Examples with Reference org.dishevelled.bio.assembly.gfa2.Reference used on opensource projects

Example 6 with Reference

use of org.dishevelled.bio.assembly.gfa2.Reference in project dishevelled-bio by heuermh.

the class TraversePaths method call.

@Override
public Integer call() throws Exception {
    BufferedReader reader = null;
    PrintWriter writer = null;
    try {
        reader = reader(inputGfa1File);
        writer = writer(outputGfa1File);
        final PrintWriter w = writer;
        Gfa1Reader.stream(reader, new Gfa1Listener() {

            @Override
            public boolean record(final Gfa1Record gfa1Record) {
                Gfa1Writer.write(gfa1Record, w);
                if (gfa1Record instanceof Path) {
                    Path path = (Path) gfa1Record;
                    int size = path.getSegments().size();
                    Reference source = null;
                    Reference target = null;
                    String overlap = null;
                    for (int i = 0; i < size; i++) {
                        target = path.getSegments().get(i);
                        if (i > 0) {
                            overlap = (path.getOverlaps() != null && path.getOverlaps().size() > i) ? path.getOverlaps().get(i - 1) : null;
                        }
                        if (source != null) {
                            Traversal traversal = new Traversal(path.getName(), i - 1, source, target, overlap, EMPTY_ANNOTATIONS);
                            Gfa1Writer.write(traversal, w);
                        }
                        source = target;
                    }
                }
                return true;
            }
        });
        return 0;
    } finally {
        try {
            reader.close();
        } catch (Exception e) {
        // ignore
        }
        try {
            writer.close();
        } catch (Exception e) {
        // ignore
        }
    }
}

Example 7 with Reference

use of org.dishevelled.bio.assembly.gfa2.Reference in project dishevelled by heuermh.

the class ImportGfa2Task method run.

@Override
public void run(final TaskMonitor taskMonitor) throws Exception {
    taskMonitor.setTitle("Import a network in Graphical Fragment Assembly (GFA) 2.0 format");
    final Map<String, Segment> segmentsById = new HashMap<String, Segment>();
    taskMonitor.setStatusMessage("Reading segments from file ...");
    try (BufferedReader readable = new BufferedReader(new FileReader(inputFile))) {
        // stream segments, building cache
        stream(readable, new Gfa2Adapter() {

            @Override
            protected boolean segment(final Segment segment) {
                segmentsById.put(segment.getId(), segment);
                return true;
            }
        });
    }
    taskMonitor.setStatusMessage("Finding reverse orientation references ...");
    final Table<String, Orientation, Segment> segmentsByOrientation = HashBasedTable.create();
    final List<Edge> edges = new ArrayList<Edge>();
    final List<Gap> gaps = new ArrayList<Gap>();
    final List<Path> paths = new ArrayList<Path>();
    try (BufferedReader readable = new BufferedReader(new FileReader(inputFile))) {
        // stream edges, gaps, and paths, looking for reverse orientation references
        stream(readable, new Gfa2Adapter() {

            private void putIfAbsent(final Reference reference) {
                Segment segment = segmentsById.get(reference.getId());
                if (segment == null) {
                    throw new RuntimeException("could not find segment by id " + reference.getId());
                }
                if (!segmentsByOrientation.contains(reference.getId(), reference.getOrientation())) {
                    segmentsByOrientation.put(reference.getId(), reference.getOrientation(), segment);
                }
            }

            @Override
            public boolean edge(final Edge edge) {
                putIfAbsent(edge.getSource());
                putIfAbsent(edge.getTarget());
                edges.add(edge);
                return true;
            }

            @Override
            public boolean gap(final Gap gap) {
                putIfAbsent(gap.getSource());
                putIfAbsent(gap.getTarget());
                gaps.add(gap);
                return true;
            }

            @Override
            public boolean path(final Path path) {
                for (Reference reference : path.getReferences()) {
                    putIfAbsent(reference);
                }
                if (loadPaths) {
                    paths.add(path);
                }
                return true;
            }
        });
    }
    logger.info("read {} segments, {} edges, {} gaps, and {} paths from {}", new Object[] { segmentsById.size(), edges.size(), gaps.size(), paths.size(), inputFile });
    segmentsById.clear();
    taskMonitor.setStatusMessage("Building Cytoscape nodes from segments ...");
    final CyNetwork network = applicationManager.getCurrentNetwork();
    final Map<String, CyNode> nodes = new HashMap<String, CyNode>(segmentsByOrientation.size());
    for (Table.Cell<String, Orientation, Segment> c : segmentsByOrientation.cellSet()) {
        String id = c.getRowKey();
        Orientation orientation = c.getColumnKey();
        Segment segment = c.getValue();
        String name = id + (orientation.isForward() ? "+" : "-");
        if (!nodes.containsKey(name)) {
            CyNode node = network.addNode();
            CyTable nodeTable = network.getDefaultNodeTable();
            CyRow nodeRow = nodeTable.getRow(node.getSUID());
            Integer length = segment.getLength();
            Integer readCount = segment.getReadCountOpt().orElse(null);
            Integer fragmentCount = segment.getFragmentCountOpt().orElse(null);
            Integer kmerCount = segment.getKmerCountOpt().orElse(null);
            String sequenceChecksum = segment.containsSequenceChecksum() ? String.valueOf(segment.getSequenceChecksum()) : null;
            String sequenceUri = segment.getSequenceUriOpt().orElse(null);
            setValue(nodeTable, nodeRow, "name", String.class, name);
            setValue(nodeTable, nodeRow, "length", Integer.class, length);
            setValue(nodeTable, nodeRow, "readCount", Integer.class, readCount);
            setValue(nodeTable, nodeRow, "fragmentCount", Integer.class, fragmentCount);
            setValue(nodeTable, nodeRow, "kmerCount", Integer.class, kmerCount);
            setValue(nodeTable, nodeRow, "sequenceChecksum", String.class, sequenceChecksum);
            setValue(nodeTable, nodeRow, "sequenceUri", String.class, sequenceUri);
            // default display length to length
            Integer displayLength = length;
            String sequence = orientation.isForward() ? segment.getSequence() : reverseComplement(segment.getSequence());
            if (sequence != null) {
                Integer sequenceLength = sequence.length();
                String displaySequence = trimFromMiddle(sequence, displaySequenceLimit);
                Integer displaySequenceLength = displaySequence.length();
                if (loadSequences) {
                    setValue(nodeTable, nodeRow, "sequence", String.class, sequence);
                }
                setValue(nodeTable, nodeRow, "sequenceLength", Integer.class, sequenceLength);
                setValue(nodeTable, nodeRow, "displaySequence", String.class, displaySequence);
                setValue(nodeTable, nodeRow, "displaySequenceLength", Integer.class, displaySequenceLength);
                // override display length with sequence length if necessary
                if (length == null || length != sequenceLength) {
                    displayLength = sequenceLength;
                }
            }
            StringBuilder sb = new StringBuilder();
            sb.append(name);
            if (displayLength != null) {
                sb.append("  ");
                sb.append(displayLength);
                sb.append(" bp");
            }
            String displayName = sb.toString();
            if (readCount != null) {
                sb.append(" ");
                sb.append(readCount);
                sb.append(" reads");
            }
            if (fragmentCount != null) {
                sb.append(" ");
                sb.append(fragmentCount);
                sb.append(" fragments");
            }
            if (kmerCount != null) {
                sb.append(" ");
                sb.append(kmerCount);
                sb.append(" kmers");
            }
            String displayLabel = sb.toString();
            setValue(nodeTable, nodeRow, "displayName", String.class, displayName);
            setValue(nodeTable, nodeRow, "displayLength", Integer.class, displayLength);
            setValue(nodeTable, nodeRow, "displayLabel", String.class, displayLabel);
            nodes.put(name, node);
        }
    }
    logger.info("converted segments and orientation to " + nodes.size() + " nodes");
    segmentsByOrientation.clear();
    taskMonitor.setStatusMessage("Building Cytoscape edges from edges and gaps ...");
    for (Edge edge : edges) {
        String sourceId = edge.getSource().getId();
        String sourceOrientation = edge.getSource().isForwardOrientation() ? "+" : "-";
        String targetId = edge.getTarget().getId();
        String targetOrientation = edge.getTarget().isForwardOrientation() ? "+" : "-";
        CyNode sourceNode = nodes.get(sourceId + sourceOrientation);
        CyNode targetNode = nodes.get(targetId + targetOrientation);
        CyEdge cyEdge = network.addEdge(sourceNode, targetNode, true);
        CyTable edgeTable = network.getDefaultEdgeTable();
        CyRow edgeRow = edgeTable.getRow(cyEdge.getSUID());
        setValue(edgeTable, edgeRow, "id", String.class, edge.getIdOpt().orElse(null));
        setValue(edgeTable, edgeRow, "type", String.class, "edge");
        setValue(edgeTable, edgeRow, "sourceId", String.class, sourceId);
        setValue(edgeTable, edgeRow, "sourceOrientation", String.class, sourceOrientation);
        setValue(edgeTable, edgeRow, "targetId", String.class, targetId);
        setValue(edgeTable, edgeRow, "targetOrientation", String.class, targetOrientation);
        setValue(edgeTable, edgeRow, "sourceStart", String.class, edge.getSourceStart().toString());
        setValue(edgeTable, edgeRow, "sourceEnd", String.class, edge.getSourceEnd().toString());
        setValue(edgeTable, edgeRow, "targetStart", String.class, edge.getTargetStart().toString());
        setValue(edgeTable, edgeRow, "targetEnd", String.class, edge.getTargetEnd().toString());
        setValue(edgeTable, edgeRow, "alignment", String.class, edge.hasAlignment() ? edge.getAlignment().toString() : null);
        setValue(edgeTable, edgeRow, "readCount", Integer.class, edge.getReadCountOpt().orElse(null));
        setValue(edgeTable, edgeRow, "fragmentCount", Integer.class, edge.getFragmentCountOpt().orElse(null));
        setValue(edgeTable, edgeRow, "kmerCount", Integer.class, edge.getKmerCountOpt().orElse(null));
        setValue(edgeTable, edgeRow, "mappingQuality", Integer.class, edge.getMappingQualityOpt().orElse(null));
        setValue(edgeTable, edgeRow, "mismatchCount", Integer.class, edge.getMismatchCountOpt().orElse(null));
    }
    logger.info("converted edges to " + edges.size() + " edges");
    for (Gap gap : gaps) {
        String sourceId = gap.getSource().getId();
        String sourceOrientation = gap.getSource().isForwardOrientation() ? "+" : "-";
        String targetId = gap.getTarget().getId();
        String targetOrientation = gap.getTarget().isForwardOrientation() ? "+" : "-";
        CyNode sourceNode = nodes.get(sourceId + sourceOrientation);
        CyNode targetNode = nodes.get(targetId + targetOrientation);
        CyEdge edge = network.addEdge(sourceNode, targetNode, true);
        CyTable edgeTable = network.getDefaultEdgeTable();
        CyRow edgeRow = edgeTable.getRow(edge.getSUID());
        setValue(edgeTable, edgeRow, "id", String.class, gap.getIdOpt().orElse(null));
        setValue(edgeTable, edgeRow, "type", String.class, "gap");
        setValue(edgeTable, edgeRow, "sourceId", String.class, sourceId);
        setValue(edgeTable, edgeRow, "sourceOrientation", String.class, sourceOrientation);
        setValue(edgeTable, edgeRow, "targetId", String.class, targetId);
        setValue(edgeTable, edgeRow, "targetOrientation", String.class, targetOrientation);
        setValue(edgeTable, edgeRow, "distance", Integer.class, gap.getDistance());
        setValue(edgeTable, edgeRow, "variance", Integer.class, gap.getVarianceOpt().orElse(null));
    }
    logger.info("converted gaps to " + gaps.size() + " edges");
    nodes.clear();
    edges.clear();
    gaps.clear();
    // pass paths to AssemblyApp if requested
    if (loadPaths && !paths.isEmpty()) {
        taskMonitor.setStatusMessage("Loading paths in path view ...");
        assemblyModel.setInputFileName(inputFile.toString());
    // todo: convert to gfa1 paths?
    // note paths in gfa2 can have references to segments, edges, or other groups
    // assemblyModel.setPaths(paths, traversalsByPathName);
    }
}

Example 8 with Reference

use of org.dishevelled.bio.assembly.gfa2.Reference in project molgenis-emx2 by molgenis.

the class SqlColumnRefArrayExecutor method createReferenceExistsCheck.

/**
 * trigger on this column to check if foreign key exists. Might be composite key, i.e., list of
 * columns
 */
private static void createReferenceExistsCheck(DSLContext jooq, Column column) {
    String schemaName = column.getSchema().getName();
    Name thisTable = name(schemaName, column.getTable().getTableName());
    Name toTable = name(column.getRefSchema(), column.getRefTableName());
    String functionName = getReferenceExistsCheckName(column);
    List<Reference> references = column.getReferences();
    String fromColumns = references.stream().map(r -> name(r.getName()).toString()).collect(Collectors.joining(","));
    String toColumns = references.stream().map(r -> name(r.getRefTo()).toString()).collect(Collectors.joining(","));
    String errorColumns = references.stream().map(r -> "COALESCE(error_row." + name(r.getRefTo()).toString() + ",'NULL')").collect(Collectors.joining("||','||"));
    String exceptFilter = references.stream().map(r -> {
        if (r.isOverlappingRef()) {
            return name(r.getRefTo()) + " = NEW." + name(r.getName());
        } else {
            return name(r.getRefTo()) + " = ANY (NEW." + name(r.getName()) + ")";
        }
    }).collect(Collectors.joining(" AND "));
    String unnestRefs = references.stream().map(r -> {
        // can be overlapping with non_array reference
        if (r.isOverlappingRef()) {
            return "NEW." + name(r.getName()) + " AS " + name(r.getRefTo());
        } else {
            return "UNNEST(NEW." + name(r.getName()) + ") AS " + name(r.getRefTo());
        }
    }).collect(Collectors.joining(","));
    String nonRefLinkFieldsAreNotNull = references.stream().filter(r -> !r.isOverlapping()).map(r2 -> "error_row." + name(r2.getRefTo()) + " IS NOT NULL ").collect(Collectors.joining(" OR "));
    jooq.execute("CREATE OR REPLACE FUNCTION {0}() RETURNS trigger AS $BODY$ " + "\nDECLARE error_row RECORD;" + "\nBEGIN" + "\n\tFOR error_row IN SELECT {1} EXCEPT SELECT {2} FROM {3} WHERE {10} LOOP" + // exclude if only refLink fields are set
    "\n\t\tIF {11} THEN" + "\n\t\t\tRAISE EXCEPTION USING ERRCODE='23503', MESSAGE = 'insert or update on table \"'||{9}||'\" violates foreign key (ref_array) constraint'" + " , DETAIL = 'Key ('||{6}||')=('|| {5} ||') is not present in table \"'||{7}||'\", column(s)('||{8}||')';" + "\n\t\tEND IF;" + "\n\tEND LOOP;" + "\n\tRETURN NEW;" + "\nEND; $BODY$ LANGUAGE plpgsql;", // 0
    name(schemaName, functionName), // 1
    keyword(unnestRefs), // 2
    keyword(toColumns), // 3
    toTable, // 4
    thisTable, // 5
    keyword(errorColumns), // 6
    inline(fromColumns), // 7
    inline(column.getRefTableName()), // 8
    inline(toColumns), // 9
    inline(column.getTableName()), // 10
    keyword(exceptFilter), // 11
    keyword(nonRefLinkFieldsAreNotNull));
    // add the trigger
    jooq.execute("CREATE CONSTRAINT TRIGGER {0} " + "\n\tAFTER INSERT OR UPDATE OF {1} ON {2} FROM {3}" + "\n\tDEFERRABLE INITIALLY IMMEDIATE " + "\n\tFOR EACH ROW EXECUTE PROCEDURE {4}()", name(functionName), keyword(fromColumns), thisTable, toTable, name(column.getTable().getSchema().getName(), functionName));
}

Example 9 with Reference

use of org.dishevelled.bio.assembly.gfa2.Reference in project dishevelled-bio by heuermh.

the class ReassemblePaths method call.

@Override
public Integer call() throws Exception {
    BufferedReader reader = null;
    PrintWriter writer = null;
    try {
        reader = reader(inputGfa1File);
        writer = writer(outputGfa1File);
        final PrintWriter w = writer;
        final List<Path> paths = new ArrayList<Path>();
        final ListMultimap<String, Traversal> traversalsByPathName = ArrayListMultimap.create();
        Gfa1Reader.stream(reader, new Gfa1Listener() {

            @Override
            public boolean record(final Gfa1Record gfa1Record) {
                if (gfa1Record instanceof Path) {
                    Path path = (Path) gfa1Record;
                    paths.add(path);
                } else if (gfa1Record instanceof Traversal) {
                    Traversal traversal = (Traversal) gfa1Record;
                    traversalsByPathName.put(traversal.getPathName(), traversal);
                } else {
                    Gfa1Writer.write(gfa1Record, w);
                }
                return true;
            }
        });
        for (Path path : paths) {
            List<Traversal> traversals = traversalsByPathName.get(path.getName());
            Collections.sort(traversals, new Comparator<Traversal>() {

                @Override
                public int compare(final Traversal t0, final Traversal t1) {
                    return t0.getOrdinal() - t1.getOrdinal();
                }
            });
            List<Reference> segments = new ArrayList<Reference>();
            List<String> overlaps = new ArrayList<String>();
            for (Traversal traversal : traversals) {
                if (segments.isEmpty()) {
                    segments.add(traversal.getSource());
                }
                segments.add(traversal.getTarget());
                if (traversal.hasOverlap()) {
                    overlaps.add(traversal.getOverlap());
                }
            }
            Gfa1Writer.write(new Path(path.getName(), segments, overlaps.isEmpty() ? null : overlaps, path.getAnnotations()), w);
        }
        return 0;
    } finally {
        try {
            reader.close();
        } catch (Exception e) {
        // ignore
        }
        try {
            writer.close();
        } catch (Exception e) {
        // ignore
        }
    }
}