void IterativeExtensions<span, Node, Edge, GraphDataVariant>::buildSequence (
const Node& node,
const Path_t<Node>& consensusRight,
size_t nbNodes,
size_t depth,
Sequence& seq
)
{
/** Shortcuts. */
Data& data = seq.getData();
size_t lenRight = consensusRight.size();
/** We compute the total size of the sequence. */
size_t fullLength = graph.getKmerSize() + consensusRight.size();
/* we need this in mapsembler: the first used kmer should be a k-1 mer. Indeed, if a first kmer is extracted from a sequence :
* -------------****** (k=6), then this node is the one linked to a new one starting with ******,
* thus with an overlap of k and not k-1.
*/
size_t offset = (depth == 0 && dont_output_first_nucleotide) ? 1 : 0;
/** We compute the total size of the sequence. */
size_t length = fullLength - offset;
/** We set the comment of the sequence. */
seq.setComment (Stringify::format ("%lli__len__%i__depth__%i", nbNodes, fullLength, depth) );
/** We set the data length. */
seq.getData().resize (length);
size_t idx=0;
size_t i = offset;
/** We dump the starting node. */
string nodeStr = graph.toString (node);
for ( ; i<nodeStr.size(); i++) { data[idx++] = nodeStr[i]; }
/** We dump the right part. */
for (i=0; i<lenRight; i++) { data[idx++] = ascii (consensusRight[i]); }
}
int main (int argc, char* argv[])
{
const size_t SPAN = KMER_SPAN(1);
/** Shortcuts. */
typedef Kmer<SPAN>::Type Type;
typedef Kmer<SPAN>::Count Count;
typedef Kmer<SPAN>::ModelCanonical ModelCanon;
typedef Kmer<SPAN>::ModelMinimizer <ModelCanon> Model;
size_t kmerSize = 33;
size_t mmerSize = 11;
/** We create a command line parser. */
OptionsParser parser ("GraphStats");
parser.push_back (new OptionOneParam (STR_URI_INPUT, "bank input", true));
try
{
/** We parse the user options. */
IProperties* options = parser.parse (argc, argv);
string filename = options->getStr (STR_URI_INPUT);
/** We create the solid kmers. */
Graph graph = Graph::create ("-in %s -kmer-size %d -bloom none -out toto.h5 -abundance-min 1", filename.c_str(), kmerSize);
/** We get the information of the solid kmers from the HDF5 file. */
Storage* storage = StorageFactory(STORAGE_HDF5).load ("toto.h5"); LOCAL (storage);
Group& dskGroup = storage->getGroup("dsk");
/** We get the solid kmers partition. */
Partition<Count>& partition = dskGroup.getPartition<Count> ("solid");
/** We create two kmers models. */
Model model (kmerSize, mmerSize);
Model modelK1 (kmerSize-1, mmerSize);
// We declare an output Bank
BankBinary outputBank (System::file().getBaseName(filename) + ".bin");
/** We create a sequence with BINARY encoding. */
Sequence seq (Data::ASCII);
/** We get an iterator over the [kmer,abundance] of solid kmers. */
Iterator<Count>* it = partition.iterator(); LOCAL (it);
/** We iterate the solid kmers. */
for (it->first(); !it->isDone(); it->next())
{
Type current = it->item().value;
cout << "kmer=" << it->item().value << " minimizer=" << model.getMinimizerValue(current)
<< " abundance=" << it->item().abundance << endl;
/** We interpret the kmer value as a Data object. */
Data data (Data::BINARY);
data.setRef ((char*) ¤t, model.getKmerSize());
modelK1.iterate (data, [&] (const Model::Kmer& k, size_t idx)
{
/** Shortcut. */
Type miniminizerCurrent = k.minimizer().value();
cout << "-> " << k.value()
<< " minimizer=" << miniminizerCurrent << " "
<< modelK1.getMmersModel().toString (miniminizerCurrent)
<< endl;
string tmp = modelK1.getMmersModel().toString (miniminizerCurrent);
/** We interpret the minimizer value as a Data object. */
seq.getData().setRef ((char*)tmp.c_str(), modelK1.getMmersModel().getKmerSize());
/** We insert the sequence into the binary bank. */
outputBank.insert (seq);
});
}
/** We flush the output bank. */
outputBank.flush();
/** We iterate the output bank. */
outputBank.iterate ([&] (const Sequence& s)
{
/** We get the kmer corresponding to the current sequence. */
ModelCanon::Kmer mini = modelK1.getMmersModel().codeSeed (s.getDataBuffer(), Data::BINARY);
cout << "mini=" << mini.value() << " " << modelK1.getMmersModel().toString (mini.value()) << endl;
});
}
catch (OptionFailure& e)
{
return e.displayErrors (std::cout);
}
catch (Exception& e)
{
std::cerr << "EXCEPTION: " << e.getMessage() << std::endl;
}
}
void operator() (Sequence& seq){
int used_windows_size=windows_size;
if (windows_size==0 || windows_size>seq.getDataSize()){ // if windows size == 0 then we use the full read length as windows
used_windows_size=seq.getDataSize();
}
if(not keep_low_complexity and not is_high_complexity(seq,kmer_size)){return;}
bool exists;
unsigned char count;
itKmer->setData (seq.getData());
vector<int> values; // For each position: number of occurrences of the kmer starting at this position.
// vector<int> covered_positions; // DEPRECATED [OLD WAY FOR COMPUTING SHARE KMER POSITIONS, FASTER BUT NON IMPLEMENTED WITH WINDOWS SIZE METHODS (max_populated_window)]
vector<bool> position_shared = vector<bool>(seq.getDataSize()); // boolean vector. A position is true if it's covered by at least a shared kmer.
for (int pos=0;pos<seq.getDataSize();pos++) position_shared[pos]=false;
int position=0;
for (itKmer->first(); !itKmer->isDone(); itKmer->next()){
quasiDico->get_value((*itKmer)->value().getVal(),exists,count);
if(!exists) {
count=0;
}
values.push_back(count);
// if (count>0) covered_positions.push_back(position); // DEPRECATED [OLD WAY FOR COMPUTING SHARE KMER POSITIONS, FASTER BUT NON IMPLEMENTED WITH WINDOWS SIZE METHODS (max_populated_window)]
if (count>0) { // TODO: OPTIMIZABLE.
for (int pos=position;pos<position+kmer_size && pos<=seq.getDataSize();pos++) position_shared[pos]=true;
}
position++;
}
// float percentage_shared_positions = 100*number_positions_covered_shared_kmer(covered_positions, seq.getDataSize())/float(seq.getDataSize()); // DEPRECATED [OLD WAY FOR COMPUTING SHARE KMER POSITIONS, FASTER BUT NON IMPLEMENTED WITH WINDOWS SIZE METHODS (max_populated_window)]
const int mpw = max_populated_window(position_shared,used_windows_size);
const float percentage_span_kmer = 100*mpw/float(used_windows_size);
// if (percentage_shared_positions !=percentage_span_kmer){cout<<percentage_shared_positions<< " == " <<percentage_span_kmer<<" ?"<<endl; exit(1);} // TO REMOVE
float mean;
int median, min, max;
if(mean_median_min_max(values, mean, median, min, max)){
string toPrint (to_string(seq.getIndex())+" "+to_string(mean)+" "+to_string(median)+" "+to_string(min)+" "+to_string(max)+" "+to_string(percentage_span_kmer));
// toPrint.append(" ");
// for(int i=0;i<seq.getDataSize() ;i++){
//
// if (position_shared[i]) {
// toPrint.append("1");
// }
// else toPrint.append("0");
// }
toPrint.append("\n");
if (percentage_span_kmer>=threshold) {
synchro->lock();
// bv->set(seq.getIndex());
fwrite(toPrint.c_str(), sizeof(char), toPrint.size(), outFile);
synchro->unlock ();
}
synchro->unlock ();
}
else{
string toPrint (to_string(seq.getIndex())+" none\n");
synchro->lock();
fwrite(toPrint.c_str(), sizeof(char), toPrint.size(), outFile);
synchro->unlock ();
}
}
void GlueCommander::output(string seq)
{
Sequence s (Data::ASCII);
s.getData().setRef ((char*)seq.c_str(), seq.size());
out->insert(s);
}
