1
|
// $Id$
|
2
|
/**
|
3
|
* Copyright (C) 2016 EDIT
|
4
|
* European Distributed Institute of Taxonomy
|
5
|
* http://www.e-taxonomy.eu
|
6
|
*
|
7
|
* The contents of this file are subject to the Mozilla Public License Version 1.1
|
8
|
* See LICENSE.TXT at the top of this package for the full license terms.
|
9
|
*/
|
10
|
package eu.etaxonomy.taxeditor.molecular.io;
|
11
|
|
12
|
|
13
|
import java.io.IOException;
|
14
|
import java.net.URI;
|
15
|
import java.util.ArrayList;
|
16
|
import java.util.Collections;
|
17
|
import java.util.Iterator;
|
18
|
import java.util.List;
|
19
|
|
20
|
import javax.xml.namespace.QName;
|
21
|
|
22
|
import eu.etaxonomy.cdm.model.molecular.Sequence;
|
23
|
import eu.etaxonomy.cdm.model.molecular.SingleReadAlignment;
|
24
|
import eu.etaxonomy.taxeditor.molecular.editor.e4.AlignmentEditorE4;
|
25
|
import eu.etaxonomy.taxeditor.molecular.l10n.Messages;
|
26
|
import info.bioinfweb.commons.bio.CharacterStateSetType;
|
27
|
import info.bioinfweb.commons.io.W3CXSConstants;
|
28
|
import info.bioinfweb.commons.text.StringUtils;
|
29
|
import info.bioinfweb.jphyloio.ReadWriteConstants;
|
30
|
import info.bioinfweb.jphyloio.ReadWriteParameterMap;
|
31
|
import info.bioinfweb.jphyloio.dataadapters.JPhyloIOEventReceiver;
|
32
|
import info.bioinfweb.jphyloio.dataadapters.MatrixDataAdapter;
|
33
|
import info.bioinfweb.jphyloio.dataadapters.ObjectListDataAdapter;
|
34
|
import info.bioinfweb.jphyloio.dataadapters.implementations.NoCharDefsNoSetsMatrixDataAdapter;
|
35
|
import info.bioinfweb.jphyloio.dataadapters.implementations.store.StoreObjectListDataAdapter;
|
36
|
import info.bioinfweb.jphyloio.events.CharacterSetIntervalEvent;
|
37
|
import info.bioinfweb.jphyloio.events.LinkedLabeledIDEvent;
|
38
|
import info.bioinfweb.jphyloio.events.SequenceTokensEvent;
|
39
|
import info.bioinfweb.jphyloio.events.TokenSetDefinitionEvent;
|
40
|
import info.bioinfweb.jphyloio.events.type.EventContentType;
|
41
|
import info.bioinfweb.jphyloio.utils.JPhyloIOWritingUtils;
|
42
|
|
43
|
|
44
|
|
45
|
/**
|
46
|
* In implementation of {@link MatrixDataAdapter} that delegates to a {@link Sequence} object.
|
47
|
* <p>
|
48
|
* Note that this adapter stores information on the {@link Sequence} in creation. Modifying the sequence or linked
|
49
|
* objects will require to create a new instance of this class. Otherwise unexpected behavior may be the consequence.
|
50
|
*
|
51
|
* @author Ben Stöver
|
52
|
* @date 29.04.2016
|
53
|
*/
|
54
|
public class CDMSequenceMatrixAdapter extends NoCharDefsNoSetsMatrixDataAdapter implements ReadWriteConstants, SingleReadAlignmentRDFXMLConstants {
|
55
|
|
56
|
private static final String NO_SEQUENCE_WITH_THE_ID_D_COULD_BE_FOUND = Messages.CDMSequenceMatrixAdapter_NO_SEQUENCE_FOUND;
|
57
|
public static final String MATRIX_ID = DEFAULT_MATRIX_ID_PREFIX + "ContigAlignment"; //$NON-NLS-1$
|
58
|
public static final String SINGLE_READ_SEQUENCE_ID_PREFIX = DEFAULT_SEQUENCE_ID_PREFIX + "SingleRead"; //$NON-NLS-1$
|
59
|
public static final String CONSENSUS_SEQUENCE_ID= DEFAULT_SEQUENCE_ID_PREFIX + "Consensus"; //$NON-NLS-1$
|
60
|
|
61
|
|
62
|
private Sequence sequence;
|
63
|
private boolean exportConsensus;
|
64
|
private List<SingleReadAlignment> singleReadList;
|
65
|
private String consensusSequenceLabel;
|
66
|
private ObjectListDataAdapter<TokenSetDefinitionEvent> tokenSetList;
|
67
|
//TODO Also allow specifying single read labels?
|
68
|
|
69
|
|
70
|
/**
|
71
|
* Creates a new instance of this class.
|
72
|
*
|
73
|
* @param sequence the <i>CDM</i> sequence object containing the data to be exported
|
74
|
* @param consensusSequenceLabel the label to be used for the consensus sequence (Maybe {@code null}.)
|
75
|
* @param exportConsensus Specify {@code true} here, if the consensus sequence shall be included in the export or {@code false}
|
76
|
* otherwise.
|
77
|
* @param exportSingleReads Specify {@code true} here, if the single reads shall be included in the export or {@code false}
|
78
|
* otherwise.
|
79
|
* @throws IllegalArgumentException if both {@code exportConsensus} and {@code exportSingleReads} are {@code false}
|
80
|
*/
|
81
|
public CDMSequenceMatrixAdapter(Sequence sequence, String consensusSequenceLabel, boolean exportConsensus, boolean exportSingleReads) {
|
82
|
super();
|
83
|
if (!exportConsensus && !exportSingleReads) {
|
84
|
throw new IllegalArgumentException(Messages.CDMSequenceMatrixAdapter_EITHER_MUST_BE_TRUE);
|
85
|
}
|
86
|
else {
|
87
|
this.sequence = sequence;
|
88
|
this.exportConsensus = exportConsensus;
|
89
|
this.consensusSequenceLabel = consensusSequenceLabel;
|
90
|
tokenSetList = createTokenSetList();
|
91
|
if (exportSingleReads) {
|
92
|
singleReadList = new ArrayList<SingleReadAlignment>(sequence.getSingleReadAlignments()); // Store references of single reads in defined order to allow random access.
|
93
|
//TODO Omit single reads that do not have an edited sequence yet?
|
94
|
}
|
95
|
else {
|
96
|
singleReadList = Collections.emptyList();
|
97
|
}
|
98
|
}
|
99
|
}
|
100
|
|
101
|
|
102
|
private ObjectListDataAdapter<TokenSetDefinitionEvent> createTokenSetList() {
|
103
|
StoreObjectListDataAdapter<TokenSetDefinitionEvent> result = new StoreObjectListDataAdapter<TokenSetDefinitionEvent>();
|
104
|
final String id = ReadWriteConstants.DEFAULT_TOKEN_SET_ID_PREFIX;
|
105
|
result.setObjectStartEvent(new TokenSetDefinitionEvent(CharacterStateSetType.DNA, id, null));
|
106
|
long length = getColumnCount(null); //TODO Change this expression, if column count should return -1 in the future.
|
107
|
if (length > 0) { // Empty character set interval events are not allowed. //TODO Handle -1 separately
|
108
|
result.getObjectContent(id).add(new CharacterSetIntervalEvent(0, length));
|
109
|
}
|
110
|
return result;
|
111
|
}
|
112
|
|
113
|
|
114
|
/**
|
115
|
* @return the sequence
|
116
|
*/
|
117
|
public Sequence getCDMSequence() {
|
118
|
return sequence;
|
119
|
}
|
120
|
|
121
|
|
122
|
private int extractSingleReadIndexFromID(String sequenceID) {
|
123
|
if (sequenceID.startsWith(SINGLE_READ_SEQUENCE_ID_PREFIX)) {
|
124
|
try {
|
125
|
return Integer.parseInt(sequenceID.substring(SINGLE_READ_SEQUENCE_ID_PREFIX.length()));
|
126
|
}
|
127
|
catch (NumberFormatException e) {} // fall through
|
128
|
}
|
129
|
return -1;
|
130
|
}
|
131
|
|
132
|
|
133
|
@Override
|
134
|
public LinkedLabeledIDEvent getStartEvent(ReadWriteParameterMap parameters) {
|
135
|
return new LinkedLabeledIDEvent(EventContentType.ALIGNMENT, MATRIX_ID, "Contig alignment", null); //$NON-NLS-1$
|
136
|
//TODO Use label according to derivate and markers.
|
137
|
}
|
138
|
|
139
|
|
140
|
@Override
|
141
|
public boolean containsLongTokens(ReadWriteParameterMap parameters) {
|
142
|
return false;
|
143
|
}
|
144
|
|
145
|
|
146
|
@Override
|
147
|
public long getColumnCount(ReadWriteParameterMap parameters) {
|
148
|
return -1; // Indicates that sequences may have different lengths. (Otherwise writing files without sequence elongation would not be possible.)
|
149
|
}
|
150
|
|
151
|
|
152
|
@Override
|
153
|
public long getSequenceCount(ReadWriteParameterMap parameters) {
|
154
|
int addend = 1;
|
155
|
if (!exportConsensus) {
|
156
|
addend = 0;
|
157
|
}
|
158
|
return singleReadList.size() + addend;
|
159
|
}
|
160
|
|
161
|
|
162
|
@Override
|
163
|
public Iterator<String> getSequenceIDIterator(ReadWriteParameterMap parameters) {
|
164
|
return new SequenceIDIterator(singleReadList.size(), exportConsensus);
|
165
|
}
|
166
|
|
167
|
|
168
|
@Override
|
169
|
public long getSequenceLength(ReadWriteParameterMap parameters, String sequenceID) throws IllegalArgumentException {
|
170
|
int singleReadIndex = extractSingleReadIndexFromID(sequenceID);
|
171
|
if (singleReadIndex >= 0) {
|
172
|
String sequence = singleReadList.get(singleReadIndex).getEditedSequence();
|
173
|
if (sequence != null) {
|
174
|
return sequence.length();
|
175
|
}
|
176
|
else { // This would happen e.g. if not edited sequence was copied from pherogram yet.
|
177
|
return 0;
|
178
|
}
|
179
|
}
|
180
|
else if (exportConsensus && CONSENSUS_SEQUENCE_ID.equals(sequenceID)) {
|
181
|
return getCDMSequence().getSequenceString().length();
|
182
|
}
|
183
|
else {
|
184
|
throw new IllegalArgumentException(String.format(NO_SEQUENCE_WITH_THE_ID_D_COULD_BE_FOUND, sequenceID));
|
185
|
}
|
186
|
}
|
187
|
|
188
|
|
189
|
@Override
|
190
|
public LinkedLabeledIDEvent getSequenceStartEvent(ReadWriteParameterMap parameters, String sequenceID) {
|
191
|
int singleReadIndex = extractSingleReadIndexFromID(sequenceID);
|
192
|
if (singleReadIndex >= 0) {
|
193
|
return new LinkedLabeledIDEvent(EventContentType.SEQUENCE, sequenceID, "Single read " + singleReadIndex, null); //$NON-NLS-1$
|
194
|
//TODO Use name displayed in derivate hierarchy or specified name as label instead?
|
195
|
}
|
196
|
else if (exportConsensus && CONSENSUS_SEQUENCE_ID.equals(sequenceID)) {
|
197
|
return new LinkedLabeledIDEvent(EventContentType.SEQUENCE, sequenceID, consensusSequenceLabel, null);
|
198
|
}
|
199
|
else {
|
200
|
throw new IllegalArgumentException(String.format(NO_SEQUENCE_WITH_THE_ID_D_COULD_BE_FOUND, sequenceID));
|
201
|
}
|
202
|
}
|
203
|
|
204
|
|
205
|
private void writeStringPart(JPhyloIOEventReceiver receiver, String string, long startColumn, long endColumn) throws IOException {
|
206
|
if (string != null) {
|
207
|
receiver.add(new SequenceTokensEvent(StringUtils.charSequenceToStringList(
|
208
|
string.substring((int)startColumn, (int)endColumn))));
|
209
|
}
|
210
|
}
|
211
|
|
212
|
|
213
|
private String createMetadataID(String sequenceID, QName predicate) {
|
214
|
return sequenceID + "META" + predicate.getLocalPart(); //$NON-NLS-1$
|
215
|
}
|
216
|
|
217
|
|
218
|
private void writeMetadataEvents(JPhyloIOEventReceiver receiver, String sequenceID, QName predicate, QName dataType, Object objectValue)
|
219
|
throws IOException {
|
220
|
|
221
|
if (objectValue != null) {
|
222
|
JPhyloIOWritingUtils.writeSimpleLiteralMetadata(receiver, createMetadataID(sequenceID, predicate), null,
|
223
|
predicate, dataType, objectValue, null);
|
224
|
}
|
225
|
}
|
226
|
|
227
|
|
228
|
@Override
|
229
|
public void writeSequencePartContentData(ReadWriteParameterMap parameters, JPhyloIOEventReceiver receiver, String sequenceID,
|
230
|
long startColumn, long endColumn) throws IOException, IllegalArgumentException {
|
231
|
|
232
|
int singleReadIndex = extractSingleReadIndexFromID(sequenceID);
|
233
|
if (singleReadIndex >= 0) {
|
234
|
SingleReadAlignment singleRead = singleReadList.get(singleReadIndex);
|
235
|
if (startColumn == 0) {
|
236
|
writeMetadataEvents(receiver, sequenceID, PREDICATE_IS_SINGLE_READ, W3CXSConstants.DATA_TYPE_BOOLEAN, new Boolean(true));
|
237
|
writeMetadataEvents(receiver, sequenceID, PREDICATE_IS_REVERSE_COMPLEMENTED, W3CXSConstants.DATA_TYPE_BOOLEAN, new Boolean(singleRead.isReverseComplement()));
|
238
|
writeMetadataEvents(receiver, sequenceID, PREDICATE_HAS_LEFT_CUT_POSITION, W3CXSConstants.DATA_TYPE_INT, singleRead.getLeftCutPosition());
|
239
|
writeMetadataEvents(receiver, sequenceID, PREDICATE_HAS_RIGHT_CUT_POSITION, W3CXSConstants.DATA_TYPE_INT, singleRead.getRightCutPosition());
|
240
|
|
241
|
URI pherogramURI = AlignmentEditorE4.getPherogramURI(singleRead.getSingleRead());
|
242
|
if (pherogramURI != null) {
|
243
|
JPhyloIOWritingUtils.writeTerminalResourceMetadata(receiver, createMetadataID(sequenceID, PREDICATE_HAS_PHEROGRAM),
|
244
|
null, PREDICATE_HAS_PHEROGRAM, pherogramURI);
|
245
|
}
|
246
|
|
247
|
writeMetadataEvents(receiver, sequenceID, PREDICATE_HAS_PHEROGRAM_ALIGNMENT, DATA_TYPE_PHERORAGM_ALIGNMENT, singleRead.getShifts());
|
248
|
//TODO This needs to be written with sequence type XML!
|
249
|
}
|
250
|
writeStringPart(receiver, singleRead.getEditedSequence(), startColumn, endColumn);
|
251
|
}
|
252
|
else if (exportConsensus && CONSENSUS_SEQUENCE_ID.equals(sequenceID)) {
|
253
|
if (startColumn == 0) {
|
254
|
writeMetadataEvents(receiver, sequenceID, PREDICATE_IS_CONSENSUS_SEQUENCE, W3CXSConstants.DATA_TYPE_BOOLEAN, new Boolean(true));
|
255
|
//TODO Possibly export additional properties of sequence (e.g. isBarcode(), getDdbjId(), ...) as metadata?
|
256
|
}
|
257
|
writeStringPart(receiver, getCDMSequence().getSequenceString(), startColumn, endColumn);
|
258
|
}
|
259
|
else {
|
260
|
throw new IllegalArgumentException(String.format(NO_SEQUENCE_WITH_THE_ID_D_COULD_BE_FOUND, sequenceID));
|
261
|
}
|
262
|
}
|
263
|
|
264
|
|
265
|
/* (non-Javadoc)
|
266
|
* @see info.bioinfweb.jphyloio.dataadapters.implementations.NoSetsMatrixDataAdapter#getTokenSets()
|
267
|
*/
|
268
|
@Override
|
269
|
public ObjectListDataAdapter<TokenSetDefinitionEvent> getTokenSets(ReadWriteParameterMap parameters) {
|
270
|
return tokenSetList;
|
271
|
}
|
272
|
}
|