, hidden regions {{start, end, col}})
+ if (gs_regions[i] == null)
+ {
+ throw new Error("Implementation error: " + i
+ + "'th sequence Cigar has no operations.");
+ }
+ g_seqs[i] = new StringBuffer((String) ((Object[]) gs_regions[i])[0]); // the
+ // visible
+ // gapped
+ // sequence
+ }
+ // Now account for insertions. (well - deletions)
+ // this is complicated because we must keep track of shifted positions in
+ // each sequence
+ ShiftList shifts = new ShiftList();
+ for (int i = 0; i < alseqs.length; i++)
+ {
+ Object[] gs_region = ((Object[]) ((Object[]) gs_regions[i])[2]);
+ if (gs_region != null)
+
+ {
+ for (int hr = 0; hr < gs_region.length; hr++)
+ {
+ int[] region = (int[]) gs_region[hr];
+ char[] insert = new char[region[1] - region[0] + 1];
+ for (int s = 0; s < insert.length; s++)
+ {
+ insert[s] = gapCharacter;
+ }
+ int inspos = shifts.shift(region[2]); // resolve insertion position in
+ // current alignment frame of
+ // reference
+ for (int s = 0; s < alseqs.length; s++)
+ {
+ if (s != i)
+ {
+ if (g_seqs[s].length() <= inspos)
+ {
+ // prefix insertion with more gaps.
+ for (int l = inspos - g_seqs[s].length(); l > 0; l--)
+ {
+ g_seqs[s].append(gapCharacter); // to debug - use a diffferent
+ // gap character here
+ }
+ }
+ g_seqs[s].insert(inspos, insert);
+ }
+ else
+ {
+ g_seqs[s].insert(inspos,
+ alseqs_string[i].substring(region[0], region[1] + 1));
+ }
+ }
+ shifts.addShift(region[2], insert.length); // update shift in
+ // alignment frame of
+ // reference
+ if (segments == null)
+ {
+ // add a hidden column for this deletion
+ colsel.hideColumns(inspos, inspos + insert.length - 1);
+ }
+ }
+ }
+ }
+ for (int i = 0; i < alseqs.length; i++)
+ {
+ int[] bounds = ((int[]) ((Object[]) gs_regions[i])[1]);
+ SequenceI ref = alseqs[i].getRefSeq();
+ seqs[i] = new Sequence(ref.getName(), g_seqs[i].toString(),
+ ref.getStart() + alseqs[i].start + bounds[0], ref.getStart()
+ + alseqs[i].start + (bounds[2] == 0 ? -1 : bounds[2]));
+ seqs[i].setDatasetSequence(ref);
+ seqs[i].setDescription(ref.getDescription());
+ }
+ if (segments != null)
+ {
+ for (int i = 0; i < segments.length; i += 3)
+ {
+ // int start=shifts.shift(segments[i]-1)+1;
+ // int end=shifts.shift(segments[i]+segments[i+1]-1)-1;
+ colsel.hideColumns(segments[i + 1], segments[i + 1]
+ + segments[i + 2] - 1);
+ }
+ }
+ return seqs;
}
+
/**
* non rigorous testing
*/
/**
- *
- * @param seq Sequence
- * @param ex_cs_gapped String
+ *
+ * @param seq
+ * Sequence
+ * @param ex_cs_gapped
+ * String
* @return String
*/
- public static String testCigar_string(Sequence seq, String ex_cs_gapped) {
+ public static String testCigar_string(Sequence seq, String ex_cs_gapped)
+ {
SeqCigar c_sgapped = new SeqCigar(seq);
String cs_gapped = c_sgapped.getCigarstring();
if (!cs_gapped.equals(ex_cs_gapped))
- System.err.println("Failed getCigarstring: incorect string '"+cs_gapped+"' != "+ex_cs_gapped);
+ {
+ System.err.println("Failed getCigarstring: incorect string '"
+ + cs_gapped + "' != " + ex_cs_gapped);
+ }
return cs_gapped;
}
- public static boolean testSeqRecovery(SeqCigar gen_sgapped, SequenceI s_gapped) {
+
+ public static boolean testSeqRecovery(SeqCigar gen_sgapped,
+ SequenceI s_gapped)
+ {
+ // this is non-rigorous - start and end recovery is not tested.
SequenceI gen_sgapped_s = gen_sgapped.getSeq('-');
- if (!gen_sgapped_s.getSequence().equals(s_gapped.getSequence())) {
- System.err.println("Couldn't reconstruct sequence.\n" +
- gen_sgapped_s.getSequence() + "\n" +
- s_gapped.getSequence());
+ if (!gen_sgapped_s.getSequence().equals(s_gapped.getSequence()))
+ {
+ System.err.println("Couldn't reconstruct sequence.\n"
+ + gen_sgapped_s.getSequenceAsString() + "\n"
+ + s_gapped.getSequenceAsString());
return false;
}
return true;
}
- public static void main(String argv[]) throws Exception {
- Sequence s=new Sequence("MySeq", "asdfktryasdtqwrtsaslldddptyipqqwaslchvhttt",39,80);
+
+ public static void main(String argv[]) throws Exception
+ {
+ String o_seq;
+ Sequence s = new Sequence("MySeq",
+ o_seq = "asdfktryasdtqwrtsaslldddptyipqqwaslchvhttt", 39, 80);
String orig_gapped;
- Sequence s_gapped=new Sequence("MySeq", orig_gapped="----asdf------ktryas---dtqwrtsasll----dddptyipqqwa----slchvhttt", 39,80);
- String ex_cs_gapped="4I4M6I6M3I11M4I12M4I9M";
+ Sequence s_gapped = new Sequence(
+ "MySeq",
+ orig_gapped = "----asdf------ktryas---dtqwrtsasll----dddptyipqqwa----slchvhttt",
+ 39, 80);
+ String ex_cs_gapped = "4I4M6I6M3I11M4I12M4I9M";
s_gapped.setDatasetSequence(s);
String sub_gapped_s;
- Sequence s_subsequence_gapped=new Sequence("MySeq", sub_gapped_s="------ktryas---dtqwrtsasll----dddptyipqqwa----slchvh", 43,77);
+ Sequence s_subsequence_gapped = new Sequence(
+ "MySeq",
+ sub_gapped_s = "------ktryas---dtqwrtsasll----dddptyipqqwa----slchvh",
+ 43, 77);
s_subsequence_gapped.setDatasetSequence(s);
SeqCigar c_null = new SeqCigar(s);
String cs_null = c_null.getCigarstring();
- if (cs_null.length()>0)
- System.err.println("Failed getCigarstring: Unexpected cigar operations:"+cs_null);
+ if (!cs_null.equals("42M"))
+ {
+ System.err
+ .println("Failed to recover ungapped sequence cigar operations:"
+ + ((cs_null == "") ? "empty string" : cs_null));
+ }
testCigar_string(s_gapped, ex_cs_gapped);
SeqCigar gen_sgapped = SeqCigar.parseCigar(s, ex_cs_gapped);
if (!gen_sgapped.getCigarstring().equals(ex_cs_gapped))
- System.err.println("Failed parseCigar("+ex_cs_gapped+")->getCigarString()->'"+gen_sgapped.getCigarstring()+"'");
+ {
+ System.err.println("Failed parseCigar(" + ex_cs_gapped
+ + ")->getCigarString()->'" + gen_sgapped.getCigarstring()
+ + "'");
+ }
testSeqRecovery(gen_sgapped, s_gapped);
// Test dataset resolution
SeqCigar sub_gapped = new SeqCigar(s_subsequence_gapped);
if (!testSeqRecovery(sub_gapped, s_subsequence_gapped))
- System.err.println("Failed recovery for subsequence of dataset sequence");
+ {
+ System.err
+ .println("Failed recovery for subsequence of dataset sequence");
+ }
// width functions
- if (sub_gapped.getWidth()!=sub_gapped_s.length())
+ if (sub_gapped.getWidth() != sub_gapped_s.length())
+ {
System.err.println("Failed getWidth()");
+ }
sub_gapped.getFullWidth();
if (sub_gapped.hasDeletedRegions())
+ {
System.err.println("hasDeletedRegions is incorrect.");
+ }
// Test start-end region SeqCigar
- SeqCigar sub_se_gp= new SeqCigar(s_subsequence_gapped, 8, 48);
- if (sub_se_gp.getWidth()!=40)
- System.err.println("SeqCigar(seq, start, end) not properly clipped alignsequence.");
- System.out.println("Original sequence align:\n"+sub_gapped_s+"\nReconstructed window from 8 to 48\n"+"XXXXXXXX"+sub_se_gp.getSequenceString('-')+"...."+"\nCigar String:"+sub_se_gp.getCigarstring()+"");
+ SeqCigar sub_se_gp = new SeqCigar(s_subsequence_gapped, 8, 48);
+ if (sub_se_gp.getWidth() != 41)
+ {
+ System.err
+ .println("SeqCigar(seq, start, end) not properly clipped alignsequence.");
+ }
+ System.out.println("Original sequence align:\n" + sub_gapped_s
+ + "\nReconstructed window from 8 to 48\n" + "XXXXXXXX"
+ + sub_se_gp.getSequenceString('-') + "..." + "\nCigar String:"
+ + sub_se_gp.getCigarstring() + "\n");
+ SequenceI ssgp = sub_se_gp.getSeq('-');
+ System.out.println("\t " + ssgp.getSequenceAsString());
+ for (int r = 0; r < 10; r++)
+ {
+ sub_se_gp = new SeqCigar(s_subsequence_gapped, 8, 48);
+ int sl = sub_se_gp.getWidth();
+ int st = sl - 1 - r;
+ for (int rs = 0; rs < 10; rs++)
+ {
+ int e = st + rs;
+ sub_se_gp.deleteRange(st, e);
+ String ssgapedseq = sub_se_gp.getSeq('-').getSequenceAsString();
+ System.out.println(st + "," + e + "\t:" + ssgapedseq);
+ st -= 3;
+ }
+ }
+ {
+ SeqCigar[] set = new SeqCigar[]
+ { new SeqCigar(s), new SeqCigar(s_subsequence_gapped, 8, 48),
+ new SeqCigar(s_gapped) };
+ Alignment al = new Alignment(set);
+ for (int i = 0; i < al.getHeight(); i++)
+ {
+ System.out.println("" + al.getSequenceAt(i).getName() + "\t"
+ + al.getSequenceAt(i).getStart() + "\t"
+ + al.getSequenceAt(i).getEnd() + "\t"
+ + al.getSequenceAt(i).getSequenceAsString());
+ }
+ }
+ {
+ System.out.println("Gapped.");
+ SeqCigar[] set = new SeqCigar[]
+ { new SeqCigar(s), new SeqCigar(s_subsequence_gapped, 8, 48),
+ new SeqCigar(s_gapped) };
+ set[0].deleteRange(20, 25);
+ Alignment al = new Alignment(set);
+ for (int i = 0; i < al.getHeight(); i++)
+ {
+ System.out.println("" + al.getSequenceAt(i).getName() + "\t"
+ + al.getSequenceAt(i).getStart() + "\t"
+ + al.getSequenceAt(i).getEnd() + "\t"
+ + al.getSequenceAt(i).getSequenceAsString());
+ }
+ }
+ // if (!ssgapedseq.equals("ryas---dtqqwa----slchvh"))
+ // System.err.println("Subseqgaped\n------ktryas---dtqwrtsasll----dddptyipqqwa----slchvhryas---dtqwrtsasll--qwa----slchvh\n"+ssgapedseq+"\n"+sub_se_gp.getCigarstring());
}
+ /**
+ * references to entities that this sequence cigar is associated with.
+ */
+ private Hashtable selGroups = null;
+
+ public void setGroupMembership(Object group)
+ {
+ if (selGroups == null)
+ {
+ selGroups = new Hashtable();
+ }
+ selGroups.put(group, new int[0]);
+ }
+
+ /**
+ * Test for and if present remove association to group.
+ *
+ * @param group
+ * @return true if group was associated and it was removed
+ */
+ public boolean removeGroupMembership(Object group)
+ {
+ if (selGroups != null && selGroups.containsKey(group))
+ {
+ selGroups.remove(group);
+ return true;
+ }
+ return false;
+ }
+
+ /**
+ * forget all associations for this sequence.
+ */
+ public void clearMemberships()
+ {
+ if (selGroups != null)
+ {
+ selGroups.clear();
+ }
+ selGroups = null;
+ }
+
+ /**
+ *
+ * @return null or array of all associated entities
+ */
+ public Object[] getAllMemberships()
+ {
+ if (selGroups == null)
+ {
+ return null;
+ }
+ Object[] mmbs = new Object[selGroups.size()];
+ Enumeration en = selGroups.keys();
+ for (int i = 0; en.hasMoreElements(); i++)
+ {
+ mmbs[i] = en.nextElement();
+ }
+ return mmbs;
+ }
+
+ /**
+ * Test for group membership
+ *
+ * @param sgr
+ * - a selection group or some other object that may be associated
+ * with seqCigar
+ * @return true if sgr is associated with this seqCigar
+ */
+ public boolean isMemberOf(Object sgr)
+ {
+ return (selGroups != null) && selGroups.get(sgr) != null;
+ }
}