Search in sources :

Example 1 with SAMSegment

use of cz1.ngs.model.SAMSegment in project polyGembler by c-zhou.

the class Anchor method run.

@Override
public void run() {
    // TODO Auto-generated method stub
    // read assembly graph file
    final GFA gfa = new GFA(query_file, asm_graph);
    qry_seqs = gfa.getSequenceMap();
    sub_seqs = Sequence.parseFastaFileAsMap(subject_file);
    myLogger.info("  GFA vertices: " + gfa.vertexSet().size());
    myLogger.info("  GFA edges   : " + gfa.edgeSet().size());
    // myLogger.info("  GFA edges --- ");
    // for(OverlapEdge olap : gfa.edgeSet())
    // myLogger.info(olap.olapInfo().toString());
    // find 'N/n's in subject/reference sequences
    // which could have impact on parsing the blast records
    sub_gaps = new HashMap<String, TreeRangeSet<Integer>>();
    for (Map.Entry<String, Sequence> entry : sub_seqs.entrySet()) {
        String seq_sn = entry.getKey();
        String seq_str = entry.getValue().seq_str();
        final TreeRangeSet<Integer> tmp_rangeSet = TreeRangeSet.create();
        for (int j = 0; j < seq_str.length(); j++) {
            if (seq_str.charAt(j) == 'N' || seq_str.charAt(j) == 'n')
                // blast record is 1-based closed coordination
                tmp_rangeSet.add(Range.closed(j + 1, j + 1).canonical(DiscreteDomain.integers()));
        }
        int seq_ln = seq_str.length();
        final TreeRangeSet<Integer> range_set = TreeRangeSet.create();
        for (Range<Integer> range : tmp_rangeSet.asRanges()) {
            int lowerend = range.hasLowerBound() ? Math.max(0, range.lowerEndpoint() - gap_buff) : 0;
            int upperend = range.hasUpperBound() ? Math.min(seq_ln, range.upperEndpoint() + gap_buff - 1) : seq_ln;
            range_set.add(Range.closed(lowerend, upperend).canonical(DiscreteDomain.integers()));
        }
        sub_gaps.put(seq_sn, range_set);
    }
    // read alignment file and place the query sequences
    final Map<String, Set<SAMSegment>> initPlace = new HashMap<String, Set<SAMSegment>>();
    final Map<String, List<SAMSegment>> initPseudoAssembly = new HashMap<String, List<SAMSegment>>();
    for (String sub_seq : sub_seqs.keySet()) initPseudoAssembly.put(sub_seq, new ArrayList<SAMSegment>());
    try {
        final SamReaderFactory factory = SamReaderFactory.makeDefault().enable(SamReaderFactory.Option.INCLUDE_SOURCE_IN_RECORDS, SamReaderFactory.Option.VALIDATE_CRC_CHECKSUMS).validationStringency(ValidationStringency.SILENT);
        final SamReader in1 = factory.open(new File(align_file));
        final SAMRecordIterator iter1 = in1.iterator();
        String qry;
        int qry_ln;
        double min_aln;
        final List<SAMSegment> buff = new ArrayList<SAMSegment>();
        SAMRecord rc = iter1.next();
        while (rc != null) {
            qry = rc.getReadName();
            qry_ln = qry_seqs.get(qry).seq_ln();
            buff.clear();
            if (!rc.getReadUnmappedFlag())
                buff.add(SAMSegment.samRecord(rc, true, qry_ln));
            while ((rc = iter1.next()) != null && rc.getReadName().equals(qry)) {
                buff.add(SAMSegment.samRecord(rc, true, qry_ln));
            }
            if (buff.isEmpty())
                continue;
            min_aln = 0.9 * buff.get(0).qlength();
            // keep alignment fragment that has qual>0
            Set<SAMSegment> init_f = new HashSet<SAMSegment>();
            Set<SAMSegment> init_r = new HashSet<SAMSegment>();
            for (SAMSegment record : buff) {
                if (record.qual() == 0 && record.qlength() < min_aln)
                    continue;
                if (record.qseqid().equals(qry))
                    init_f.add(record);
                else
                    init_r.add(record);
                initPseudoAssembly.get(record.sseqid()).add(record);
            }
            if (!init_f.isEmpty())
                initPlace.put(qry, init_f);
            if (!init_r.isEmpty())
                initPlace.put(qry + "'", init_r);
        }
        iter1.close();
        in1.close();
    } catch (IOException e) {
        // TODO Auto-generated catch block
        e.printStackTrace();
    }
    // Collections.sort(initPseudoAssembly.get("1_pilon"), new AlignmentSegment.SubjectCoordinationComparator());
    // if(debug) {
    // for(SAMSegment record : initPseudoAssembly.get("1_pilon")) {
    // System.out.println(record.qseqid()+":"+record.sstart()+"-"+record.send());
    // }
    // }
    final Set<SAMSegment> contained = new HashSet<SAMSegment>();
    final Set<SAMSegment> placed = new HashSet<SAMSegment>();
    final int flank_size = 10000;
    int distance;
    for (String sub_seq : sub_seqs.keySet()) {
        // sub_seq = "Chr10";
        if (sub_seq.equals("Chr00"))
            continue;
        myLogger.info(">>>>>>>>>>>>>" + sub_seq + "<<<<<<<<<<<<<<<<");
        final List<SAMSegment> seq_by_sub = initPseudoAssembly.get(sub_seq);
        Collections.sort(seq_by_sub, new AlignmentSegment.SubjectCoordinationComparator());
        placed.clear();
        int nSeq = seq_by_sub.size();
        double edge_penalty, edge_score;
        SAMSegment root_seq, source_seq, target_seq;
        Set<SAMSegment> target_seqs;
        Set<OverlapEdge> outgoing;
        TraceableEdge edge;
        String root_seqid, source_seqid, target_seqid;
        TraceableVertex<String> root_vertex, source_vertex, target_vertex;
        Deque<SAMSegment> deque = new ArrayDeque<SAMSegment>();
        final List<TraceableVertex<String>> traceable = new ArrayList<TraceableVertex<String>>();
        for (int i = 0; i < nSeq; i++) {
            root_seq = seq_by_sub.get(i);
            root_seqid = root_seq.qseqid();
            if (placed.contains(root_seq))
                continue;
            final TraceableDirectedWeightedPseudograph<String> razor = new TraceableDirectedWeightedPseudograph<String>(TraceableEdge.class);
            // final ListenableDirectedWeightedGraph<TraceableVertex<String>, DefaultWeightedEdge> razor =
            // new ListenableDirectedWeightedGraph<TraceableVertex<String>, DefaultWeightedEdge>(DefaultWeightedEdge.class);
            // JGraphModelAdapter<TraceableVertex<String>, DefaultWeightedEdge> jgAdapter =
            // new JGraphModelAdapter<TraceableVertex<String>, DefaultWeightedEdge>(razor);
            // JGraph jgraph = new JGraph(jgAdapter);
            deque.clear();
            deque.push(root_seq);
            contained.clear();
            while (!deque.isEmpty()) {
                source_seq = deque.pop();
                source_seqid = source_seq.qseqid();
                if (contained.contains(source_seq))
                    continue;
                contained.add(source_seq);
                source_vertex = new TraceableVertex<String>(source_seqid);
                source_vertex.setSAMSegment(source_seq);
                if (!razor.containsVertex(source_vertex))
                    razor.addVertex(source_vertex);
                outgoing = gfa.outgoingEdgesOf(source_seqid);
                for (OverlapEdge out : outgoing) {
                    target_seqid = gfa.getEdgeTarget(out);
                    if (!initPlace.containsKey(target_seqid))
                        continue;
                    target_seqs = initPlace.get(target_seqid);
                    distance = Integer.MAX_VALUE;
                    target_seq = null;
                    for (SAMSegment seq : target_seqs) {
                        int d = AlignmentSegment.sdistance(source_seq, seq);
                        if (d < distance) {
                            distance = d;
                            target_seq = seq;
                        }
                    }
                    if (distance <= flank_size) {
                        target_vertex = new TraceableVertex<String>(target_seqid);
                        target_vertex.setSAMSegment(target_seq);
                        if (!razor.containsVertex(target_vertex))
                            razor.addVertex(target_vertex);
                        if (razor.containsEdge(source_vertex, target_vertex))
                            continue;
                        edge = razor.addEdge(source_vertex, target_vertex);
                        // calculate edge weight
                        // higher weight edges are those,
                        /**
                         **
                         *							//       1.  large/long alignment segments vertices
                         *							// TODO: 2*. small gaps on the reference
                         *							edge_weight = qry_seqs.get(source_seqid).seq_ln()+
                         *									qry_seqs.get(target_seqid).seq_ln()-
                         *									gfa.getEdge(source_seqid, target_seqid).olap();
                         */
                        // TODO: 1*. large/long alignment segments vertices
                        // 2.  small gaps on the reference
                        edge_penalty = AlignmentSegment.sdistance(source_seq, target_seq);
                        edge.setPenalty(edge_penalty);
                        edge_score = qry_seqs.get(source_seqid).seq_ln() + qry_seqs.get(target_seqid).seq_ln() - gfa.getEdge(source_seqid, target_seqid).olap();
                        edge.setScore(edge_score);
                        deque.push(target_seq);
                    }
                }
            }
            if (ddebug)
                myLogger.info(root_seqid + " " + razor.vertexSet().size() + " " + razor.edgeSet().size() + " done");
            // JFrame frame = new JFrame();
            // frame.getContentPane().add(jgraph);
            // frame.setDefaultCloseOperation(JFrame.EXIT_ON_CLOSE);
            // frame.pack();
            // frame.setVisible(true);
            // "pseudo"-DFS to find the route with the highest score
            final Map<String, TraceableVertex<String>> razv_map = new HashMap<String, TraceableVertex<String>>();
            for (TraceableVertex<String> v : razor.vertexSet()) razv_map.put(v.getId(), v);
            // we use a bidirectional hashmap to simulate the deque
            // this is because we may need to do deletions
            // Deque<TraceableVertex<String>> queue = new ArrayDeque<TraceableVertex<String>>();
            final TreeBidiMap<Long, TraceableVertex<String>> bidiQ = new TreeBidiMap<Long, TraceableVertex<String>>();
            root_vertex = razv_map.get(root_seqid);
            root_vertex.setSAMSegment(root_seq);
            root_vertex.setScore(qry_seqs.get(root_seqid).seq_ln());
            root_vertex.setPenalty(0);
            root_vertex.setStatus(true);
            bidiQ.put(0L, root_vertex);
            double max_ws = Double.NEGATIVE_INFINITY, source_penalty, target_penalty, source_score, target_score, penalty, score, target_ws, source_ws, ws;
            int source_ln;
            Set<TraceableEdge> out_edges;
            TraceableVertex<String> opt_vertex = null;
            long sizeQ;
            boolean isLeaf;
            if (ddebug)
                for (TraceableEdge e : razor.edgeSet()) myLogger.info(e.toString() + "(" + razor.getEdgeSource(e).getSAMSegment().toString() + "|" + razor.getEdgeTarget(e).getSAMSegment().toString() + "|" + e.getScore() + "-" + e.getPenalty() + ")");
            while (!bidiQ.isEmpty()) {
                sizeQ = bidiQ.lastKey();
                source_vertex = bidiQ.get(sizeQ);
                bidiQ.remove(sizeQ);
                source_ln = qry_seqs.get(source_vertex.getId()).seq_ln();
                source_score = source_vertex.getScore() - source_ln;
                source_penalty = source_vertex.getPenalty();
                source_ws = source_score - source_penalty;
                isLeaf = true;
                out_edges = razor.outgoingEdgesOf(source_vertex);
                for (TraceableEdge out : out_edges) {
                    // this is not right because graph edges are immutable?
                    // target_vertex = razor.getEdgeTarget(out);
                    target_vertex = razv_map.get(razor.getEdgeTarget(out).getId());
                    target_score = target_vertex.getScore();
                    target_penalty = target_vertex.getPenalty();
                    target_ws = target_score - target_penalty;
                    edge_penalty = out.getPenalty();
                    penalty = source_penalty + edge_penalty;
                    edge_score = out.getScore();
                    score = source_score + edge_score;
                    ws = score - penalty;
                    if (edge_penalty > flank_size || target_vertex.getStatus() && (ws <= target_ws || isLoopback(razor, source_vertex, target_vertex)))
                        continue;
                    isLeaf = false;
                    target_vertex.setBackTrace(source_vertex);
                    target_vertex.setScore(score);
                    target_vertex.setPenalty(penalty);
                    target_vertex.setStatus(true);
                    bidiQ.put(sizeQ++, target_vertex);
                }
                if (isLeaf && source_ws > max_ws) {
                    penalty = source_vertex.getPenalty();
                    score = source_vertex.getScore();
                    max_ws = source_ws;
                    opt_vertex = source_vertex;
                    if (ddebug) {
                        String trace = opt_vertex.toString() + ":" + opt_vertex.getSAMSegment().sstart() + "-" + opt_vertex.getSAMSegment().send() + "(" + opt_vertex.getScore() + "-" + opt_vertex.getPenalty() + ")";
                        TraceableVertex<String> optx = opt_vertex;
                        while ((optx = optx.getBackTrace()) != null) {
                            trace += "," + optx.toString() + ":" + optx.getSAMSegment().sstart() + "-" + optx.getSAMSegment().send() + "(" + optx.getScore() + "-" + optx.getPenalty() + ")";
                        }
                        myLogger.info("trace back [" + score + ", " + penalty + "]: " + trace);
                    }
                }
            }
            traceable.add(opt_vertex);
            Set<TraceableVertex<String>> optx = new HashSet<TraceableVertex<String>>();
            optx.add(opt_vertex);
            while ((opt_vertex = opt_vertex.getBackTrace()) != null) optx.add(opt_vertex);
            for (TraceableVertex<String> v : optx) placed.add(v.getSAMSegment());
        }
        // sort traceable by size
        Collections.sort(traceable, new Comparator<TraceableVertex<String>>() {

            @Override
            public int compare(TraceableVertex<String> t0, TraceableVertex<String> t1) {
                // TODO Auto-generated method stub
                return Double.compare(t1.getScore(), t0.getScore());
            }
        });
        if (debug) {
            for (TraceableVertex<String> opt_vertex : traceable) {
                double score = opt_vertex.getScore();
                double penalty = opt_vertex.getPenalty();
                String trace = opt_vertex.toString() + ":" + opt_vertex.getSAMSegment().sstart() + "-" + opt_vertex.getSAMSegment().send() + "(" + opt_vertex.getScore() + "-" + opt_vertex.getPenalty() + ")";
                while ((opt_vertex = opt_vertex.getBackTrace()) != null) {
                    trace += "," + opt_vertex.toString() + ":" + opt_vertex.getSAMSegment().sstart() + "-" + opt_vertex.getSAMSegment().send() + "(" + opt_vertex.getScore() + "-" + opt_vertex.getPenalty() + ")";
                }
                myLogger.info("trace back [" + score + ", " + penalty + "]: " + trace);
            }
        }
        // we generate a compound alignment record for each traceable
        for (TraceableVertex<String> opt_vertex : traceable) {
        }
    }
}
Also used : SAMRecordIterator(htsjdk.samtools.SAMRecordIterator) HashMap(java.util.HashMap) ArrayList(java.util.ArrayList) ArrayList(java.util.ArrayList) List(java.util.List) HashSet(java.util.HashSet) OverlapEdge(cz1.ngs.model.OverlapEdge) Sequence(cz1.ngs.model.Sequence) SAMSegment(cz1.ngs.model.SAMSegment) AlignmentSegment(cz1.ngs.model.AlignmentSegment) TraceableVertex(cz1.ngs.model.TraceableVertex) SAMRecord(htsjdk.samtools.SAMRecord) HashMap(java.util.HashMap) Map(java.util.Map) BidiMap(org.apache.commons.collections4.BidiMap) TreeBidiMap(org.apache.commons.collections4.bidimap.TreeBidiMap) File(java.io.File) GFA(cz1.ngs.model.GFA) TreeRangeSet(com.google.common.collect.TreeRangeSet) HashSet(java.util.HashSet) Set(java.util.Set) TreeRangeSet(com.google.common.collect.TreeRangeSet) TraceableEdge(cz1.ngs.model.TraceableEdge) SamReader(htsjdk.samtools.SamReader) TraceableDirectedWeightedPseudograph(cz1.ngs.model.TraceableDirectedWeightedPseudograph) TreeBidiMap(org.apache.commons.collections4.bidimap.TreeBidiMap) SamReaderFactory(htsjdk.samtools.SamReaderFactory) IOException(java.io.IOException) ArrayDeque(java.util.ArrayDeque)

Aggregations

TreeRangeSet (com.google.common.collect.TreeRangeSet)1 AlignmentSegment (cz1.ngs.model.AlignmentSegment)1 GFA (cz1.ngs.model.GFA)1 OverlapEdge (cz1.ngs.model.OverlapEdge)1 SAMSegment (cz1.ngs.model.SAMSegment)1 Sequence (cz1.ngs.model.Sequence)1 TraceableDirectedWeightedPseudograph (cz1.ngs.model.TraceableDirectedWeightedPseudograph)1 TraceableEdge (cz1.ngs.model.TraceableEdge)1 TraceableVertex (cz1.ngs.model.TraceableVertex)1 SAMRecord (htsjdk.samtools.SAMRecord)1 SAMRecordIterator (htsjdk.samtools.SAMRecordIterator)1 SamReader (htsjdk.samtools.SamReader)1 SamReaderFactory (htsjdk.samtools.SamReaderFactory)1 File (java.io.File)1 IOException (java.io.IOException)1 ArrayDeque (java.util.ArrayDeque)1 ArrayList (java.util.ArrayList)1 HashMap (java.util.HashMap)1 HashSet (java.util.HashSet)1 List (java.util.List)1