void Assemble(HashGraph &hash_graph)
{
cout << "kmers " << hash_graph.num_vertices() << " "<< hash_graph.num_edges() << endl;
int kmer_size = hash_graph.kmer_size();
double min_cover = max(1, (kmer_size == option.mink ? option.min_count : option.min_support));
Histgram<int> hist = hash_graph.coverage_histgram();
double expected_coverage = hist.mean();
deque<Sequence> contigs;
deque<ContigInfo> contig_infos;
hash_graph.Assemble(contigs, contig_infos);
hash_graph.clear();
{
HashGraph tmp_hash_graph;
tmp_hash_graph.swap(hash_graph);
}
ContigGraph contig_graph(kmer_size, contigs, contig_infos);
contigs.clear();
contig_infos.clear();
contig_graph.RemoveDeadEnd(option.min_contig);
int bubble = contig_graph.RemoveBubble();
cout << "merge bubble " << bubble << endl;
contig_graph.MergeSimilarPath();
if (!option.is_no_coverage)
contig_graph.RemoveLocalLowCoverage(min_cover, option.min_contig, 0.1);
contig_graph.SortVertices();
contig_graph.GetContigs(contigs, contig_infos);
WriteSequence(option.graph_file(kmer_size), contigs);
contigs.clear();
contig_infos.clear();
if (!option.is_no_coverage)
{
double ratio = (kmer_size < option.maxk) ? 0.5 : 0.2;
if (ratio < 2.0 / expected_coverage)
ratio = 2.0 / expected_coverage;
contig_graph.IterateLocalCoverage(option.min_contig, ratio, min_cover, 1e100, 1.1);
contig_graph.MergeSimilarPath();
}
deque<Sequence> multi_contigs;
deque<ContigInfo> multi_contig_infos;
contig_graph.SortVertices();
contig_graph.GetContigs(multi_contigs, multi_contig_infos);
PrintN50(multi_contigs);
//WriteSequence(option.contig_file(kmer_size), multi_contigs);
WriteContig(option.contig_file(kmer_size), multi_contigs, multi_contig_infos, FormatString("contig-%d", kmer_size));
//WriteContigInfo(option.contig_info_file(kmer_size), multi_contig_infos);
}
void Assemble(HashGraph &hash_graph)
{
cout << "kmers " << hash_graph.num_vertices() << " "<< hash_graph.num_edges() << endl;
int kmer_size = hash_graph.kmer_size();
double min_cover = max(1, (kmer_size == option.mink ? option.min_count : option.min_support));
Histgram<int> hist = hash_graph.coverage_histgram();
//double expected_coverage = hist.mean();
deque<Sequence> contigs;
deque<ContigInfo> contig_infos;
hash_graph.Assemble(contigs, contig_infos);
hash_graph.clear();
{
HashGraph tmp_hash_graph;
tmp_hash_graph.swap(hash_graph);
}
ContigGraph contig_graph(kmer_size, contigs, contig_infos);
contigs.clear();
contig_infos.clear();
if (!option.is_no_coverage)
{
contig_graph.RemoveStandAlone(kmer_size);
int bubble = contig_graph.RemoveBubble();
cout << "merge bubble " << bubble << endl;
contig_graph.RemoveLocalLowCoverage(min_cover, option.min_contig, 0.1);
}
contig_graph.SortVertices();
contig_graph.GetContigs(contigs, contig_infos);
WriteSequence(option.graph_file(kmer_size), contigs);
contigs.clear();
contig_infos.clear();
if (!option.is_no_coverage)
{
double ratio = 0.25;
deque<Sequence> multi_contigs;
deque<ContigInfo> multi_contig_infos;
contig_graph.GetContigs(multi_contigs, multi_contig_infos);
PrintN50(multi_contigs);
contig_graph.Trim(10);
contig_graph.MergeSimilarPath();
contig_graph.GetContigs(multi_contigs, multi_contig_infos);
contig_graph.InitializeTable();
contig_graph.IterateComponentCoverage2(option.min_contig, ratio, min_cover, 1e100, 1.1, max_component_size);
contig_graph.GetContigs(multi_contigs, multi_contig_infos);
contig_graph.Trim(10);
contig_graph.Prune(kmer_size);
contig_graph.GetContigs(multi_contigs, multi_contig_infos);
contig_graph.MergeSimilarPath();
}
deque<Sequence> multi_contigs;
deque<ContigInfo> multi_contig_infos;
contig_graph.SortVertices();
contig_graph.GetContigs(multi_contigs, multi_contig_infos);
PrintN50(multi_contigs);
WriteSequence(option.contig_file(kmer_size), multi_contigs);
WriteContigInfo(option.contig_info_file(kmer_size), multi_contig_infos);
deque<Sequence> transcripts;
FindIsoforms(contig_graph, transcripts);
int index = 0;
for (unsigned i = 0; i < transcripts.size(); ++i)
{
if (transcripts[i].size() >= 300)
transcripts[index++] = transcripts[i];
}
transcripts.resize(index);
PrintN50(transcripts);
WriteSequence(option.transcript_file(kmer_size), transcripts, FormatString("transcript-%d", kmer_size));
}