Assemble contigs. * @return the number of contigs assembled */
| 42 | * @return the number of contigs assembled |
| 43 | */ |
| 44 | static inline |
| 45 | size_t assemble(SequenceCollectionHash* seqCollection, |
| 46 | FastaWriter* fileWriter = NULL) |
| 47 | { |
| 48 | typedef SequenceCollectionHash Graph; |
| 49 | typedef graph_traits<Graph>::vertex_descriptor V; |
| 50 | typedef Graph::SymbolSetPair SymbolSetPair; |
| 51 | |
| 52 | Timer timer("Assemble"); |
| 53 | |
| 54 | size_t kmerCount = 0; |
| 55 | unsigned contigID = 0; |
| 56 | size_t assembledKmer = 0; |
| 57 | size_t lowCoverageKmer = 0; |
| 58 | size_t lowCoverageContigs = 0; |
| 59 | |
| 60 | for (Graph::iterator iter = seqCollection->begin(); |
| 61 | iter != seqCollection->end(); ++iter) { |
| 62 | if (iter->second.deleted()) |
| 63 | continue; |
| 64 | kmerCount++; |
| 65 | |
| 66 | extDirection dir; |
| 67 | SeqContiguity status = checkSeqContiguity(*iter, dir, true); |
| 68 | if (status == SC_CONTIGUOUS) |
| 69 | continue; |
| 70 | else if(status == SC_ISLAND) |
| 71 | { |
| 72 | BranchRecord currBranch(SENSE); |
| 73 | currBranch.push_back(*iter); |
| 74 | currBranch.terminate(BS_NOEXT); |
| 75 | size_t removed = assembleContig(seqCollection, |
| 76 | fileWriter, currBranch, contigID++); |
| 77 | assembledKmer += currBranch.size(); |
| 78 | if (removed > 0) { |
| 79 | lowCoverageContigs++; |
| 80 | lowCoverageKmer += removed; |
| 81 | } |
| 82 | continue; |
| 83 | } |
| 84 | assert(status == SC_ENDPOINT); |
| 85 | |
| 86 | BranchRecord currBranch(dir); |
| 87 | currBranch.push_back(*iter); |
| 88 | V currSeq = iter->first; |
| 89 | extendBranch(currBranch, currSeq, |
| 90 | iter->second.getExtension(dir)); |
| 91 | assert(currBranch.isActive()); |
| 92 | while(currBranch.isActive()) |
| 93 | { |
| 94 | SymbolSetPair extRec; |
| 95 | int multiplicity = -1; |
| 96 | bool success = seqCollection->getSeqData( |
| 97 | currSeq, extRec, multiplicity); |
| 98 | assert(success); |
| 99 | (void)success; |
| 100 | processLinearExtensionForBranch(currBranch, |
| 101 | currSeq, extRec, multiplicity, UINT_MAX); |
nothing calls this directly
no test coverage detected