#ifndef MOTHUR_H
#define MOTHUR_H
-using namespace std;
/*
/* This file contains all the standard incudes we use in the project as well as some common utilities. */
+//#include <cstddef>
//io libraries
#include <iostream>
#include <iomanip>
#include <fstream>
#include <sstream>
+#include <signal.h>
+
//exception
#include <stdexcept>
#include <map>
#include <string>
#include <list>
+#include <string.h>
//math
#include <cmath>
#include <math.h>
#include <algorithm>
+#include <numeric>
+
+//misc
+#include <cerrno>
+#include <ctime>
+#include <limits>
+
+#ifdef USE_MPI
+ #include "mpi.h"
+#endif
+/***********************************************************************/
+
+#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
+ #include <sys/wait.h>
+ #include <sys/time.h>
+ #include <sys/resource.h>
+ #include <sys/types.h>
+ #include <sys/stat.h>
+ #include <unistd.h>
+
+ #ifdef USE_READLINE
+ #include <readline/readline.h>
+ #include <readline/history.h>
+ #endif
+
+#else
+ #include <conio.h> //allows unbuffered screen capture from stdin
+ #include <direct.h> //get cwd
+ #include <windows.h>
+ #include <psapi.h>
+ #include <direct.h>
+ #include <tchar.h>
+
+#endif
+
+using namespace std;
+
+#define exp(x) (exp((double) x))
+#define sqrt(x) (sqrt((double) x))
+#define log10(x) (log10((double) x))
+#define log2(x) (log10(x)/log10(2))
+#define isnan(x) ((x) != (x))
+#define isinf(x) (fabs(x) == std::numeric_limits<double>::infinity())
-typedef unsigned long long ull;
+
+typedef unsigned long ull;
+typedef unsigned short intDist;
struct IntNode {
int lvalue;
int rcoef;
IntNode* left;
IntNode* right;
-};
-/***********************************************************************/
-
-// snagged from http://www.parashift.com/c++-faq-lite/misc-technical-issues.html#faq-39.2
-// works for now, but there should be a way to do it without killing the whole program
-
-class BadConversion : public runtime_error {
-public:
- BadConversion(const string& s) : runtime_error(s){ }
+ IntNode(int lv, int rv, IntNode* l, IntNode* r) : lvalue(lv), rvalue(rv), left(l), right(r) {};
+ IntNode() {};
};
-//**********************************************************************************************************************
-
-template<typename T>
-inline void convert(const string& s, T& x, bool failIfLeftoverChars = true){
- istringstream i(s);
- char c;
- if (!(i >> x) || (failIfLeftoverChars && i.get(c)))
- throw BadConversion(s);
-}
-//**********************************************************************************************************************
-
-template<typename T>
-inline bool convertTest(const string& s, T& x, bool failIfLeftoverChars = true){
- istringstream i(s);
- char c;
- if (!(i >> x) || (failIfLeftoverChars && i.get(c)))
- {
- cout << "'" << s << "' is unable to be converted into an integer.\n";
- return false;
- }
- return true;
-}
-
-//**********************************************************************************************************************
-
-template<typename T>
-string toString(const T&x){
- stringstream output;
- output << x;
- return output.str();
-}
-
-//**********************************************************************************************************************
+struct ThreadNode {
+ int* pid;
+ IntNode* left;
+ IntNode* right;
+};
-template<typename T>
-string toHex(const T&x){
- stringstream output;
+struct diffPair {
+ float prob;
+ float reverseProb;
- output << hex << x;
-
- return output.str();
-}
-//**********************************************************************************************************************
+ diffPair() {
+ prob = 0; reverseProb = 0;
+ }
+ diffPair(float p, float rp) {
+ prob = p;
+ reverseProb = rp;
+ }
+};
-template<typename T>
-string toString(const T&x, int i){
- stringstream output;
+/**********************************************************/
+struct CommonHeader {
+ unsigned int magicNumber;
+ string version;
+ unsigned long long indexOffset;
+ unsigned int indexLength;
+ unsigned int numReads;
+ unsigned short headerLength;
+ unsigned short keyLength;
+ unsigned short numFlowsPerRead;
+ int flogramFormatCode;
+ string flowChars; //length depends on number flow reads
+ string keySequence; //length depends on key length
- output.precision(i);
- output << fixed << x;
-
- return output.str();
-}
-
-
-/***********************************************************************/
-
-inline void gobble(istream& f){
+ CommonHeader(){ magicNumber=0; indexOffset=0; indexLength=0; numReads=0; headerLength=0; keyLength=0; numFlowsPerRead=0; flogramFormatCode='s'; }
+ ~CommonHeader() { }
+};
+/**********************************************************/
+struct Header {
+ unsigned short headerLength;
+ unsigned short nameLength;
+ unsigned int numBases;
+ unsigned short clipQualLeft;
+ unsigned short clipQualRight;
+ unsigned short clipAdapterLeft;
+ unsigned short clipAdapterRight;
+ string name; //length depends on nameLength
+ string timestamp;
+ string region;
+ string xy;
- char d;
- while(isspace(d=f.get())) {;}
- f.putback(d);
+ Header() { headerLength=0; nameLength=0; numBases=0; clipQualLeft=0; clipQualRight=0; clipAdapterLeft=0; clipAdapterRight=0; }
+ ~Header() { }
+};
+/**********************************************************/
+struct seqRead {
+ vector<unsigned short> flowgram;
+ vector<unsigned int> flowIndex;
+ string bases;
+ vector<unsigned int> qualScores;
-}
+ seqRead() { }
+ ~seqRead() { }
+};
/***********************************************************************/
-
-inline float roundDist(float dist, int precision){
-
- return int(dist * precision + 0.5)/float(precision);
-
-}
-
+struct PDistCell{
+ ull index;
+ float dist;
+ PDistCell() : index(0), dist(0) {};
+ PDistCell(ull c, float d) : index(c), dist(d) {}
+};
/***********************************************************************/
-
-inline int getNumNames(string names){
-
- int count = 0;
-
- if(names != ""){
- count = 1;
- for(int i=0;i<names.size();i++){
- if(names[i] == ','){
- count++;
- }
- }
- }
+struct consTax{
+ string name;
+ string taxonomy;
+ int abundance;
+ consTax() : name(""), taxonomy("unknown"), abundance(0) {};
+ consTax(string n, string t, int a) : name(n), taxonomy(t), abundance(a) {}
+};
+/***********************************************************************/
+struct consTax2{
+ string taxonomy;
+ int abundance;
+ consTax2() : taxonomy("unknown"), abundance(0) {};
+ consTax2(string t, int a) : taxonomy(t), abundance(a) {}
+};
+/************************************************************/
+struct clusterNode {
+ int numSeq;
+ int parent;
+ int smallChild; //used to make linkTable work with list and rabund. represents bin number of this cluster node
+ clusterNode(int num, int par, int kid) : numSeq(num), parent(par), smallChild(kid) {};
+};
+/************************************************************/
+struct seqDist {
+ int seq1;
+ int seq2;
+ double dist;
+ seqDist() {}
+ seqDist(int s1, int s2, double d) : seq1(s1), seq2(s2), dist(d) {}
+ ~seqDist() {}
+};
+/************************************************************/
+struct distlinePair {
+ int start;
+ int end;
- return count;
+};
+/************************************************************/
+struct seqPriorityNode {
+ int numIdentical;
+ string seq;
+ string name;
+ seqPriorityNode() {}
+ seqPriorityNode(int n, string s, string nm) : numIdentical(n), seq(s), name(nm) {}
+ ~seqPriorityNode() {}
+};
+/************************************************************/
+struct compGroup {
+ string group1;
+ string group2;
+ compGroup() {}
+ compGroup(string s, string nm) : group1(s), group2(nm) {}
+ string getCombo() { return group1+"-"+group2; }
+ ~compGroup() {}
+};
+/***************************************************************/
+struct spearmanRank {
+ string name;
+ float score;
+ spearmanRank(string n, float s) : name(n), score(s) {}
+};
+//***********************************************************************
+inline bool compareIndexes(PDistCell left, PDistCell right){
+ return (left.index > right.index);
}
-
-/**************************************************************************************************/
-
-inline vector<vector<double> > binomial(int maxOrder){
-
- vector<vector<double> > binomial(maxOrder+1);
-
- for(int i=0;i<=maxOrder;i++){
- binomial[i].resize(maxOrder+1);
- binomial[i][0]=1;
- binomial[0][i]=0;
- }
- binomial[0][0]=1;
-
- binomial[1][0]=1;
- binomial[1][1]=1;
-
- for(int i=2;i<=maxOrder;i++){
- binomial[1][i]=0;
- }
-
- for(int i=2;i<=maxOrder;i++){
- for(int j=1;j<=maxOrder;j++){
- if(i==j){ binomial[i][j]=1; }
- if(j>i) { binomial[i][j]=0; }
- else { binomial[i][j]=binomial[i-1][j-1]+binomial[i-1][j]; }
- }
- }
-
- return binomial;
+//********************************************************************************************************************
+inline bool compareSpearman(spearmanRank left, spearmanRank right){
+ return (left.score < right.score);
}
-
-/***********************************************************************/
-
-inline string getRootName(string longName){
+//********************************************************************************************************************
+inline double max(double left, double right){
+ if (left > right) { return left; }
+ else { return right; }
+}
+//********************************************************************************************************************
+inline double max(int left, double right){
+ double value = left;
+ if (left > right) { return value; }
+ else { return right; }
+}
+//********************************************************************************************************************
+inline double max(double left, int right){
+ double value = right;
+ if (left > value) { return left; }
+ else { return value; }
+}
+//********************************************************************************************************************
+//sorts highest to lowest
+inline bool compareSeqPriorityNodes(seqPriorityNode left, seqPriorityNode right){
+ if (left.numIdentical > right.numIdentical) {
+ return true;
+ }else if (left.numIdentical == right.numIdentical) {
+ if (left.seq > right.seq) { return true; }
+ else { return false; }
+ }
+ return false;
+}
- string rootName = longName;
-
- if(longName.find_last_of(".") != longName.npos){
- int pos = longName.find_last_of('.')+1;
- rootName = longName.substr(0, pos);
- }
-
- return rootName;
+/************************************************************/
+//sorts lowest to highest
+inline bool compareDistLinePairs(distlinePair left, distlinePair right){
+ return (left.end < right.end);
+}
+//********************************************************************************************************************
+//sorts lowest to highest
+inline bool compareSequenceDistance(seqDist left, seqDist right){
+ return (left.dist < right.dist);
+}
+//********************************************************************************************************************
+//returns sign of double
+inline double sign(double temp){
+ //find sign
+ if (temp > 0) { return 1.0; }
+ else if (temp < 0) { return -1.0; }
+ return 0;
}
/***********************************************************************/
-inline string getSimpleName(string longName){
-
- string simpleName = longName;
-
- if(longName.find_last_of("/") != longName.npos){
- int pos = longName.find_last_of('/')+1;
- simpleName = longName.substr(pos, longName.length());
- }
+// snagged from http://www.parashift.com/c++-faq-lite/misc-technical-issues.html#faq-39.2
+// works for now, but there should be a way to do it without killing the whole program
- return simpleName;
-}
-/***********************************************************************/
+class BadConversion : public runtime_error {
+public:
+ BadConversion(const string& s) : runtime_error(s){ }
+};
-inline int factorial(int num){
- int total = 1;
+//**********************************************************************************************************************
+template<typename T>
+void convert(const string& s, T& x, bool failIfLeftoverChars = true){
- for (int i = 1; i <= num; i++) {
- total *= i;
- }
+ istringstream i(s);
+ char c;
+ if (!(i >> x) || (failIfLeftoverChars && i.get(c)))
+ throw BadConversion(s);
- return total;
}
+//**********************************************************************************************************************
+template <typename T> int sgn(T val){ return (val > T(0)) - (val < T(0)); }
+//**********************************************************************************************************************
-/***********************************************************************/
-
-inline string getPathName(string longName){
-
- string rootPathName = longName;
+template<typename T>
+bool convertTestFloat(const string& s, T& x, bool failIfLeftoverChars = true){
+
+ istringstream i(s);
+ char c;
+ if (!(i >> x) || (failIfLeftoverChars && i.get(c)))
+ {
+ return false;
+ }
+ return true;
- if(longName.find_last_of("/") != longName.npos){
- int pos = longName.find_last_of('/')+1;
- rootPathName = longName.substr(0, pos);
- }
-
- return rootPathName;
}
-/***********************************************************************/
-
-inline int openInputFile(string fileName, ifstream& fileHandle){
+//**********************************************************************************************************************
- fileHandle.open(fileName.c_str());
- if(!fileHandle) {
- cerr << "Error: Could not open " << fileName << endl;
- return 1;
- }
- else {
- return 0;
- }
+template<typename T>
+bool convertTest(const string& s, T& x, bool failIfLeftoverChars = true){
+
+ istringstream i(s);
+ char c;
+ if (!(i >> x) || (failIfLeftoverChars && i.get(c)))
+ {
+ return false;
+ }
+ return true;
}
-
-/***********************************************************************/
-
-inline int openOutputFile(string fileName, ofstream& fileHandle){
+//**********************************************************************************************************************
+template<typename T>
+string toString(const T&x){
+
+ stringstream output;
+ output << x;
+ return output.str();
- fileHandle.open(fileName.c_str(), ios::trunc);
- if(!fileHandle) {
- cerr << "Error: Could not open " << fileName << endl;
- return 1;
- }
- else {
- return 0;
- }
-
}
-/***********************************************************************/
+//**********************************************************************************************************************
-//This function parses the estimator options and puts them in a vector
-inline void splitAtDash(string& estim, vector<string>& container) {
- try {
- string individual;
+template<typename T>
+string toHex(const T&x){
+
+ stringstream output;
- while (estim.find_first_of('-') != -1) {
- individual = estim.substr(0,estim.find_first_of('-'));
- if ((estim.find_first_of('-')+1) <= estim.length()) { //checks to make sure you don't have dash at end of string
- estim = estim.substr(estim.find_first_of('-')+1, estim.length());
- container.push_back(individual);
- }
- }
- //get last one
- container.push_back(estim);
- }
- catch(exception& e) {
- cout << "Standard Error: " << e.what() << " has occurred in the mothur class Function splitAtDash. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
- exit(1);
- }
- catch(...) {
- cout << "An unknown error has occurred in the mothur class function splitAtDash. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
- exit(1);
- }
+ output << hex << x;
+ return output.str();
+
}
+//**********************************************************************************************************************
-/***********************************************************************/
-//This function parses the label options and puts them in a set
-inline void splitAtDash(string& estim, set<string>& container) {
- try {
- string individual;
-
- while (estim.find_first_of('-') != -1) {
- individual = estim.substr(0,estim.find_first_of('-'));
- if ((estim.find_first_of('-')+1) <= estim.length()) { //checks to make sure you don't have dash at end of string
- estim = estim.substr(estim.find_first_of('-')+1, estim.length());
- container.insert(individual);
- }
- }
- //get last one
- container.insert(estim);
- }
- catch(exception& e) {
- cout << "Standard Error: " << e.what() << " has occurred in the mothur class Function splitAtDash. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
- exit(1);
- }
- catch(...) {
- cout << "An unknown error has occurred in the mothur class function splitAtDash. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
- exit(1);
- }
-
-}
-/***********************************************************************/
-//This function parses the line options and puts them in a set
-inline void splitAtDash(string& estim, set<int>& container) {
- try {
- string individual;
- int lineNum;
+template<typename T>
+string toString(const T&x, int i){
+
+ stringstream output;
- while (estim.find_first_of('-') != -1) {
- individual = estim.substr(0,estim.find_first_of('-'));
- if ((estim.find_first_of('-')+1) <= estim.length()) { //checks to make sure you don't have dash at end of string
- estim = estim.substr(estim.find_first_of('-')+1, estim.length());
- convert(individual, lineNum); //convert the string to int
- container.insert(lineNum);
- }
- }
- //get last one
- convert(estim, lineNum); //convert the string to int
- container.insert(lineNum);
- }
- catch(exception& e) {
- cout << "Standard Error: " << e.what() << " has occurred in the mothur class Function splitAtDash. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
- exit(1);
- }
- catch(...) {
- cout << "An unknown error has occurred in the mothur class function splitAtDash. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
- exit(1);
- }
-
-}
-/***********************************************************************/
-//This function parses the a string and puts peices in a vector
-inline void splitAtComma(string& estim, vector<string>& container) {
- try {
- string individual;
+ output.precision(i);
+ output << fixed << x;
- while (estim.find_first_of(',') != -1) {
- individual = estim.substr(0,estim.find_first_of(','));
- if ((estim.find_first_of(',')+1) <= estim.length()) { //checks to make sure you don't have comma at end of string
- estim = estim.substr(estim.find_first_of(',')+1, estim.length());
- container.push_back(individual);
- }
- }
- //get last one
- container.push_back(estim);
- }
- catch(exception& e) {
- cout << "Standard Error: " << e.what() << " has occurred in the mothur class Function splitAtComma. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
- exit(1);
- }
- catch(...) {
- cout << "An unknown error has occurred in the mothur class function splitAtComma. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
- exit(1);
- }
-}
-/***********************************************************************/
-
-//This function splits up the various option parameters
-inline void splitAtComma(string& prefix, string& suffix){
- try {
- prefix = suffix.substr(0,suffix.find_first_of(','));
- if ((suffix.find_first_of(',')+2) <= suffix.length()) { //checks to make sure you don't have comma at end of string
- suffix = suffix.substr(suffix.find_first_of(',')+1, suffix.length());
- string space = " ";
- while(suffix.at(0) == ' ')
- suffix = suffix.substr(1, suffix.length());
- }
-
- }
- catch(exception& e) {
- cout << "Standard Error: " << e.what() << " has occurred in the mothur class Function splitAtComma. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
- exit(1);
- }
- catch(...) {
- cout << "An unknown error has occurred in the mothur class function splitAtComma. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
- exit(1);
- }
-
+ return output.str();
+
}
-/***********************************************************************/
-
-//This function separates the key value from the option value i.e. dist=96_...
-inline void splitAtEquals(string& key, string& value){
- try {
- if(value.find_first_of('=') != -1){
- key = value.substr(0,value.find_first_of('='));
- if ((value.find_first_of('=')+1) <= value.length()) {
- value = value.substr(value.find_first_of('=')+1, value.length());
- }
- }else{
- key = value;
- value = 1;
- }
- }
- catch(exception& e) {
- cout << "Standard Error: " << e.what() << " has occurred in the mothur class Function splitAtEquals. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
- exit(1);
- }
- catch(...) {
- cout << "An unknown error has occurred in the mothur class function splitAtEquals. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
- exit(1);
- }
+//**********************************************************************************************************************
+template<class T>
+T fromString(const string& s){
+ istringstream stream (s);
+ T t;
+ stream >> t;
+ return t;
}
-/**************************************************************************************************/
-
-inline bool inUsersGroups(string groupname, vector<string> Groups) {
- try {
- for (int i = 0; i < Groups.size(); i++) {
- if (groupname == Groups[i]) { return true; }
- }
- return false;
- }
- catch(exception& e) {
- cout << "Standard Error: " << e.what() << " has occurred in the mothur class Function inUsersGroups. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
- exit(1);
- }
- catch(...) {
- cout << "An unknown error has occurred in the mothur class function inUsersGroups. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
- exit(1);
- }
-}
-
-/**************************************************************************************************/
+//**********************************************************************************************************************
#endif