https://github.com/bcgsc/ntCard
Raw File
Tip revision: e8d401c263cbfbceb868263157ba2cab277084ca authored by Johnathan Wong on 24 January 2020, 23:25:27 UTC
release 1.2.0 (#37)
Tip revision: e8d401c
ntcard.cpp
#include "Uncompress.h"
#include "vendor/ntHash/ntHashIterator.hpp"
#include "vendor/ntHash/stHashIterator.hpp"

#include <cassert>
#include <cmath>
#include <cstdlib>
#include <fstream>
#include <getopt.h>
#include <iomanip>
#include <iostream>
#include <sstream>
#include <string>
#include <vector>

#ifdef _OPENMP
#include <omp.h>
#endif

#define PROGRAM "ntCard"

static const char VERSION_MESSAGE[] =
    PROGRAM " 1.1.0 \n"
            "Written by Hamid Mohamadi.\n"
            "Copyright 2018 Hamid Mohamadi, Licensed under MIT License\n";

static const char USAGE_MESSAGE[] =
    "Usage: " PROGRAM " [OPTION]... FILE(S)...\n"
    "Estimates k-mer coverage histogram in FILE(S).\n\n"
    "Acceptable file formats: fastq, fasta, sam, bam and in compressed formats gz, bz, zip, xz.\n"
    "A list of files containing file names in each row can be passed with @ prefix.\n"

    "\n"
    " Options:\n"
    "\n"
    "  -t, --threads=N	use N parallel threads [1] (N>=2 should be used when input files are >=2)\n"
    "  -k, --kmer=N	the length of kmer \n"
    "  -g, --gap=N	the length of gap in the gap seed [0]. g mod 2 must equal k mod 2 unless g == "
    "0\n"
    "           	-g does not support multiple k currently.\n"
    "  -c, --cov=N	the maximum coverage of kmer in output [1000]\n"
    "  -p, --pref=STRING    the prefix for output file name(s)\n"
    "  -o, --output=STRING	the name for output file name (used when output should be a single "
    "file)\n"
    "      --help	display this help and exit\n"
    "      --version	output version information and exit\n"
    "\n"
    "Report bugs to https://github.com/bcgsc/ntCard/issues\n";

using namespace std;

namespace opt {
unsigned nThrd = 1;
unsigned kmLen = 64;
unsigned gap = 0;
size_t rBuck;
unsigned rBits = 27;
unsigned sBits = 11;
unsigned sMask;
unsigned covMax = 1000;
size_t nSamp = 2;
size_t nK = 0;
string prefix;
string output;
bool samH = true;
std::vector<std::vector<unsigned> > seedSet;
}

static const char shortopts[] = "t:s:r:k:c:l:p:f:o:g:";

enum
{
	OPT_HELP = 1,
	OPT_VERSION
};

static const struct option longopts[] = { { "threads", required_argument, NULL, 't' },
	                                      { "kmer", required_argument, NULL, 'k' },
	                                      { "gap", required_argument, NULL, 'g' },
	                                      { "cov", required_argument, NULL, 'c' },
	                                      { "rbit", required_argument, NULL, 'r' },
	                                      { "sbit", required_argument, NULL, 's' },
	                                      { "output", required_argument, NULL, 'o' },
	                                      { "pref", required_argument, NULL, 'p' },
	                                      { "help", no_argument, NULL, OPT_HELP },
	                                      { "version", no_argument, NULL, OPT_VERSION },
	                                      { NULL, 0, NULL, 0 } };

size_t
getInf(const char* inFile)
{
	std::ifstream in(inFile, std::ifstream::ate | std::ifstream::binary);
	return in.tellg();
}

bool
isNumber(const string& seq)
{
	string::const_iterator itr = seq.begin();
	while (itr != seq.end() && isdigit(*itr))
		++itr;
	return (itr == seq.end() && !seq.empty());
}

unsigned
getftype(std::ifstream& in, std::string& samSeq)
{
	std::string hseq;
	getline(in, hseq);
	if (hseq[0] == '>') {
		return 1;
	}
	if (hseq[0] == '@') {
		if ((hseq[1] == 'H' && hseq[2] == 'D') || (hseq[1] == 'S' && hseq[2] == 'Q') ||
		    (hseq[1] == 'R' && hseq[2] == 'G') || (hseq[1] == 'P' && hseq[2] == 'G') ||
		    (hseq[1] == 'C' && hseq[2] == 'O')) {
			return 2;
		} else
			return 0;
	}
	std::istringstream alnSec(hseq);
	std::string s1, s2, s3, s4, s5, s6, s7, s8, s9, s10, s11;
	alnSec >> s1 >> s2 >> s3 >> s4 >> s5 >> s6 >> s7 >> s8 >> s9 >> s10 >> s11;
	if (isNumber(s2) && isNumber(s5)) {
		opt::samH = false;
		samSeq = hseq;
		return 2;
	}
	return 3;
}

inline void
ntComp(const uint64_t hVal, uint16_t* t_Counter)
{
	uint64_t indBit = opt::nSamp;
	if (hVal >> (63 - opt::sBits) == 1)
		indBit = 0;
	if (hVal >> (64 - opt::sBits) == opt::sMask)
		indBit = 1;
	if (indBit < opt::nSamp) {
		size_t shVal = hVal & (opt::rBuck - 1);
#pragma omp atomic
		++t_Counter[indBit * opt::rBuck + shVal];
	}
}

inline void
ntRead(const string& seq, const std::vector<unsigned>& kList, uint16_t* t_Counter, size_t totKmer[])
{
	for (unsigned k = 0; k < kList.size(); k++) {
		ntHashIterator itr(seq, 1, kList[k]);
		while (itr != itr.end()) {
			ntComp((*itr)[0], t_Counter + k * opt::nSamp * opt::rBuck);
			++itr;
			++totKmer[k];
		}
	}
}

inline void
stRead(const string& seq, const std::vector<unsigned>& kList, uint16_t* t_Counter, size_t totKmer[])
{
	for (unsigned k = 0; k < kList.size(); k++) {
		stHashIterator itr(seq, opt::seedSet, 1, 1, kList[k]);
		while (itr != itr.end()) {
			ntComp((const uint64_t)(*itr)[0], t_Counter + k * opt::nSamp * opt::rBuck);
			++itr;
			++totKmer[k];
		}
	}
}

void
getEfq(std::ifstream& in, const std::vector<unsigned>& kList, uint16_t* t_Counter, size_t totKmer[])
{
	bool good = true;
	for (string seq, hseq; good;) {
		good = static_cast<bool>(getline(in, seq));
		good = static_cast<bool>(getline(in, hseq));
		good = static_cast<bool>(getline(in, hseq));
		if (good && opt::gap == 0) {
			ntRead(seq, kList, t_Counter, totKmer);
		}
		if (good && opt::gap != 0) {
			stRead(seq, kList, t_Counter, totKmer);
		}
		good = static_cast<bool>(getline(in, hseq));
	}
}

void
getEfa(std::ifstream& in, const std::vector<unsigned>& kList, uint16_t* t_Counter, size_t totKmer[])
{
	bool good = true;
	for (string seq, hseq; good;) {
		string line;
		good = static_cast<bool>(getline(in, seq));
		while (good && seq[0] != '>') {
			line += seq;
			good = static_cast<bool>(getline(in, seq));
		}
		if (opt::gap == 0) {
			ntRead(seq, kList, t_Counter, totKmer);
		} else {
			stRead(seq, kList, t_Counter, totKmer);
		}
	}
}

void
getEsm(
    std::ifstream& in,
    const std::vector<unsigned>& kList,
    const std::string& samSeq,
    uint16_t* t_Counter,
    size_t totKmer[])
{
	std::string samLine, seq;
	std::string s1, s2, s3, s4, s5, s6, s7, s8, s9, s11;
	if (opt::samH) {
		while (getline(in, samLine))
			if (samLine[0] != '@')
				break;
	} else
		samLine = samSeq;
	do {
		std::istringstream iss(samLine);
		iss >> s1 >> s2 >> s3 >> s4 >> s5 >> s6 >> s7 >> s8 >> s9 >> seq >> s11;
		if (opt::gap == 0) {
			ntRead(seq, kList, t_Counter, totKmer);
		} else {
			stRead(seq, kList, t_Counter, totKmer);
		}
	} while (getline(in, samLine));
}

void
compEst(const uint16_t* t_Counter, double& F0Mean, double fMean[])
{
	unsigned p[opt::nSamp][65536];
	for (size_t i = 0; i < opt::nSamp; i++)
		for (size_t j = 0; j < 65536; j++)
			p[i][j] = 0;

	for (size_t i = 0; i < opt::nSamp; i++)
		for (size_t j = 0; j < opt::rBuck; j++)
			++p[i][t_Counter[i * opt::rBuck + j]];

	double pMean[65536];
	for (size_t i = 0; i < 65536; i++)
		pMean[i] = 0.0;
	for (size_t i = 0; i < 65536; i++) {
		for (size_t j = 0; j < opt::nSamp; j++)
			pMean[i] += p[j][i];
		pMean[i] /= 1.0 * opt::nSamp;
	}

	F0Mean = (ssize_t)(
	    (opt::rBits * log(2) - log(pMean[0])) * 1.0 * ((size_t)1 << (opt::sBits + opt::rBits)));
	for (size_t i = 0; i < 65536; i++)
		fMean[i] = 0;
	if (pMean[0] == 0) {
		return;
	}
	fMean[1] = -1.0 * pMean[1] / (pMean[0] * (log(pMean[0]) - opt::rBits * log(2)));
	for (size_t i = 2; i < 65536; i++) {
		double sum = 0.0;
		for (size_t j = 1; j < i; j++)
			sum += j * pMean[i - j] * fMean[j];
		fMean[i] = -1.0 * pMean[i] / (pMean[0] * (log(pMean[0]) - opt::rBits * log(2))) -
		           sum / (i * pMean[0]);
	}
	for (size_t i = 1; i < 65536; i++)
		fMean[i] = abs((ssize_t)(fMean[i] * F0Mean));
}

void
outDefault(const std::vector<unsigned>& kList, const size_t totalKmers[], const uint16_t* t_Counter)
{
	std::ofstream histFiles[opt::nK];
	for (unsigned k = 0; k < opt::nK; k++) {
		std::stringstream hstm;
		hstm << opt::prefix << "_k" << kList[k] << ".hist";
		histFiles[k].open((hstm.str()).c_str());
	}
#pragma omp parallel for schedule(dynamic)
	for (unsigned k = 0; k < kList.size(); k++) {
		double F0Mean = 0.0;
		double fMean[65536];
		compEst(t_Counter + k * opt::nSamp * opt::rBuck, F0Mean, fMean);
		histFiles[k] << "F1\t" << totalKmers[k] << "\n";
		histFiles[k] << "F0\t" << (size_t)F0Mean << "\n";
		for (size_t i = 1; i <= opt::covMax; i++)
			histFiles[k] << i << "\t" << (size_t)fMean[i] << "\n";
	}
	for (unsigned k = 0; k < opt::nK; k++)
		histFiles[k].close();
}

void
outCompact(const std::vector<unsigned>& kList, const size_t totalKmers[], const uint16_t* t_Counter)
{
	ofstream histFile(opt::output.c_str());
	histFile << "k\tf\tn\n";
	for (unsigned k = 0; k < kList.size(); k++) {
		double F0Mean = 0.0;
		double fMean[65536];
		compEst(t_Counter + k * opt::nSamp * opt::rBuck, F0Mean, fMean);
		std::cerr << "k=" << kList[k] << "\tF1\t" << totalKmers[k] << "\n";
		std::cerr << "k=" << kList[k] << "\tF0\t" << (size_t)F0Mean << "\n";
		for (size_t i = 1; i <= opt::covMax; i++)
			histFile << kList[k] << "\t" << i << "\t" << (size_t)fMean[i] << "\n";
	}
	histFile.close();
}

int
main(int argc, char** argv)
{

	double sTime = omp_get_wtime();

	vector<unsigned> kList;
	bool die = false;
	for (int c; (c = getopt_long(argc, argv, shortopts, longopts, NULL)) != -1;) {
		std::istringstream arg(optarg != NULL ? optarg : "");
		switch (c) {
		case '?':
			die = true;
			break;
		case 't':
			arg >> opt::nThrd;
			break;
		case 's':
			arg >> opt::sBits;
			break;
		case 'r':
			arg >> opt::rBits;
			break;
		case 'c':
			arg >> opt::covMax;
			if (opt::covMax > 65535)
				opt::covMax = 65535;
			break;
		case 'p':
			arg >> opt::prefix;
			break;
		case 'o':
			arg >> opt::output;
			break;
		case 'g':
			arg >> opt::gap;
			break;
		case 'k': {
			std::string token;
			while (getline(arg, token, ',')) {
				unsigned myK;
				std::stringstream ss(token);
				ss >> myK;
				kList.push_back(myK);
				++opt::nK;
			}
			break;
		}
		case OPT_HELP:
			std::cerr << USAGE_MESSAGE;
			exit(EXIT_SUCCESS);
		case OPT_VERSION:
			std::cerr << VERSION_MESSAGE;
			exit(EXIT_SUCCESS);
		}
		if (optarg != NULL && !arg.eof()) {
			std::cerr << PROGRAM ": invalid option: `-" << (char)c << optarg << "'\n";
			exit(EXIT_FAILURE);
		}
	}
	if (argc - optind < 1) {
		std::cerr << PROGRAM ": missing arguments\n";
		die = true;
	}

	if (opt::gap != 0 && (opt::gap % 2 != kList[0] % 2)) {
		std::cerr << PROGRAM "Gap size and kmer must have the same modulus\n";
		die = true;
	}

	if (opt::nK == 0) {
		std::cerr << PROGRAM ": missing argument -k ... \n";
		die = true;
	}

	if (opt::prefix.empty() && opt::output.empty()) {
		std::cerr << PROGRAM ": missing argument -p/-o ... \n";
		die = true;
	}

	if (opt::gap != 0 && kList.size() != 1) {
		std::cerr << PROGRAM ": -g does not support multiple k currently.\n";
		die = true;
	}

	if (die) {
		std::cerr << "Try `" << PROGRAM << " --help' for more information.\n";
		exit(EXIT_FAILURE);
	}

	if (opt::gap != 0) {
		std::string gap(opt::gap, '0');
		std::string nonGap((kList[0] - opt::gap) / 2, '1');
		std::vector<std::string> seedString;
		seedString.push_back(nonGap + gap + nonGap);
		opt::seedSet = stHashIterator::parseSeed(seedString);
	}

	vector<string> inFiles;
	for (int i = optind; i < argc; ++i) {
		string file(argv[i]);
		if (file[0] == '@') {
			string inName;
			ifstream inList(file.substr(1, file.length()).c_str());
			while (getline(inList, inName))
				inFiles.push_back(inName);
		} else
			inFiles.push_back(file);
	}

	size_t totalSize = 0;
	for (unsigned file_i = 0; file_i < inFiles.size(); ++file_i)
		totalSize += getInf(inFiles[file_i].c_str());
	if (totalSize < 50000000000)
		opt::sBits = 7;

	size_t totalKmers[kList.size()];
	for (unsigned k = 0; k < kList.size(); k++)
		totalKmers[k] = 0;

	opt::rBuck = ((size_t)1) << opt::rBits;
	opt::sMask = (((size_t)1) << (opt::sBits - 1)) - 1;
	uint16_t* t_Counter = new uint16_t[opt::nK * opt::nSamp * opt::rBuck]();

#ifdef _OPENMP
	omp_set_num_threads(opt::nThrd);
#endif

#pragma omp parallel for schedule(dynamic)
	for (unsigned file_i = 0; file_i < inFiles.size(); ++file_i) {
		size_t totKmer[kList.size()];
		for (unsigned k = 0; k < kList.size(); k++)
			totKmer[k] = 0;
		std::ifstream in(inFiles[file_i].c_str());
		std::string samSeq;
		unsigned ftype = getftype(in, samSeq);
		if (ftype == 0)
			getEfq(in, kList, t_Counter, totKmer);
		else if (ftype == 1)
			getEfa(in, kList, t_Counter, totKmer);
		else if (ftype == 2)
			getEsm(in, kList, samSeq, t_Counter, totKmer);
		else {
			std::cerr << "Error in reading file: " << inFiles[file_i] << std::endl;
			exit(EXIT_FAILURE);
		}
		in.close();
		for (unsigned k = 0; k < kList.size(); k++)
#pragma omp atomic
			totalKmers[k] += totKmer[k];
	}

	if (opt::output.empty())
		outDefault(kList, totalKmers, t_Counter);
	else
		outCompact(kList, totalKmers, t_Counter);

	delete[] t_Counter;

	std::cerr << "Runtime(sec): " << setprecision(4) << fixed << omp_get_wtime() - sTime << "\n";
	return 0;
}
back to top