// put contigs from contfile into contlist
void Contiglist::add_contigs( std::string contigsfiles ){
std::stringstream ss;
ss.str( contigsfiles );
std::string filename;
std::string buffer("");
std::string line("");
std::string contig_id("");
while( getline( ss, filename, ',' )){
Log::Inst()->log_it( "contigsfile: " + filename );
// open contig file
std::ifstream cont( filename );
// read in contig objects
while( getline( cont, line ) ){
if( line[0] == '>' && buffer.length() != 0 ){
if( buffer.length() > 2*initial_trim + contig_sub_len ){
buffer = buffer.substr( initial_trim, buffer.length() - 2*initial_trim );
contigs.push_back( Contig( reads, buffer, contig_id ));
if( verbose ){
Log::Inst()->log_it( "Added contig: " + contig_id );
}
}
else if( buffer.length() > contig_sub_len ){
int trim = (buffer.length() - contig_sub_len) / 2;
buffer = buffer.substr( trim, buffer.length() - 2*trim );
contigs.push_back( Contig( reads, buffer, contig_id ));
if( verbose ){
Log::Inst()->log_it( "Added contig: " + contig_id );
}
}
buffer = "";
contig_id = line.substr(1);
}
else if ( line[0] == '>' ){
contig_id = line.substr(1);
}
else{
buffer += line;
}
}
// close contig file
cont.close();
}
// insert last line into contigs list
if( buffer.length() != 0 ){
contigs.push_back( Contig( reads, buffer, contig_id ) );
}
}
//JEREMIAH
bool SGVisitorContig::visit(StringGraph* /*pGraph*/, Vertex* pVertex)
{
//m_ct.push_back(Contig(pVertex->getID(), pVertex->getParentID(), pVertex->getSeq().toString(),
// pVertex->getSeq().length()));
m_ct.push_back(Contig(pVertex->getID(), pVertex->getSeq().toString()));
return false;
}
Contig
read_contig(const std::string& filename,
const std::string& name)
{
SeqType sequence=read_fasta(filename,name);
return Contig(name,sequence);
}
Contig
read_contig(const std::string& fasta_filename,
const std::string& quality_filename, const std::string& name)
{
SeqType sequence=read_fasta(fasta_filename,name);
QualSeqType quality=read_quality(quality_filename,name);
return Contig(name,sequence,quality);
}
int CopyNumEstimator::run(const Properties& options, const Arguments& arguments) {
int r = 0;
if ((r = checkOptions(options)) != 0) {
return r;
}
double lambda = options.get< double >("lambda");
LOG4CXX_DEBUG(logger, "copy_num_estimator begin");
ContigList contigs;
// read contigs
{
std::istream* is = &std::cin;
if (options.find("i") != options.not_found()) {
std::string file = options.get< std::string >("i");
is = new std::ifstream(file.c_str());
}
if (is == &std::cin) {
std::cin.sync_with_stdio(false);
}
size_t edge_num = 0;
CondensedDeBruijnGraphReader reader(*is);
CondensedDeBruijnGraphEdge edge;
while (reader.read(edge)) {
int copy_num = (int)std::round(edge.coverage / lambda);
if (copy_num >= 0) {
contigs.push_back(Contig(edge.seq, copy_num));
}
++edge_num;
}
LOG4CXX_DEBUG(logger, boost::format("Number of condensed edges = %d") % edge_num);
LOG4CXX_DEBUG(logger, boost::format("Number of Contigs = %d") % contigs.size());
if (is != &std::cin) {
delete is;
}
}
// write contigs
{
std::string file = options.get< std::string >("G");
if (!WriteContigs(file, contigs)) {
return 1;
}
}
// write components
{
std::string file = options.get< std::string >("C");
std::ofstream stream(file.c_str());
if (!stream) {
return 1;
}
size_t contig_idx = 0, component_idx = 0;
BOOST_FOREACH(const Contig& contig, contigs) {
if (contig.copy_num <= 1) {
stream << boost::format(">component\t%d\n%d\n\n") % component_idx % contig_idx;
++component_idx;
}
++contig_idx;
}
LOG4CXX_DEBUG(logger, boost::format("Number of Uniques = %d") % component_idx);
}
LOG4CXX_DEBUG(logger, "copy_num_estimator end");
return r;
}
