C++ KmerCounter::get_kmer_count方法代码示例

kmer_int_type_t IRKE::extract_best_seed(vector<kmer_int_type_t> &kmer_vec,
                                        KmerCounter &kcounter,
                                        float min_connectivity)
{

    unsigned int kmer_length = kcounter.get_kmer_length();

    unsigned int best_kmer_count = 0;
    kmer_int_type_t best_seed = 0;

    for (unsigned int i = 0; i < kmer_vec.size(); i++) {

        kmer_int_type_t kmer = kmer_vec[i];
        unsigned int count = kcounter.get_kmer_count(kmer);

        if (count > best_kmer_count && is_good_seed_kmer(kmer, count, kmer_length, min_connectivity)) {
            best_kmer_count = count;
            best_seed = kmer;
        }
    }

    if (IRKE_COMMON::MONITOR >= 2) {
        cerr << "Parallel method found better seed: " << kcounter.get_kmer_string(best_seed) << " with count: "
            << best_kmer_count << endl;
    }

    return (best_seed);
}

vector<unsigned int> compute_kmer_coverage(string& sequence, KmerCounter& kcounter) {
    vector<unsigned int> coverage;
    if(IRKE_COMMON::MONITOR) {
        cerr << "processing sequence: " << sequence << endl;
    }
    for (int i = 0; i <= (int) sequence.length() - KMER_SIZE; i++) {
        // cerr << "i: " << i << ", <= " << sequence.length() - KMER_SIZE << endl;
        string kmer = sequence.substr(i, KMER_SIZE);
        if(IRKE_COMMON::MONITOR >= 2) {
            for (int j = 0; j <= i; j++) {
                cerr << " ";
            }
            cerr << kmer << endl;
        }
        unsigned int kmer_count = 0;
        if(!contains_non_gatc(kmer)) {
            kmer_count = kcounter.get_kmer_count(kmer);
        }

        // Note, in the jellyfish run, we restrain it to min kmer coverage of 2. 
        // If we don't find a kmer catalogued, it must have a kmer count of 1.
        if (kmer_count < 1) {
            kmer_count = 1;
        }
        
        coverage.push_back(kmer_count);
    }
    return(coverage);
}

string IRKE::reconstruct_path_sequence(KmerCounter& kcounter, vector<kmer_int_type_t>& path, vector<unsigned int>& cov_counter) {
	
	if (path.size() == 0) {
		return("");
	}
	
	string seq = kcounter.get_kmer_string(path[0]);
	cov_counter.push_back( kcounter.get_kmer_count(path[0]) );
	
	for (unsigned int i = 1; i < path.size(); i++) {
		string kmer = kcounter.get_kmer_string(path[i]);
		seq += kmer.substr(kmer.length()-1, 1);
		
		cov_counter.push_back( kcounter.get_kmer_count(path[i]) );
	}
	
	return(seq);
}

vector<kmer_int_type_t>  IRKE::build_inchworm_contig_from_seed(kmer_int_type_t kmer, KmerCounter &kcounter,
                                                               float min_connectivity, unsigned int &total_counts,
                                                               bool)
{

    unsigned int kmer_count = kcounter.get_kmer_count(kmer);

    /* Extend to the right */

    unsigned int kmer_length = kcounter.get_kmer_length();

    Kmer_visitor visitor(kmer_length, DOUBLE_STRANDED_MODE);
    Path_n_count_pair selected_path_n_pair_forward = inchworm(kcounter, 'F', kmer, visitor, min_connectivity);

    visitor.clear();
    // add selected path to visitor

    vector<kmer_int_type_t> &forward_path = selected_path_n_pair_forward.first;

    if (IRKE_COMMON::MONITOR >= 2) {
        cerr << "Forward path contains: " << forward_path.size() << " kmers. " << endl;
    }


    for (unsigned int i = 0; i < forward_path.size(); i++) {
        kmer_int_type_t kmer = forward_path[i];
        visitor.add(kmer);

        if (IRKE_COMMON::MONITOR >= 2) {
            cerr << "\tForward path kmer: " << kcounter.get_kmer_string(kmer) << endl;
        }

    }


    /* Extend to the left */
    visitor.erase(kmer); // reset the seed

    Path_n_count_pair selected_path_n_pair_reverse = inchworm(kcounter, 'R', kmer, visitor, min_connectivity);
    if (IRKE_COMMON::MONITOR >= 2) {
        vector<kmer_int_type_t> &reverse_path = selected_path_n_pair_reverse.first;
        cerr << "Reverse path contains: " << reverse_path.size() << " kmers. " << endl;
        for (unsigned int i = 0; i < reverse_path.size(); i++) {
            cerr << "\tReverse path kmer: " << kcounter.get_kmer_string(reverse_path[i]) << endl;
        }
    }


    total_counts = selected_path_n_pair_forward.second + selected_path_n_pair_reverse.second + kmer_count;

    vector<kmer_int_type_t> &reverse_path = selected_path_n_pair_reverse.first;

    vector<kmer_int_type_t> joined_path = _join_forward_n_reverse_paths(reverse_path, kmer, forward_path);

    return (joined_path);
}

bool IRKE::exceeds_min_connectivity (KmerCounter& kcounter, string kmerA, string kmerB, float min_connectivity) {

	kmer_int_type_t valA = kmer_to_intval(kmerA);
	kmer_int_type_t valB = kmer_to_intval(kmerB);

	Kmer_Occurence_Pair pairA(valA, kcounter.get_kmer_count(valA));
	Kmer_Occurence_Pair pairB(valB, kcounter.get_kmer_count(valB));

	return exceeds_min_connectivity(kcounter, pairA, pairB, min_connectivity);

}

vector<unsigned int> compute_kmer_coverage(string& sequence, KmerCounter& kcounter) {
    if(IRKE_COMMON::MONITOR) {
        cerr << "processing sequence: " << sequence << endl;
    }
    if (sequence.length() < KMER_SIZE)
    {
        // Can't rely on length() - KMER_SIZE for this as length is unsigned
        cerr << "Sequence: " << sequence << "is smaller than " << KMER_SIZE << " base pairs, skipping" << endl;
	return vector<unsigned int>();
    }

    vector<unsigned int> coverage;
    for (size_t i = 0; i <= sequence.length() - KMER_SIZE; i++) {
        // cerr << "i: " << i << ", <= " << sequence.length() - KMER_SIZE << endl;
        string kmer = sequence.substr(i, KMER_SIZE);
        if(IRKE_COMMON::MONITOR >= 2) {
            for (size_t j = 0; j <= i; j++) {
                cerr << " ";
            }
            cerr << kmer << endl;
        }
        unsigned int kmer_count = 0;
        if(!contains_non_gatc(kmer)) {
            kmer_count = kcounter.get_kmer_count(kmer);
        }

        // Note, in the jellyfish run, we restrain it to min kmer coverage of 2. 
        // If we don't find a kmer catalogued, it must have a kmer count of 1.
        if (kmer_count < 1) {
            kmer_count = 1;
        }
        
        coverage.push_back(kmer_count);
    }
    return(coverage);
}

Path_n_count_pair IRKE::inchworm (KmerCounter& kcounter, char direction, kmer_int_type_t kmer, Kmer_visitor& visitor, float min_connectivity) {
	
	// cout << "inchworm" << endl;
	
	Path_n_count_pair entire_path;
	
	unsigned int inchworm_round = 0;
	
	unsigned long num_total_kmers = kcounter.size();
	
	Kmer_visitor eliminator(kcounter.get_kmer_length(), DOUBLE_STRANDED_MODE);
	
	while (true) {
		
		inchworm_round++;
		eliminator.clear();
		
		if (inchworm_round > num_total_kmers) {
			throw(string ("Error, inchworm rounds have exceeded the number of possible seed kmers"));
		}
		
		if (IRKE_COMMON::MONITOR >= 3) {
			cerr << endl << "Inchworm round(" << string(1,direction) << "): " << inchworm_round << " searching kmer: " << kmer << endl;
			string kmer_str = kcounter.get_kmer_string(kmer);
			cerr << kcounter.describe_kmer(kmer_str) << endl;
		}
		
		visitor.erase(kmer); // seed kmer must be not visited already.
		
		Kmer_Occurence_Pair kmer_pair(kmer, kcounter.get_kmer_count(kmer));
		Path_n_count_pair best_path = inchworm_step(kcounter, direction, kmer_pair, visitor, eliminator, inchworm_round, 0, min_connectivity, MAX_RECURSION);
		
		if (best_path.second > 0) {
			// append info to entire path in reverse order, so starts just after seed kmer
			vector<kmer_int_type_t>& kmer_list = best_path.first;
			
			unsigned int num_kmers = kmer_list.size();
			int first_index = num_kmers - 1;
			int last_index = 0;
			if (CRAWL) {
				last_index = first_index - CRAWL_LENGTH + 1;
				if (last_index < 0) {
					last_index = 0;
				}
			}
			
			for (int i = first_index; i >= last_index; i--) {
				kmer_int_type_t kmer_extend = kmer_list[i];
				entire_path.first.push_back(kmer_extend);
				visitor.add(kmer_extend);
				entire_path.second += kcounter.get_kmer_count(kmer_extend);

			}
			
			kmer = entire_path.first[ entire_path.first.size() -1 ];
			
		}
		else {
			// no extension possible
			break;
		}
	}
	
	if (IRKE_COMMON::MONITOR >= 3) 
		cerr << endl;
	
	
	return(entire_path);
}

//.........这里部分代码省略.........
		visitor.clear();
		// add selected path to visitor
		
		vector<kmer_int_type_t>& forward_path = selected_path_n_pair_forward.first;
		if (IRKE_COMMON::MONITOR >= 2) {
            cerr << "Forward path contains: " << forward_path.size() << " kmers. " << endl;
        }


        for (unsigned int i = 0; i < forward_path.size(); i++) {
			kmer_int_type_t kmer = forward_path[i];
			visitor.add(kmer);
            
            if (IRKE_COMMON::MONITOR >= 2) {
                cerr << "\tForward path kmer: " << kcounter.get_kmer_string(kmer) << endl;
            }
            
		}
		
		
		/* Extend to the left */ 
		visitor.erase(kmer); // reset the seed
		
		Path_n_count_pair selected_path_n_pair_reverse = inchworm(kcounter, 'R', kmer, visitor, min_connectivity);
        if (IRKE_COMMON::MONITOR >= 2) {
            vector<kmer_int_type_t>& reverse_path = selected_path_n_pair_reverse.first;
            cerr << "Reverse path contains: " << reverse_path.size() << " kmers. " << endl;
            for (unsigned int i = 0; i < reverse_path.size(); i++) {
                cerr  << "\tReverse path kmer: " << kcounter.get_kmer_string(reverse_path[i]) << endl; 
            }
        }
        
		
		unsigned int total_counts = selected_path_n_pair_forward.second + selected_path_n_pair_reverse.second + kcounter.get_kmer_count(kmer); 
		
		vector<kmer_int_type_t>& reverse_path = selected_path_n_pair_reverse.first;
		
		vector<kmer_int_type_t> joined_path = _join_forward_n_reverse_paths(reverse_path, kmer, forward_path);
		
		// report sequence reconstructed from path.
		
		vector<unsigned int> assembly_base_coverage;
		string sequence = reconstruct_path_sequence(kcounter, joined_path, assembly_base_coverage);
		
		unsigned int avg_cov =  static_cast<unsigned int> ( (float)total_counts/(sequence.length()-kcounter.get_kmer_length() +1) + 0.5);
		
		/*
		  cout << "Inchworm-reconstructed sequence, length: " << sequence.length() 
		  << ", avgCov: " << avg_cov
		  << " " << sequence << endl;
		*/
		
		
		
		if (sequence.length() >= MIN_ASSEMBLY_LENGTH && avg_cov >= MIN_ASSEMBLY_COVERAGE) {
			
			INCHWORM_ASSEMBLY_COUNTER++;
			
			stringstream headerstream;
			
			
			headerstream << ">a" << INCHWORM_ASSEMBLY_COUNTER << ";" << avg_cov 
						 << " K: " << kmer_length
						 << " length: " << sequence.length();
			
			string header = headerstream.str();

Path_n_count_pair IRKE::inchworm(KmerCounter &kcounter,
                                 char direction,
                                 kmer_int_type_t kmer,
                                 Kmer_visitor &visitor,
                                 float min_connectivity)
{

    // cout << "inchworm" << endl;

    Path_n_count_pair entire_path;
    entire_path.second = 0; // init cumulative path coverage

    unsigned int inchworm_round = 0;

    unsigned long num_total_kmers = kcounter.size();

    Kmer_visitor eliminator(kcounter.get_kmer_length(), DOUBLE_STRANDED_MODE);

    while (true) {


        if (IRKE_COMMON::__DEVEL_rand_fracture) {

            // terminate extension with probability of __DEVEL_rand_fracture_prob

            float prob_to_fracture = rand() / (float) RAND_MAX;
            //cerr << "prob: " << prob_to_fracture << endl;

            if (prob_to_fracture <= IRKE_COMMON::__DEVEL_rand_fracture_prob) {

                // cerr << "Fracturing at iworm round: " << inchworm_round << " given P: " << prob_to_fracture << endl;

                return (entire_path);
            }
        }

        inchworm_round++;
        eliminator.clear();

        if (inchworm_round > num_total_kmers) {
            throw (string("Error, inchworm rounds have exceeded the number of possible seed kmers"));
        }

        if (IRKE_COMMON::MONITOR >= 3) {
            cerr << endl << "Inchworm round(" << string(1, direction) << "): " << inchworm_round << " searching kmer: "
                << kmer << endl;
            string kmer_str = kcounter.get_kmer_string(kmer);
            cerr << kcounter.describe_kmer(kmer_str) << endl;
        }

        visitor.erase(kmer); // seed kmer must be not visited already.

        Kmer_Occurence_Pair kmer_pair(kmer, kcounter.get_kmer_count(kmer));
        Path_n_count_pair best_path = inchworm_step(kcounter,
                                                    direction,
                                                    kmer_pair,
                                                    visitor,
                                                    eliminator,
                                                    inchworm_round,
                                                    0,
                                                    min_connectivity,
                                                    MAX_RECURSION);

        vector<kmer_int_type_t> &kmer_list = best_path.first;
        unsigned int num_kmers = kmer_list.size();

        if ((IRKE_COMMON::__DEVEL_zero_kmer_on_use && num_kmers >= 1) || best_path.second > 0) {
            // append info to entire path in reverse order, so starts just after seed kmer

            int first_index = num_kmers - 1;
            int last_index = 0;
            if (CRAWL) {
                last_index = first_index - CRAWL_LENGTH + 1;
                if (last_index < 0) {
                    last_index = 0;
                }
            }

            for (int i = first_index; i >= last_index; i--) {
                kmer_int_type_t kmer_extend = kmer_list[i];
                entire_path.first.push_back(kmer_extend);
                visitor.add(kmer_extend);
                //entire_path.second += kcounter.get_kmer_count(kmer_extend);

                // selected here, zero out:


                if (IRKE_COMMON::__DEVEL_zero_kmer_on_use) {
                    kcounter.clear_kmer(kmer_extend);
                }

            }

            kmer = entire_path.first[entire_path.first.size() - 1];

            entire_path.second += best_path.second;

        }
        else {
            // no extension possible
//.........这里部分代码省略.........

//.........这里部分代码省略.........
        cerr << "Done opening file. " << itmp.tmp_filename << endl;
    }


    //-------------------
    // Build contigs.
    //-------------------

#pragma omp parallel for private (myTid) schedule (dynamic, 1000)
    for (unsigned int i = 0; i < kmers.size(); i++) {

        // cerr << "round: " << i << endl;

        myTid = omp_get_thread_num();

        unsigned long kmer_counter_size = kcounter.size();
        if (kmer_counter_size > init_size) {

            // string s = "after.kmers";
            // kcounter.dump_kmers_to_file(s);

            stringstream error;
            error << stacktrace() << "Error, Kcounter size has grown from " << init_size
                << " to " << kmer_counter_size << endl;
            throw (error.str());
        }


        //kmer_int_type_t kmer = kmers[i]->first;
        //unsigned int kmer_count = kmers[i]->second;

        kmer_int_type_t kmer = kmers[i].first;
        // unsigned int kmer_count = kmers[i].second;  // NO!!!  Use for sorting, but likely zeroed out in the hashtable after contig construction
        unsigned int kmer_count = kcounter.get_kmer_count(kmer);

        if (!is_good_seed_kmer(kmer, kmer_count, kmer_length, min_connectivity)) {
            continue;
        }

        // cout << "SEED kmer: " << kcounter.get_kmer_string(kmer) << ", count: " << kmer_count << endl;

        if (IRKE_COMMON::MONITOR >= 2) {
            cerr << "SEED kmer: " << kcounter.get_kmer_string(kmer) << ", count: " << kmer_count << endl;
        }

        if (IRKE_COMMON::MONITOR >= 2) {
#pragma omp critical
            cerr << "Seed for thread: " << myTid << " is " << kcounter.get_kmer_string(kmer) << " with count: "
                << kmer_count << endl;
        }

        unsigned int total_counts;
        vector<kmer_int_type_t> joined_path =
            build_inchworm_contig_from_seed(kmer, kcounter, min_connectivity, total_counts, PARALLEL_IWORM);

        if (PARALLEL_IWORM && TWO_PHASE) {
            // get a new seed based on the draft contig
            // choose the 'good' kmer with highest abundance

            kmer_int_type_t new_seed = extract_best_seed(joined_path, kcounter, min_connectivity);

            if (kcounter.get_kmer_count(new_seed) == 0) {
                continue; // must have been zapped by another thread
            }

            joined_path =