eu.etaxonomy.taxeditor.molecular/src/main/java/eu/etaxonomy/taxeditor/molecular/io/CDMSequenceMatrixAdapter.java

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