ColoredCDBG.hpp

Go to the documentation of this file.

#ifndef BIFROST_COLOREDCDBG_HPP
#define BIFROST_COLOREDCDBG_HPP
 
#include <iostream>
#include <random>
#include <unordered_map>
#include <unordered_set>
 
#include "CompactedDBG.hpp"
#include "DataManager.hpp"
 
#include "getRSS.h"
 
struct CCDBG_Build_opt : CDBG_Build_opt {
 
    string filename_colors_in;
 
    bool outputColors;
 
    CCDBG_Build_opt() : outputColors(true) {}
};
 
template<typename U = void> using UnitigColorMap = UnitigMap<DataAccessor<U>, DataStorage<U>>;
template<typename U = void> using const_UnitigColorMap = const_UnitigMap<DataAccessor<U>, DataStorage<U>>;
 
template<typename Unitig_data_t> //Curiously Recurring Template Pattern (CRTP)
class CCDBG_Data_t {
 
    typedef Unitig_data_t U;
 
    public:
 
        void clear(const UnitigColorMap<U>& um_dest){}
 
        void concat(const UnitigColorMap<U>& um_dest, const UnitigColorMap<U>& um_src){}
 
        void merge(const UnitigColorMap<U>& um_dest, const const_UnitigColorMap<U>& um_src){}
 
        void extract(const UnitigColors* uc_dest, const UnitigColorMap<U>& um_src, const bool last_extraction){}
 
        string serialize(const const_UnitigColorMap<U>& um_src) const {
 
            return string();
        }
};
 
template<typename Unitig_data_t = void>
class ColoredCDBG : public CompactedDBG<DataAccessor<Unitig_data_t>, DataStorage<Unitig_data_t>> {
 
    static_assert(is_void<Unitig_data_t>::value || is_base_of<CCDBG_Data_t<Unitig_data_t>, Unitig_data_t>::value,
                  "Type Unitig_data_t of data associated with vertices of class ColoredCDBG<Unitig_data_t> must "
                  " be void (no data) or a class extending class CCDBG_Data_t");
 
    typedef Unitig_data_t U;
 
    template<typename U> friend class DataAccessor;
 
    public:
 
        ColoredCDBG(int kmer_length = DEFAULT_K, int minimizer_length = -1);
 
        ColoredCDBG(const ColoredCDBG& o);
 
        ColoredCDBG(ColoredCDBG&& o);
 
        ColoredCDBG& operator=(const ColoredCDBG& o);
 
        ColoredCDBG& operator=(ColoredCDBG&& o);
 
        bool operator==(const ColoredCDBG& o) const;
 
        inline bool operator!=(const ColoredCDBG& o) const;
 
        ColoredCDBG& operator+=(const ColoredCDBG& o);
 
        void clear();
 
        bool buildGraph(const CCDBG_Build_opt& opt);
 
        bool buildColors(const CCDBG_Build_opt& opt);
 
        bool write(const string& prefix_output_fn, const size_t nb_threads = 1, const bool write_index_file = true, const bool compress_output = false, const bool verbose = false) const;
 
        bool read(const string& input_graph_fn, const string& input_colors_fn, const size_t nb_threads = 1, const bool verbose = false);
 
        bool read(const string& input_graph_fn, const string& input_index_fn, const string& input_colors_fn, const size_t nb_threads = 1, const bool verbose = false);
 
        bool merge(const ColoredCDBG& o, const size_t nb_threads = 1, const bool verbose = false);
 
        bool merge(ColoredCDBG&& o, const size_t nb_threads = 1, const bool verbose = false);
 
        bool merge(const vector<ColoredCDBG>& v, const size_t nb_threads = 1, const bool verbose = false);
 
        bool merge(vector<ColoredCDBG>&& v, const size_t nb_threads = 1, const bool verbose = false);
 
        string getColorName (const size_t color_id) const;
 
        vector<string> getColorNames() const;
 
        inline size_t getNbColors() const { return this->getData()->getNbColors(); }
 
        bool search(const vector<string>& query_filenames, const string& out_filename_prefix,
                    const double ratio_kmers, const bool inexact_search, const size_t nb_threads,
                    const bool verbose = false) const;
 
    private:
 
        void checkColors(const vector<string>& filename_seq_in) const;
        bool loadColors(const string& input_graph_fn, const string& input_colors_fn, const size_t nb_threads, const bool verbose);
 
        void initUnitigColors(const CCDBG_Build_opt& opt, const size_t max_nb_hash = 31);
        void buildUnitigColors(const size_t nb_threads);
        //void buildUnitigColors2(const size_t nb_threads);
 
        void resizeDataUC(const size_t sz, const size_t nb_threads = 1, const size_t max_nb_hash = 31);
 
        bool invalid;
};
 
#include "ColoredCDBG.tcc"
 
#endif