bump to stable version

rtk/IO.cpp

@@ -5,23 +5,23 @@ std::mutex rarefyMutex;
 void lineCntOut(const string inF, const string outF, const string arg4){
 	ifstream in(inF.c_str());
 	ofstream out(outF.c_str(), ios::out);
-	if (!in){ 
+	if (!in){
 #ifdef notRpackage
-cerr << "Can't open infile " << inF << endl; std::exit(99); 
+cerr << "Can't open infile " << inF << endl; std::exit(99);
 #endif
 }
-	if (!out){ 
+	if (!out){
 #ifdef notRpackage
-cerr << "Can't open outfile " << outF << endl; std::exit(99); 
+cerr << "Can't open outfile " << outF << endl; std::exit(99);
 #endif
 }
 	//read file that contains nicely ordered all indexes of lines to be extracted
 	string line;
 	vector<uint> srtTar;
 	ifstream idxS(arg4.c_str());
-	if (!idxS){ 
+	if (!idxS){
 #ifdef notRpackage
-cerr << "Can't open outfile " << arg4 << endl; std::exit(99); 
+cerr << "Can't open outfile " << arg4 << endl; std::exit(99);
 #endif
 }
 	while (getline(idxS, line, '\n')) {
@@ -35,7 +35,7 @@ cerr << "Can't open outfile " << arg4 << endl; std::exit(99);
 	sort(srtTar.begin(), srtTar.end());
 
 	//sort through new file
-	if (!out){ 
+	if (!out){
 #ifdef notRpackage
 cerr << "Can't open outfile " << outF << endl; std::exit(99);
 #endif
@@ -53,7 +53,7 @@ cerr << "Can't open outfile " << outF << endl; std::exit(99);
 
 	in.close(); out.close();
 	if (j != srtTar.size()){
-		
+
 #ifdef notRpackage
 cerr << "Missed " << (srtTar.size() - j) << " entries." << endl;
 #endif
@@ -66,24 +66,24 @@ num_threads(nt), richness(-1), Shannon(-1.f){
 	for (uint i = 0; i < vec.size(); i++){
 		cumSum += vec[i];
 	}
-	if (verbose){ 
+	if (verbose){
 #ifdef notRpackage
 cerr << (long)cumSum << " allocating ";
 #endif
  }
 	//arr = (int*) malloc((int)cumSum * sizeof(int));
 	//arr = new unsigned short[(int)cumSum];
 	arr.resize((long)cumSum);
-	if (verbose){ 
+	if (verbose){
 #ifdef notRpackage
-cerr << "memory"; 
+cerr << "memory";
 #endif
 }
 	totSum = cumSum;
 	long k(0); uint posInVec(-1);
 	//numFeatures = 0;
 	for (size_t i = 0; i< vec.size(); i++){
-		
+
 		long maxG = (long)vec[i];
 		IDs.push_back( std::to_string(i));
 
@@ -99,9 +99,9 @@ cerr << "memory";
 	}
 	posInVec++;
 	numFeatures = posInVec;
-	if (verbose){ 
+	if (verbose){
 	#ifdef notRpackage
-	cerr << "..\n"; 
+	cerr << "..\n";
 	#endif
 	}
 }
@@ -119,17 +119,12 @@ smplVec::smplVec(const string inF, const int nt) :IDs(0),totSum(0), num_threads(
 		vec.push_back(num); IDs.push_back(ID);
 	}
 	in.close();
-	//
-#ifdef notRpackage
-cerr<<"tt";std::vector<unsigned short> v((int)cumSum);
-#endif
+
 	if (verbose){
-#ifdef notRpackage
-cerr<<(long)cumSum<<" allocating ";
-#endif
-}
-	//arr = (int*) malloc((int)cumSum * sizeof(int));
-	//arr = new unsigned short[(int)cumSum];
+		#ifdef notRpackage
+		cerr<<(long)cumSum<<" allocating ";
+		#endif
+	}
 	arr.resize((long)cumSum);
 	if (verbose){
 	#ifdef notRpackage
@@ -139,7 +134,7 @@ cerr<<(long)cumSum<<" allocating ";
 	totSum = cumSum;
 	long k(0); uint posInVec(0);
 	for (size_t i = 0; i< vec.size(); i++){
-		
+
 		long maxG = (long)vec[i];
 		maxG += k;
 		if (maxG == 0){ continue; }//not really a feature, doesnt need ot be counted as cat
@@ -461,9 +456,9 @@ double smplVec::calc_eveness(const vector<uint>& vec){
 void DivEsts::print2file(const string file){
 	if (richness.size()<1){return;}
 	ofstream out(file.c_str());
-	if (!out){ 
+	if (!out){
 #ifdef notRpackage
-cerr << "Couldn't open diversity estimate file " << file << endl; std::exit(99); 
+cerr << "Couldn't open diversity estimate file " << file << endl; std::exit(99);
 #endif
 }
 	out<<"Richness\t"<<richness[0];
@@ -496,14 +491,14 @@ void printDivMat(const string outF, vector<DivEsts*>& inD, bool printDIV ){
 
 	string outFmedian = outF + "median_alpha_diversity.tsv";
 	ofstream out(outFmedian.c_str());
-	if (!out){ 
+	if (!out){
 #ifdef notRpackage
-cerr << "Couldn't open diversity estimate matrix " << outF << endl; std::exit(99); 
+cerr << "Couldn't open diversity estimate matrix " << outF << endl; std::exit(99);
 #endif
 }
 	out << "Smpl\tRichness\tShannon\tSimpson\tInv. Simpson\tChao1\tEveness\n";
 	for (size_t i = 0; i < inD.size(); i++){
-		if (inD[i] == NULL){ 
+		if (inD[i] == NULL){
 #ifdef notRpackage
 cerr << "Empty vector at index " << i << "in div mat building.\n";
 #endif
@@ -599,9 +594,9 @@ cerr << "Empty vector at index " << i << "in div mat building.\n";
 }
 void printRareMat(const string outF, vector< map< uint, uint >>& rMat, vector< string >& sampleNames, vector < string >& rowId){
 	ofstream out(outF.c_str());
-	if (!out){ 
+	if (!out){
 #ifdef notRpackage
-cerr << "Couldn't open rarefy matrix file " << outF << endl; std::exit(99); 
+cerr << "Couldn't open rarefy matrix file " << outF << endl; std::exit(99);
 #endif
 }
 
@@ -636,9 +631,9 @@ string printSimpleMap(map<uint, uint> vec, string outF, string id, vector<string
 	// to the disk.
 	// this way we dont need memory to do
 	ofstream out(outF.c_str(),  ios::binary);
-	if (!out){ 
+	if (!out){
 #ifdef notRpackage
-cerr << "Couldn't open tmpvec file " << outF << endl; std::exit(99); 
+cerr << "Couldn't open tmpvec file " << outF << endl; std::exit(99);
 #endif
 }
 	for(uint i = 0; i < rowNames.size(); i++){
@@ -689,9 +684,9 @@ void reassembleTmpMat(vector<string> inF, vector< string > rowNames, vector< str
 	}
 
 	ofstream out(outF.c_str());
-	if (!out){ 
+	if (!out){
 #ifdef notRpackage
-cerr << "Couldn't open tmpvec file " << outF << endl; std::exit(99); 
+cerr << "Couldn't open tmpvec file " << outF << endl; std::exit(99);
 #endif
 }
 	out << "Rarefied";

rtk/Rare.cpp

@@ -1,13 +1,7 @@
-// rarefaction.cpp
-// usage: rarefaction in_matrix outfile
-//outfile contains richness and sample sum; in same dir several files are created
-//C:\Users\Falk\SkyDrive\science\data\test\test.mat.subs
-
-//#include "Matrix.h"
 #include "ClStr2Mat.h"
 #include "Rare.h"
 
-const char* rar_ver="0.7 beta";
+const char* rar_ver="0.9";
 
 
 rareStruct* calcDivRar(int i, Matrix* Mo, DivEsts* div, long rareDep,
@@ -39,6 +33,8 @@ rareStruct* calcDivRar(int i, Matrix* Mo, DivEsts* div, long rareDep,
 rareStruct* calcDivRarVec(int i, vector<string> fileNames, DivEsts* div, long rareDep,
 	vector<vector<uint>>* abundInRow, vector<vector<uint>>* occuencesInRow, string outF,
 	int repeats, int writeFiles){
+
+		exit(0);
 //	cout << i << " ";
 	smplVec* cur = new smplVec(fileNames[i],4);
 
@@ -176,7 +172,7 @@ xtra("") {
 		else if (!strcmp(argv[i], "-xtra"))
 			xtra = (argv[++i]);
 
-		
+
 
 	}
 
@@ -280,6 +276,7 @@ void rareExtremLowMem(string inF, string outF, int writeFiles, string arg4, int
 		}else if( i == 3){
 			cout << "..." << std::endl ;
 		}
+
 		uint toWhere = done + numThr - 1;
 		if ((uint)((uint)fileNames.size() - 2 ) < toWhere){
 			toWhere = fileNames.size() - 2;
@@ -297,6 +294,7 @@ void rareExtremLowMem(string inF, string outF, int writeFiles, string arg4, int
 		tmpRS = calcDivRarVec(i, fileNames,  div, rareDep, &abundInRow, &occuencesInRow, outF, repeats, writeFiles);
 		i++;
 
+
 		// process created data, first threads, then main thread
 		i = done;
 		for (; i < toWhere; i++){
@@ -384,50 +382,14 @@ int main(int argc, char* argv[])
 	string arg4 = std::to_string(opts->depth);
 	bool verbose = opts->verbose;
 
-	//mode = argv[1];
 
-	/*
-	long rareDep = 1000;	int repeats (1);
-	//bool splitMode(false),mergeMode(false),sumUpMode(false);
-	string inF = argv[2];
-	string outF = argv[3];
-	// hardcode mode, as the others are excluded for newRow
-	// can be removed, as soon as other functionality is ready
-	// to be used. June 2016
-	//string mode = argv[1];
-	string mode = argv[1];
-	string arg4 = "";
-	uint numThr = 1; //number of threads to use
-
-
-	if (argc>=5){
-		arg4 = argv[4];
-	}
-
-	if (argc>5){
-		repeats = atoi(argv[5]);
-	}
-	int writeFiles = repeats;
-	if (argc>6){
-		writeFiles = atoi(argv[6]);
-	}
-	if (argc > 7){
-		numThr = atoi(argv[7]);
-	}
-	if( writeFiles > repeats){
-		cerr << "Can not write more files than repeats. Set writefiles to repeats" << std::endl;
-		int writeFiles = repeats;
-	}
-
-	// flag to store data on disk
-	bool storeBinary = true;
-	vector < vector < string > > tmpMatFiles (writeFiles );
-	if (argc > 8){
-		storeBinary = atoi(argv[8]);
-	}*/
 
 	//all modes that classify as rarefactions:
 	if (mode == "swap" || mode == "memory") {
+		if(opts->input == ""){
+			cerr << "No input file specified. Exiting" << std::endl;
+			exit(0);
+		}
 		opts->print_rare_details();
 	}
 
@@ -566,7 +528,7 @@ int main(int argc, char* argv[])
 		// this will be used for Chao2 estimation
 		vector<vector<uint>> abundInRow(opts->repeats, vector<uint>(Mo->rowNum(),0));
 		vector<vector<uint>> occuencesInRow(opts->repeats, vector<uint>(Mo->rowNum(),0));
-		
+
 		//object to keep matrices
 		vector < vector < string > > tmpMatFiles(opts->write);
 		//cerr << "TH";
@@ -601,7 +563,7 @@ int main(int argc, char* argv[])
 				divvs[i] 		= tmpRS->div;
 				string curS 	= Mo->getSampleName(i);
 				//divvs[i-done]->print2file(outF + curS + "_alpha_div.tsv");
-				
+
 				// add the matrices to the container
 				if(NoOfMatrices > 0){
 					if(opts->writeSwap){