X-Git-Url: https://git.donarmstrong.com/?p=mothur.git;a=blobdiff_plain;f=distancecommand.cpp;h=9243df3873a9687d43da9da940b32da1ce9a5091;hp=4bdeaaeb5d9b388ac56fc2553f57dc10c0d598d9;hb=cf9987b67aa49777a4c91c2d21f96e58bf17aa82;hpb=220dc345e493cddc569521111ce32ac4d965ab7f

diff --git a/distancecommand.cpp b/distancecommand.cpp
index 4bdeaae..9243df3 100644
--- a/distancecommand.cpp
+++ b/distancecommand.cpp
@@ -8,26 +8,21 @@
  */
 
 #include "distancecommand.h"
-#include "ignoregaps.h"
-#include "eachgapdist.h"
-#include "eachgapignore.h"
-#include "onegapdist.h"
-#include "onegapignore.h"
 
 //**********************************************************************************************************************
 vector<string> DistanceCommand::setParameters(){	
 	try {
-		CommandParameter pcolumn("column", "InputTypes", "", "", "none", "none", "OldFastaColumn",false,false); parameters.push_back(pcolumn);
-		CommandParameter poldfasta("oldfasta", "InputTypes", "", "", "none", "none", "OldFastaColumn",false,false); parameters.push_back(poldfasta);
-		CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pfasta);
-		CommandParameter poutput("output", "Multiple", "column-lt-square", "column", "", "", "",false,false); parameters.push_back(poutput);
-		CommandParameter pcalc("calc", "Multiple", "nogaps-eachgap-onegap", "onegap", "", "", "",false,false); parameters.push_back(pcalc);
-		CommandParameter pcountends("countends", "Boolean", "", "T", "", "", "",false,false); parameters.push_back(pcountends);
-		CommandParameter pcompress("compress", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(pcompress);
-		CommandParameter pprocessors("processors", "Number", "", "1", "", "", "",false,false); parameters.push_back(pprocessors);
-		CommandParameter pcutoff("cutoff", "Number", "", "1.0", "", "", "",false,false); parameters.push_back(pcutoff);
-		CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
-		CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
+		CommandParameter pcolumn("column", "InputTypes", "", "", "none", "none", "OldFastaColumn","column",false,false); parameters.push_back(pcolumn);
+		CommandParameter poldfasta("oldfasta", "InputTypes", "", "", "none", "none", "OldFastaColumn","",false,false); parameters.push_back(poldfasta);
+		CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "none", "none","phylip-column",false,true, true); parameters.push_back(pfasta);
+		CommandParameter poutput("output", "Multiple", "column-lt-square-phylip", "column", "", "", "","phylip-column",false,false, true); parameters.push_back(poutput);
+		CommandParameter pcalc("calc", "Multiple", "nogaps-eachgap-onegap", "onegap", "", "", "","",false,false); parameters.push_back(pcalc);
+		CommandParameter pcountends("countends", "Boolean", "", "T", "", "", "","",false,false); parameters.push_back(pcountends);
+		CommandParameter pcompress("compress", "Boolean", "", "F", "", "", "","",false,false); parameters.push_back(pcompress);
+		CommandParameter pprocessors("processors", "Number", "", "1", "", "", "","",false,false, true); parameters.push_back(pprocessors);
+		CommandParameter pcutoff("cutoff", "Number", "", "1.0", "", "", "","",false,false, true); parameters.push_back(pcutoff);
+		CommandParameter pinputdir("inputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(pinputdir);
+		CommandParameter poutputdir("outputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(poutputdir);
 		
 		vector<string> myArray;
 		for (int i = 0; i < parameters.size(); i++) {	myArray.push_back(parameters[i].name);		}
@@ -64,6 +59,22 @@ string DistanceCommand::getHelpString(){
 	}
 }
 //**********************************************************************************************************************
+string DistanceCommand::getOutputPattern(string type) {
+    try {
+        string pattern = "";
+        
+        if (type == "phylip") {  pattern = "[filename],[outputtag],dist"; } 
+        else if (type == "column") { pattern = "[filename],dist"; }
+        else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true;  }
+        
+        return pattern;
+    }
+    catch(exception& e) {
+        m->errorOut(e, "DistanceCommand", "getOutputPattern");
+        exit(1);
+    }
+}
+//**********************************************************************************************************************
 DistanceCommand::DistanceCommand(){	
 	try {
 		abort = true; calledHelp = true; 
@@ -184,42 +195,23 @@ DistanceCommand::DistanceCommand(string option) {
 			convert(temp, countends); 
 			
 			temp = validParameter.validFile(parameters, "cutoff", false);		if(temp == "not found"){	temp = "1.0"; }
-			convert(temp, cutoff); 
+			m->mothurConvert(temp, cutoff); 
 			
 			temp = validParameter.validFile(parameters, "processors", false);	if (temp == "not found"){	temp = m->getProcessors();	}
 			m->setProcessors(temp);
-			convert(temp, processors);
+			m->mothurConvert(temp, processors);
 			
 			temp = validParameter.validFile(parameters, "compress", false);		if(temp == "not found"){  temp = "F"; }
 			convert(temp, compress);
 
 			output = validParameter.validFile(parameters, "output", false);		if(output == "not found"){	output = "column"; }
+            if (output == "phylip") { output = "lt";  }
 			
 			if (((column != "") && (oldfastafile == "")) || ((column == "") && (oldfastafile != ""))) { m->mothurOut("If you provide column or oldfasta, you must provide both."); m->mothurOutEndLine(); abort=true; }
 			
 			if ((column != "") && (oldfastafile != "") && (output != "column")) { m->mothurOut("You have provided column and oldfasta, indicating you want to append distances to your column file. Your output must be in column format to do so."); m->mothurOutEndLine(); abort=true; }
 			
 			if ((output != "column") && (output != "lt") && (output != "square")) { m->mothurOut(output + " is not a valid output form. Options are column, lt and square. I will use column."); m->mothurOutEndLine(); output = "column"; }
-			
-			ValidCalculators validCalculator;
-			
-			if (m->isTrue(countends) == true) {
-				for (int i=0; i<Estimators.size(); i++) {
-					if (validCalculator.isValidCalculator("distance", Estimators[i]) == true) { 
-						if (Estimators[i] == "nogaps")			{	distCalculator = new ignoreGaps();	}
-						else if (Estimators[i] == "eachgap")	{	distCalculator = new eachGapDist();	}
-						else if (Estimators[i] == "onegap")		{	distCalculator = new oneGapDist();	}
-					}
-				}
-			}else {
-				for (int i=0; i<Estimators.size(); i++) {
-					if (validCalculator.isValidCalculator("distance", Estimators[i]) == true) { 
-						if (Estimators[i] == "nogaps")		{	distCalculator = new ignoreGaps();					}
-						else if (Estimators[i] == "eachgap"){	distCalculator = new eachGapIgnoreTermGapDist();	}
-						else if (Estimators[i] == "onegap")	{	distCalculator = new oneGapIgnoreTermGapDist();		}
-					}
-				}
-			}
 
 		}
 				
@@ -252,14 +244,17 @@ int DistanceCommand::execute(){
 		if (!alignDB.sameLength()) {  m->mothurOut("[ERROR]: your sequences are not the same length, aborting."); m->mothurOutEndLine(); return 0; }
 		
 		string outputFile;
-				
+        
+        map<string, string> variables; 
+        variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(fastafile));
 		if (output == "lt") { //does the user want lower triangle phylip formatted file 
-			outputFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "phylip.dist";
-			remove(outputFile.c_str()); outputTypes["phylip"].push_back(outputFile);
+            variables["[outputtag]"] = "phylip";
+			outputFile = getOutputFileName("phylip", variables);
+			m->mothurRemove(outputFile); outputTypes["phylip"].push_back(outputFile);
 			
 			//output numSeqs to phylip formatted dist file
 		}else if (output == "column") { //user wants column format
-			outputFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "dist";
+			outputFile = getOutputFileName("column", variables);
 			outputTypes["column"].push_back(outputFile);
 			
 			//so we don't accidentally overwrite
@@ -268,10 +263,11 @@ int DistanceCommand::execute(){
 				rename(column.c_str(), tempcolumn.c_str());
 			}
 			
-			remove(outputFile.c_str());
+			m->mothurRemove(outputFile);
 		}else { //assume square
-			outputFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "square.dist";
-			remove(outputFile.c_str());
+			variables["[outputtag]"] = "square";
+			outputFile = getOutputFileName("phylip", variables);
+			m->mothurRemove(outputFile);
 			outputTypes["phylip"].push_back(outputFile);
 		}
 		
@@ -314,11 +310,11 @@ int DistanceCommand::execute(){
 				
 				driverMPI(start, end, outMPI, cutoff); 
 				
-				if (m->control_pressed) { outputTypes.clear(); MPI_File_close(&outMPI); delete distCalculator;  return 0; }
+				if (m->control_pressed) { outputTypes.clear(); MPI_File_close(&outMPI);   return 0; }
 			
 				//wait on chidren
 				for(int i = 1; i < processors; i++) { 
-					if (m->control_pressed) { outputTypes.clear();  MPI_File_close(&outMPI);  delete distCalculator;  return 0; }
+					if (m->control_pressed) { outputTypes.clear();  MPI_File_close(&outMPI);    return 0; }
 					
 					char buf[5];
 					MPI_Recv(buf, 5, MPI_CHAR, i, tag, MPI_COMM_WORLD, &status); 
@@ -327,7 +323,7 @@ int DistanceCommand::execute(){
 				//do your part
 				driverMPI(start, end, outMPI, cutoff); 
 				
-				if (m->control_pressed) { outputTypes.clear();  MPI_File_close(&outMPI);  delete distCalculator;  return 0; }
+				if (m->control_pressed) { outputTypes.clear();  MPI_File_close(&outMPI);   return 0; }
 			
 				char buf[5];
 				strcpy(buf, "done"); 
@@ -342,12 +338,12 @@ int DistanceCommand::execute(){
 			
 				//do your part
 				string outputMyPart;
-				unsigned long int mySize;
+				unsigned long long mySize;
 				
 				if (output != "square"){ driverMPI(start, end, outputFile, mySize); }
 				else { driverMPI(start, end, outputFile, mySize, output); }
 	
-				if (m->control_pressed) {  outputTypes.clear();  delete distCalculator;  return 0; }
+				if (m->control_pressed) {  outputTypes.clear();   return 0; }
 				
 				int amode=MPI_MODE_APPEND|MPI_MODE_WRONLY|MPI_MODE_CREATE; //
 				MPI_File outMPI;
@@ -364,9 +360,9 @@ int DistanceCommand::execute(){
 
 				//wait on chidren
 				for(int b = 1; b < processors; b++) { 
-					unsigned long int fileSize;
+					unsigned long long fileSize;
 					
-					if (m->control_pressed) { outputTypes.clear();  MPI_File_close(&outMPI);  delete distCalculator;  return 0; }
+					if (m->control_pressed) { outputTypes.clear();  MPI_File_close(&outMPI);   return 0; }
 					
 					MPI_Recv(&fileSize, 1, MPI_LONG, b, tag, MPI_COMM_WORLD, &status); 
 					
@@ -392,11 +388,11 @@ int DistanceCommand::execute(){
 				MPI_File_close(&outMPI);
 			}else { //you are a child process
 				//do your part
-				unsigned long int size;
+				unsigned long long size;
 				if (output != "square"){ driverMPI(start, end, (outputFile + toString(pid) + ".temp"), size); }
 				else { driverMPI(start, end, (outputFile + toString(pid) + ".temp"), size, output); }
 				
-				if (m->control_pressed) { delete distCalculator;  return 0; }
+				if (m->control_pressed) {  return 0; }
 			
 				//tell parent you are done.
 				MPI_Send(&size, 1, MPI_LONG, 0, tag, MPI_COMM_WORLD);
@@ -405,14 +401,14 @@ int DistanceCommand::execute(){
 		MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
 #else		
 				
-	#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+	//#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 		//if you don't need to fork anything
 		if(processors == 1){
 			if (output != "square") {  driver(0, numSeqs, outputFile, cutoff); }
 			else { driver(0, numSeqs, outputFile, "square");  }
 		}else{ //you have multiple processors
 			
-			unsigned long int numDists = 0;
+			unsigned long long numDists = 0;
 			
 			if (output == "square") {
 				 numDists = numSeqs * numSeqs;
@@ -442,14 +438,14 @@ int DistanceCommand::execute(){
 			
 			createProcesses(outputFile); 
 		}
-	#else
+	//#else
 		//ifstream inFASTA;
-		if (output != "square") {  driver(0, numSeqs, outputFile, cutoff); }
-		else { driver(0, numSeqs, outputFile, "square");  }
-	#endif
+		//if (output != "square") {  driver(0, numSeqs, outputFile, cutoff); }
+		//else { driver(0, numSeqs, outputFile, "square");  }
+	//#endif
 	
 #endif
-		if (m->control_pressed) { outputTypes.clear();  delete distCalculator; remove(outputFile.c_str()); return 0; }
+		if (m->control_pressed) { outputTypes.clear();  m->mothurRemove(outputFile); return 0; }
 		
 		#ifdef USE_MPI
 			MPI_Comm_rank(MPI_COMM_WORLD, &pid); 
@@ -472,17 +468,17 @@ int DistanceCommand::execute(){
 			if (outputFile == column) { 
 				string tempcolumn = column + ".old";
 				m->appendFiles(tempcolumn, outputFile);
-				remove(tempcolumn.c_str());
+				m->mothurRemove(tempcolumn);
 			}else{
 				m->appendFiles(outputFile, column);
-				remove(outputFile.c_str());
+				m->mothurRemove(outputFile);
 				outputFile = column;
 			}
 			
 			if (outputDir != "") { 
 				string newOutputName = outputDir + m->getSimpleName(outputFile);
 				rename(outputFile.c_str(), newOutputName.c_str());
-				remove(outputFile.c_str());
+				m->mothurRemove(outputFile);
 				outputFile = newOutputName;
 			}
 		}
@@ -492,9 +488,7 @@ int DistanceCommand::execute(){
 			}
 		#endif
 		
-		if (m->control_pressed) { outputTypes.clear();  delete distCalculator; remove(outputFile.c_str()); return 0; }
-		
-		delete distCalculator;
+		if (m->control_pressed) { outputTypes.clear();  m->mothurRemove(outputFile); return 0; }
 		
 		//set phylip file as new current phylipfile
 		string current = "";
@@ -510,7 +504,7 @@ int DistanceCommand::execute(){
 		}
 		
 		m->mothurOutEndLine();
-		m->mothurOut("Output File Name: "); m->mothurOutEndLine();
+		m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 		m->mothurOut(outputFile); m->mothurOutEndLine();
 		m->mothurOutEndLine();
 		m->mothurOut("It took " + toString(time(NULL) - startTime) + " to calculate the distances for " + toString(numSeqs) + " sequences."); m->mothurOutEndLine();
@@ -534,7 +528,7 @@ int DistanceCommand::execute(){
 /**************************************************************************************************/
 void DistanceCommand::createProcesses(string filename) {
 	try {
-#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 		int process = 1;
 		processIDS.clear();
 		
@@ -567,25 +561,83 @@ void DistanceCommand::createProcesses(string filename) {
 			int temp = processIDS[i];
 			wait(&temp);
 		}
+#else
+		//////////////////////////////////////////////////////////////////////////////////////////////////////
+		//Windows version shared memory, so be careful when passing variables through the distanceData struct. 
+		//Above fork() will clone, so memory is separate, but that's not the case with windows, 
+		//that's why the distance calculator was moved inside of the driver to make separate copies.
+		//////////////////////////////////////////////////////////////////////////////////////////////////////
+		
+		vector<distanceData*> pDataArray; //[processors-1];
+		DWORD   dwThreadIdArray[processors-1];
+		HANDLE  hThreadArray[processors-1]; 
+		
+		//Create processor-1 worker threads.
+		for( int i=0; i<processors-1; i++ ){
+			
+			// Allocate memory for thread data.
+			distanceData* tempDist = new distanceData(lines[i+1].start, lines[i+1].end, (filename + toString(i) + ".temp"), cutoff, alignDB, Estimators, m, output, numNewFasta, countends);
+			pDataArray.push_back(tempDist);
+			processIDS.push_back(i);
+			
+			//MyDistThreadFunction is in header. It must be global or static to work with the threads.
+			//default security attributes, thread function name, argument to thread function, use default creation flags, returns the thread identifier
+			hThreadArray[i] = CreateThread(NULL, 0, MyDistThreadFunction, pDataArray[i], 0, &dwThreadIdArray[i]);   
+		}
+		
+		//do your part
+		if (output != "square") {  driver(lines[0].start, lines[0].end, filename, cutoff); }
+		else { driver(lines[0].start, lines[0].end, filename, "square"); }
+		
+		//Wait until all threads have terminated.
+		WaitForMultipleObjects(processors-1, hThreadArray, TRUE, INFINITE);
+		
+		//Close all thread handles and free memory allocations.
+		for(int i=0; i < pDataArray.size(); i++){
+            if (pDataArray[i]->count != (pDataArray[i]->endLine-pDataArray[i]->startLine)) {
+                m->mothurOut("[ERROR]: process " + toString(i) + " only processed " + toString(pDataArray[i]->count) + " of " + toString(pDataArray[i]->endLine-pDataArray[i]->startLine) + " sequences assigned to it, quitting. \n"); m->control_pressed = true; 
+            }
+			CloseHandle(hThreadArray[i]);
+			delete pDataArray[i];
+		}
+#endif
 		
 		//append and remove temp files
 		for (int i=0;i<processIDS.size();i++) { 
 			m->appendFiles((filename + toString(processIDS[i]) + ".temp"), filename);
-			remove((filename + toString(processIDS[i]) + ".temp").c_str());
+			m->mothurRemove((filename + toString(processIDS[i]) + ".temp"));
 		}
-#endif
+		
 	}
 	catch(exception& e) {
 		m->errorOut(e, "DistanceCommand", "createProcesses");
 		exit(1);
 	}
 }
-
 /**************************************************************************************************/
 /////// need to fix to work with calcs and sequencedb
 int DistanceCommand::driver(int startLine, int endLine, string dFileName, float cutoff){
 	try {
-
+		ValidCalculators validCalculator;
+		Dist* distCalculator;
+		if (m->isTrue(countends) == true) {
+			for (int i=0; i<Estimators.size(); i++) {
+				if (validCalculator.isValidCalculator("distance", Estimators[i]) == true) { 
+					if (Estimators[i] == "nogaps")			{	distCalculator = new ignoreGaps();	}
+					else if (Estimators[i] == "eachgap")	{	distCalculator = new eachGapDist();	}
+					else if (Estimators[i] == "onegap")		{	distCalculator = new oneGapDist();	}
+				}
+			}
+		}else {
+			for (int i=0; i<Estimators.size(); i++) {
+				if (validCalculator.isValidCalculator("distance", Estimators[i]) == true) { 
+					if (Estimators[i] == "nogaps")		{	distCalculator = new ignoreGaps();					}
+					else if (Estimators[i] == "eachgap"){	distCalculator = new eachGapIgnoreTermGapDist();	}
+					else if (Estimators[i] == "onegap")	{	distCalculator = new oneGapIgnoreTermGapDist();		}
+				}
+			}
+		}
+		
 		int startTime = time(NULL);
 		
 		//column file
@@ -605,7 +657,7 @@ int DistanceCommand::driver(int startLine, int endLine, string dFileName, float
 			}
 			for(int j=0;j<i;j++){
 				
-				if (m->control_pressed) { outFile.close(); return 0;  }
+				if (m->control_pressed) { delete distCalculator; outFile.close(); return 0;  }
 				
 				//if there was a column file given and we are appending, we don't want to calculate the distances that are already in the column file
 				//the alignDB contains the new sequences and then the old, so if i an oldsequence and j is an old sequence then break out of this loop
@@ -623,13 +675,14 @@ int DistanceCommand::driver(int startLine, int endLine, string dFileName, float
 			if (output == "lt") { outFile << endl; }
 			
 			if(i % 100 == 0){
-				m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
+				m->mothurOutJustToScreen(toString(i) + "\t" + toString(time(NULL) - startTime)+"\n"); 
 			}
 			
 		}
-		m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
+		m->mothurOutJustToScreen(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)+"\n");
 		
 		outFile.close();
+		delete distCalculator;
 		
 		return 1;
 	}
@@ -642,7 +695,26 @@ int DistanceCommand::driver(int startLine, int endLine, string dFileName, float
 /////// need to fix to work with calcs and sequencedb
 int DistanceCommand::driver(int startLine, int endLine, string dFileName, string square){
 	try {
-
+		ValidCalculators validCalculator;
+		Dist* distCalculator;
+		if (m->isTrue(countends) == true) {
+			for (int i=0; i<Estimators.size(); i++) {
+				if (validCalculator.isValidCalculator("distance", Estimators[i]) == true) { 
+					if (Estimators[i] == "nogaps")			{	distCalculator = new ignoreGaps();	}
+					else if (Estimators[i] == "eachgap")	{	distCalculator = new eachGapDist();	}
+					else if (Estimators[i] == "onegap")		{	distCalculator = new oneGapDist();	}
+				}
+			}
+		}else {
+			for (int i=0; i<Estimators.size(); i++) {
+				if (validCalculator.isValidCalculator("distance", Estimators[i]) == true) { 
+					if (Estimators[i] == "nogaps")		{	distCalculator = new ignoreGaps();					}
+					else if (Estimators[i] == "eachgap"){	distCalculator = new eachGapIgnoreTermGapDist();	}
+					else if (Estimators[i] == "onegap")	{	distCalculator = new oneGapIgnoreTermGapDist();		}
+				}
+			}
+		}
+		
 		int startTime = time(NULL);
 		
 		//column file
@@ -662,7 +734,7 @@ int DistanceCommand::driver(int startLine, int endLine, string dFileName, string
 			
 			for(int j=0;j<alignDB.getNumSeqs();j++){
 				
-				if (m->control_pressed) { outFile.close(); return 0;  }
+				if (m->control_pressed) { delete distCalculator; outFile.close(); return 0;  }
 				
 				distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
 				double dist = distCalculator->getDist();
@@ -673,13 +745,14 @@ int DistanceCommand::driver(int startLine, int endLine, string dFileName, string
 			outFile << endl; 
 			
 			if(i % 100 == 0){
-				m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
+				m->mothurOutJustToScreen(toString(i) + "\t" + toString(time(NULL) - startTime)+"\n");
 			}
 			
 		}
-		m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
+		m->mothurOutJustToScreen(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)+"\n");
 		
 		outFile.close();
+		delete distCalculator;
 		
 		return 1;
 	}
@@ -693,6 +766,28 @@ int DistanceCommand::driver(int startLine, int endLine, string dFileName, string
 /////// need to fix to work with calcs and sequencedb
 int DistanceCommand::driverMPI(int startLine, int endLine, MPI_File& outMPI, float cutoff){
 	try {
+		
+		ValidCalculators validCalculator;
+		Dist* distCalculator;
+		if (m->isTrue(countends) == true) {
+			for (int i=0; i<Estimators.size(); i++) {
+				if (validCalculator.isValidCalculator("distance", Estimators[i]) == true) { 
+					if (Estimators[i] == "nogaps")			{	distCalculator = new ignoreGaps();	}
+					else if (Estimators[i] == "eachgap")	{	distCalculator = new eachGapDist();	}
+					else if (Estimators[i] == "onegap")		{	distCalculator = new oneGapDist();	}
+				}
+			}
+		}else {
+			for (int i=0; i<Estimators.size(); i++) {
+				if (validCalculator.isValidCalculator("distance", Estimators[i]) == true) { 
+					if (Estimators[i] == "nogaps")		{	distCalculator = new ignoreGaps();					}
+					else if (Estimators[i] == "eachgap"){	distCalculator = new eachGapIgnoreTermGapDist();	}
+					else if (Estimators[i] == "onegap")	{	distCalculator = new oneGapIgnoreTermGapDist();		}
+				}
+			}
+		}
+		
+		
 		MPI_Status status;
 		int startTime = time(NULL);
 		
@@ -702,7 +797,7 @@ int DistanceCommand::driverMPI(int startLine, int endLine, MPI_File& outMPI, flo
 	
 			for(int j=0;j<i;j++){
 				
-				if (m->control_pressed) {  return 0;  }
+				if (m->control_pressed) {  delete distCalculator; return 0;  }
 				
 				//if there was a column file given and we are appending, we don't want to calculate the distances that are already in the column file
 				//the alignDB contains the new sequences and then the old, so if i an oldsequence and j is an old sequence then break out of this loop
@@ -717,8 +812,7 @@ int DistanceCommand::driverMPI(int startLine, int endLine, MPI_File& outMPI, flo
 			}
 			
 			if(i % 100 == 0){
-				//m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
-				cout << i << '\t' << (time(NULL) - startTime) << endl;
+				m->mothurOutJustToScreen(toString(i) + "\t" + toString(time(NULL) - startTime)+"\n"); 
 			}
 			
 			 
@@ -734,8 +828,8 @@ int DistanceCommand::driverMPI(int startLine, int endLine, MPI_File& outMPI, flo
 			
 		}
 		
-		//m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
-		cout << (endLine-1) << '\t' << (time(NULL) - startTime) << endl;		
+		m->mothurOutJustToScreen(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)+"\n");
+		delete distCalculator;
 		return 1;
 	}
 	catch(exception& e) {
@@ -745,8 +839,29 @@ int DistanceCommand::driverMPI(int startLine, int endLine, MPI_File& outMPI, flo
 }
 /**************************************************************************************************/
 /////// need to fix to work with calcs and sequencedb
-int DistanceCommand::driverMPI(int startLine, int endLine, string file, unsigned long int& size){
+int DistanceCommand::driverMPI(int startLine, int endLine, string file, unsigned long long& size){
 	try {
+		ValidCalculators validCalculator;
+		Dist* distCalculator;
+		if (m->isTrue(countends) == true) {
+			for (int i=0; i<Estimators.size(); i++) {
+				if (validCalculator.isValidCalculator("distance", Estimators[i]) == true) { 
+					if (Estimators[i] == "nogaps")			{	distCalculator = new ignoreGaps();	}
+					else if (Estimators[i] == "eachgap")	{	distCalculator = new eachGapDist();	}
+					else if (Estimators[i] == "onegap")		{	distCalculator = new oneGapDist();	}
+				}
+			}
+		}else {
+			for (int i=0; i<Estimators.size(); i++) {
+				if (validCalculator.isValidCalculator("distance", Estimators[i]) == true) { 
+					if (Estimators[i] == "nogaps")		{	distCalculator = new ignoreGaps();					}
+					else if (Estimators[i] == "eachgap"){	distCalculator = new eachGapIgnoreTermGapDist();	}
+					else if (Estimators[i] == "onegap")	{	distCalculator = new oneGapIgnoreTermGapDist();		}
+				}
+			}
+		}
+		
+		
 		MPI_Status status;
 		
 		MPI_File outMPI;
@@ -778,7 +893,7 @@ int DistanceCommand::driverMPI(int startLine, int endLine, string file, unsigned
 			
 			for(int j=0;j<i;j++){
 				
-				if (m->control_pressed) {  return 0;  }
+				if (m->control_pressed) { delete distCalculator; return 0;  }
 				
 				distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
 				double dist = distCalculator->getDist();
@@ -790,9 +905,7 @@ int DistanceCommand::driverMPI(int startLine, int endLine, string file, unsigned
 
 		
 			if(i % 100 == 0){
-				//m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
-				cout << i << '\t' << (time(NULL) - startTime) << endl;
-			}
+				m->mothurOutJustToScreen(toString(i) + "\t" + toString(time(NULL) - startTime)+"\n"); 			}
 			
 			
 			//send results to parent
@@ -806,9 +919,10 @@ int DistanceCommand::driverMPI(int startLine, int endLine, string file, unsigned
 			delete buf;
 		}
 		
-		//m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
-		cout << (endLine-1) << '\t' << (time(NULL) - startTime) << endl;
+		m->mothurOutJustToScreen(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)+"\n");
+		
 		MPI_File_close(&outMPI);
+		delete distCalculator;
 		
 		return 1;
 	}
@@ -819,8 +933,28 @@ int DistanceCommand::driverMPI(int startLine, int endLine, string file, unsigned
 }
 /**************************************************************************************************/
 /////// need to fix to work with calcs and sequencedb
-int DistanceCommand::driverMPI(int startLine, int endLine, string file, unsigned long int& size, string square){
+int DistanceCommand::driverMPI(int startLine, int endLine, string file, unsigned long long& size, string square){
 	try {
+		ValidCalculators validCalculator;
+		Dist* distCalculator;
+		if (m->isTrue(countends) == true) {
+			for (int i=0; i<Estimators.size(); i++) {
+				if (validCalculator.isValidCalculator("distance", Estimators[i]) == true) { 
+					if (Estimators[i] == "nogaps")			{	distCalculator = new ignoreGaps();	}
+					else if (Estimators[i] == "eachgap")	{	distCalculator = new eachGapDist();	}
+					else if (Estimators[i] == "onegap")		{	distCalculator = new oneGapDist();	}
+				}
+			}
+		}else {
+			for (int i=0; i<Estimators.size(); i++) {
+				if (validCalculator.isValidCalculator("distance", Estimators[i]) == true) { 
+					if (Estimators[i] == "nogaps")		{	distCalculator = new ignoreGaps();					}
+					else if (Estimators[i] == "eachgap"){	distCalculator = new eachGapIgnoreTermGapDist();	}
+					else if (Estimators[i] == "onegap")	{	distCalculator = new oneGapIgnoreTermGapDist();		}
+				}
+			}
+		}
+		
 		MPI_Status status;
 		
 		MPI_File outMPI;
@@ -852,7 +986,7 @@ int DistanceCommand::driverMPI(int startLine, int endLine, string file, unsigned
 			
 			for(int j=0;j<alignDB.getNumSeqs();j++){
 				
-				if (m->control_pressed) {  return 0;  }
+				if (m->control_pressed) { delete distCalculator; return 0;  }
 				
 				distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
 				double dist = distCalculator->getDist();
@@ -864,8 +998,7 @@ int DistanceCommand::driverMPI(int startLine, int endLine, string file, unsigned
 
 		
 			if(i % 100 == 0){
-				//m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
-				cout << i << '\t' << (time(NULL) - startTime) << endl;
+				m->mothurOutJustToScreen(toString(i) + "\t" + toString(time(NULL) - startTime)+"\n"); 
 			}
 			
 			
@@ -880,10 +1013,10 @@ int DistanceCommand::driverMPI(int startLine, int endLine, string file, unsigned
 			delete buf;
 		}
 		
-		//m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
-		cout << (endLine-1) << '\t' << (time(NULL) - startTime) << endl;
-		MPI_File_close(&outMPI);
+		m->mothurOutJustToScreen(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)+"\n");
 		
+		MPI_File_close(&outMPI);
+		delete distCalculator;
 		return 1;
 	}
 	catch(exception& e) {
@@ -900,7 +1033,7 @@ int DistanceCommand::convertMatrix(string outputFile) {
 		string outfile = m->getRootName(outputFile) + "sorted.dist.temp";
 		
 		//use the unix sort 
-		#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+		#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 			string command = "sort -n " + outputFile + " -o " + outfile;
 			system(command.c_str());
 		#else //sort using windows sort
@@ -935,7 +1068,7 @@ int DistanceCommand::convertMatrix(string outputFile) {
 		//m->openInputFile(outfile, in);
 		
 		while(!in.eof()) {
-			if (m->control_pressed) { in.close(); remove(outfile.c_str()); out.close(); return 0; }
+			if (m->control_pressed) { in.close(); m->mothurRemove(outfile); out.close(); return 0; }
 			
 			in >> first >> second >> dist; m->gobble(in);
 				
@@ -970,7 +1103,7 @@ int DistanceCommand::convertMatrix(string outputFile) {
 		in.close();
 		out.close();
 		
-		remove(outfile.c_str());
+		m->mothurRemove(outfile);
 		
 		return 1;
 		
@@ -980,7 +1113,7 @@ int DistanceCommand::convertMatrix(string outputFile) {
 		exit(1);
 	}
 }
-/**************************************************************************************************
+**************************************************************************************************
 int DistanceCommand::convertToLowerTriangle(string outputFile) {
 	try{
 
@@ -988,7 +1121,7 @@ int DistanceCommand::convertToLowerTriangle(string outputFile) {
 		string outfile = m->getRootName(outputFile) + "sorted.dist.temp";
 		
 		//use the unix sort 
-		#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+		#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 			string command = "sort -n " + outputFile + " -o " + outfile;
 			system(command.c_str());
 		#else //sort using windows sort
@@ -1025,7 +1158,7 @@ int DistanceCommand::convertToLowerTriangle(string outputFile) {
 		//m->openInputFile(outfile, in);
 		
 		while(!in.eof()) {
-			if (m->control_pressed) { in.close(); remove(outfile.c_str()); out.close(); return 0; }
+			if (m->control_pressed) { in.close(); m->mothurRemove(outfile); out.close(); return 0; }
 			
 			in >> first >> second >> dist; m->gobble(in);
 				
@@ -1064,7 +1197,7 @@ int DistanceCommand::convertToLowerTriangle(string outputFile) {
 		in.close();
 		out.close();
 		
-		remove(outfile.c_str());
+		m->mothurRemove(outfile);
 		
 		return 1;
 		
@@ -1074,7 +1207,7 @@ int DistanceCommand::convertToLowerTriangle(string outputFile) {
 		exit(1);
 	}
 }
-/**************************************************************************************************/
+**************************************************************************************************/
 //its okay if the column file does not contain all the names in the fasta file, since some distance may have been above a cutoff,
 //but no sequences can be in the column file that are not in oldfasta. also, if a distance is above the cutoff given then remove it.
 //also check to make sure the 2 files have the same alignment length.
@@ -1135,7 +1268,7 @@ bool DistanceCommand::sanityCheck() {
 		string name1, name2;
 		float dist;
 		while (!inDist.eof()) {
-			if (m->control_pressed) {  inDist.close(); outDist.close(); remove(outputFile.c_str()); return good;  }
+			if (m->control_pressed) {  inDist.close(); outDist.close(); m->mothurRemove(outputFile); return good;  }
 		
 			inDist >> name1 >> name2 >> dist; m->gobble(inDist);
 			
@@ -1152,10 +1285,10 @@ bool DistanceCommand::sanityCheck() {
 		outDist.close();
 		
 		if (good) {
-			remove(column.c_str());
+			m->mothurRemove(column);
 			rename(outputFile.c_str(), column.c_str());
 		}else{
-			remove(outputFile.c_str()); //temp file is bad because file mismatch above
+			m->mothurRemove(outputFile); //temp file is bad because file mismatch above
 		}
 		
 		return good;