From: westcott <westcott>
Date: Tue, 22 Jun 2010 12:40:41 +0000 (+0000)
Subject: added mpi code to cluster.split command
X-Git-Url: https://git.donarmstrong.com/?p=mothur.git;a=commitdiff_plain;h=4ee9d3c973fb056f3105c26306da3b5d7dfb8108

added mpi code to cluster.split command
---

diff --git a/degapseqscommand.cpp b/degapseqscommand.cpp
new file mode 100644
index 0000000..c6c4b55
--- /dev/null
+++ b/degapseqscommand.cpp
@@ -0,0 +1,160 @@
+/*
+ *  degapseqscommand.cpp
+ *  Mothur
+ *
+ *  Created by westcott on 6/21/10.
+ *  Copyright 2010 Schloss Lab. All rights reserved.
+ *
+ */
+
+#include "degapseqscommand.h"
+#include "sequence.hpp"
+
+//***************************************************************************************************************
+
+DegapSeqsCommand::DegapSeqsCommand(string option)  {
+	try {
+		abort = false;
+		
+		//allow user to run help
+		if(option == "help") { help(); abort = true; }
+		
+		else {
+			//valid paramters for this command
+			string Array[] =  {"fasta", "outputdir","inputdir"};
+			vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
+			
+			OptionParser parser(option);
+			map<string,string> parameters = parser.getParameters();
+			
+			ValidParameters validParameter;
+			map<string,string>::iterator it;
+			
+			//check to make sure all parameters are valid for command
+			for (it = parameters.begin(); it != parameters.end(); it++) { 
+				if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
+			}
+			
+			//if the user changes the input directory command factory will send this info to us in the output parameter 
+			string inputDir = validParameter.validFile(parameters, "inputdir", false);		
+			if (inputDir == "not found"){	inputDir = "";		}
+			
+			//check for required parameters
+			fastafile = validParameter.validFile(parameters, "fasta", false);
+			if (fastafile == "not found") { fastafile = ""; m->mothurOut("fasta is a required parameter for the degap.seqs command."); m->mothurOutEndLine(); abort = true;  }
+			else { 
+				splitAtDash(fastafile, fastaFileNames);
+				
+				//go through files and make sure they are good, if not, then disregard them
+				for (int i = 0; i < fastaFileNames.size(); i++) {
+					if (inputDir != "") {
+						string path = hasPath(fastaFileNames[i]);
+						//if the user has not given a path then, add inputdir. else leave path alone.
+						if (path == "") {	fastaFileNames[i] = inputDir + fastaFileNames[i];		}
+					}
+	
+					ifstream in;
+					int ableToOpen = openInputFile(fastaFileNames[i], in);
+					in.close();
+					
+					if (ableToOpen == 1) { 
+						m->mothurOut(fastaFileNames[i] + " will be disregarded."); m->mothurOutEndLine(); 
+						//erase from file list
+						fastaFileNames.erase(fastaFileNames.begin()+i);
+						i--;
+					}
+				}
+				
+				//make sure there is at least one valid file left
+				if (fastaFileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
+			}
+
+			
+			//if the user changes the output directory command factory will send this info to us in the output parameter 
+			outputDir = validParameter.validFile(parameters, "outputdir", false);		if (outputDir == "not found"){	
+				outputDir = "";	
+				outputDir += hasPath(fastafile); //if user entered a file with a path then preserve it	
+			}
+
+		}
+	}
+	catch(exception& e) {
+		m->errorOut(e, "DegapSeqsCommand", "DegapSeqsCommand");
+		exit(1);
+	}
+}
+//**********************************************************************************************************************
+
+void DegapSeqsCommand::help(){
+	try {
+		m->mothurOut("The degap.seqs command reads a fastafile and removes all gap characters.\n");
+		m->mothurOut("The degap.seqs command parameter is fasta.\n");
+		m->mothurOut("The fasta parameter allows you to enter the fasta file containing your sequences, and is required. \n");
+		m->mothurOut("You may enter multiple fasta files by separating their names with dashes. ie. fasta=abrecovery.fasta-amzon.fasta \n");
+		m->mothurOut("The degap.seqs command should be in the following format: \n");
+		m->mothurOut("degap.seqs(fasta=yourFastaFile) \n");	
+		m->mothurOut("Example: degap.seqs(fasta=abrecovery.align) \n");
+		m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n\n");	
+	}
+	catch(exception& e) {
+		m->errorOut(e, "DegapSeqsCommand", "help");
+		exit(1);
+	}
+}
+
+//***************************************************************************************************************
+
+DegapSeqsCommand::~DegapSeqsCommand(){	/*	do nothing	*/	}
+
+//***************************************************************************************************************
+
+
+int DegapSeqsCommand::execute(){
+	try{
+		
+		if (abort == true) { return 0; }
+		
+		for (int s = 0; s < fastaFileNames.size(); s++) {
+				
+			m->mothurOut("Degapping sequences from " + fastaFileNames[s] + " ..." ); m->mothurOutEndLine();
+			ifstream inFASTA;
+			openInputFile(fastaFileNames[s], inFASTA);
+			
+			ofstream outFASTA;
+			string degapFile = outputDir + getRootName(getSimpleName(fastaFileNames[s])) + "ng.fasta";
+			openOutputFile(degapFile, outFASTA);
+			
+			while(!inFASTA.eof()){
+				if (m->control_pressed) {  inFASTA.close();  outFASTA.close(); remove(degapFile.c_str()); for (int j = 0; j < outputNames.size(); j++) {	remove(outputNames[j].c_str());	} return 0; }
+				 
+				Sequence currSeq(inFASTA);  gobble(inFASTA);
+				if (currSeq.getName() != "") {
+					outFASTA << ">" << currSeq.getName() << endl;
+					outFASTA << currSeq.getUnaligned() << endl;
+				}
+			}
+			inFASTA.close();
+			outFASTA.close();
+			
+			outputNames.push_back(degapFile);
+			
+			if (m->control_pressed) {  remove(degapFile.c_str()); for (int j = 0; j < outputNames.size(); j++) {	remove(outputNames[j].c_str());	} return 0; }
+		}
+		
+		m->mothurOutEndLine();
+		m->mothurOut("Output File Name: "); m->mothurOutEndLine();
+		for (int i = 0; i < outputNames.size(); i++) {	m->mothurOut(outputNames[i]); m->mothurOutEndLine();	}	
+		m->mothurOutEndLine();
+
+		
+		return 0;
+		
+	}
+	catch(exception& e) {
+		m->errorOut(e, "DegapSeqsCommand", "execute");
+		exit(1);
+	}
+}
+
+//***************************************************************************************************************
+
diff --git a/degapseqscommand.h b/degapseqscommand.h
new file mode 100644
index 0000000..192d6eb
--- /dev/null
+++ b/degapseqscommand.h
@@ -0,0 +1,33 @@
+#ifndef DEGAPSEQSCOMMAND_H
+#define DEGAPSEQSCOMMAND_H
+
+/*
+ *  degapseqscommand.h
+ *  Mothur
+ *
+ *  Created by westcott on 6/21/10.
+ *  Copyright 2010 Schloss Lab. All rights reserved.
+ *
+ */
+
+
+#include "command.hpp"
+
+class DegapSeqsCommand : public Command {
+public:
+	DegapSeqsCommand(string);
+	~DegapSeqsCommand();
+	int execute();
+	void help();
+	
+private:
+
+	bool abort;
+	string fastafile, outputDir;
+	vector<string> outputNames;
+	vector<string> fastaFileNames;
+	
+};
+
+#endif
+
diff --git a/getrelabundcommand.cpp b/getrelabundcommand.cpp
new file mode 100644
index 0000000..4fd95b2
--- /dev/null
+++ b/getrelabundcommand.cpp
@@ -0,0 +1,233 @@
+/*
+ *  getrelabundcommand.cpp
+ *  Mothur
+ *
+ *  Created by westcott on 6/21/10.
+ *  Copyright 2010 Schloss Lab. All rights reserved.
+ *
+ */
+
+#include "getrelabundcommand.h"
+
+//**********************************************************************************************************************
+
+GetRelAbundCommand::GetRelAbundCommand(string option) {
+	try {
+		globaldata = GlobalData::getInstance();
+		abort = false;
+		allLines = 1;
+		labels.clear();
+		
+		//allow user to run help
+		if(option == "help") { help(); abort = true; }
+		
+		else {
+			//valid paramters for this command
+			string AlignArray[] =  {"groups","label","scale","outputdir","inputdir"};
+			vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
+			
+			OptionParser parser(option);
+			map<string,string> parameters = parser.getParameters();
+			
+			ValidParameters validParameter;
+			
+			//check to make sure all parameters are valid for command
+			for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) { 
+				if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
+			}
+			
+			//if the user changes the output directory command factory will send this info to us in the output parameter 
+			outputDir = validParameter.validFile(parameters, "outputdir", false);		if (outputDir == "not found"){	
+				outputDir = "";	
+				outputDir += hasPath(globaldata->inputFileName); //if user entered a file with a path then preserve it	
+			}
+			
+			//make sure the user has already run the read.otu command
+			if ((globaldata->getSharedFile() == "")) {
+				 m->mothurOut("You must read a list and a group, or a shared file before you can use the get.relabund command."); m->mothurOutEndLine(); abort = true; 
+			}
+
+			//check for optional parameter and set defaults
+			// ...at some point should added some additional type checking...
+			label = validParameter.validFile(parameters, "label", false);			
+			if (label == "not found") { label = ""; }
+			else { 
+				if(label != "all") {  splitAtDash(label, labels);  allLines = 0;  }
+				else { allLines = 1;  }
+			}
+			
+			//if the user has not specified any labels use the ones from read.otu
+			if (label == "") {  
+				allLines = globaldata->allLines; 
+				labels = globaldata->labels; 
+			}
+			
+			groups = validParameter.validFile(parameters, "groups", false);			
+			if (groups == "not found") { groups = ""; }
+			else { 
+				splitAtDash(groups, Groups);
+				globaldata->Groups = Groups;
+			}
+			
+			scale = validParameter.validFile(parameters, "scale", false);				if (scale == "not found") { scale = "totalgroup"; }
+			
+			if ((scale != "totalgroup") && (scale != "totalgroup") && (scale != "totalgroup") && (scale != "totalgroup")) {
+				m->mothurOut(scale + " is not a valid scaling option for the get.relabund command. Choices are totalgroup, totalotu, averagegroup, averageotu."); m->mothurOutEndLine(); abort = true; 
+			}
+		}
+
+	}
+	catch(exception& e) {
+		m->errorOut(e, "GetRelAbundCommand", "GetRelAbundCommand");
+		exit(1);
+	}
+}
+
+//**********************************************************************************************************************
+
+void GetRelAbundCommand::help(){
+	try {
+		m->mothurOut("The get.relabund command can only be executed after a successful read.otu command of a list and group or shared file.\n");
+		m->mothurOut("The get.relabund command parameters are groups, scale and label.  No parameters are required.\n");
+		m->mothurOut("The groups parameter allows you to specify which of the groups in your groupfile you would like included. The group names are separated by dashes.\n");
+		m->mothurOut("The label parameter allows you to select what distance levels you would like, and are also separated by dashes.\n");
+		m->mothurOut("The scale parameter allows you to select what scale you would like to use. Choices are totalgroup, totalotu, averagegroup, averageotu, default is totalgroup.\n");
+		m->mothurOut("The get.relabund command should be in the following format: get.relabund(groups=yourGroups, label=yourLabels).\n");
+		m->mothurOut("Example get.relabund(groups=A-B-C, scale=log10).\n");
+		m->mothurOut("The default value for groups is all the groups in your groupfile, and all labels in your inputfile will be used.\n");
+		m->mothurOut("The get.relabund command outputs a .relabund file.\n");
+		m->mothurOut("Note: No spaces between parameter labels (i.e. groups), '=' and parameters (i.e.yourGroups).\n\n");
+
+	}
+	catch(exception& e) {
+		m->errorOut(e, "GetRelAbundCommand", "help");
+		exit(1);
+	}
+}
+
+//**********************************************************************************************************************
+
+GetRelAbundCommand::~GetRelAbundCommand(){
+}
+
+//**********************************************************************************************************************
+
+int GetRelAbundCommand::execute(){
+	try {
+	
+		if (abort == true) { return 0; }
+		
+		string outputFileName = outputDir + getRootName(getSimpleName(globaldata->inputFileName)) + "relabund";
+		ofstream out;
+		openOutputFile(outputFileName, out);
+		
+		read = new ReadOTUFile(globaldata->inputFileName);	
+		read->read(&*globaldata); 
+		input = globaldata->ginput;
+		lookup = input->getSharedRAbundVectors();
+		string lastLabel = lookup[0]->getLabel();
+		
+		//if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
+		set<string> processedLabels;
+		set<string> userLabels = labels;
+
+		//as long as you are not at the end of the file or done wih the lines you want
+		while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
+			
+			if (m->control_pressed) {  for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  } globaldata->Groups.clear(); delete read;  out.close(); remove(outputFileName.c_str()); return 0; }
+	
+			if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){			
+
+				m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
+				getRelAbundance(lookup, out);
+				
+				processedLabels.insert(lookup[0]->getLabel());
+				userLabels.erase(lookup[0]->getLabel());
+			}
+			
+			if ((anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
+				string saveLabel = lookup[0]->getLabel();
+			
+				for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }  
+				lookup = input->getSharedRAbundVectors(lastLabel);
+				m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
+				
+				getRelAbundance(lookup, out);
+				
+				processedLabels.insert(lookup[0]->getLabel());
+				userLabels.erase(lookup[0]->getLabel());
+				
+				//restore real lastlabel to save below
+				lookup[0]->setLabel(saveLabel);
+			}
+			
+			lastLabel = lookup[0]->getLabel();
+			//prevent memory leak
+			for (int i = 0; i < lookup.size(); i++) {  delete lookup[i]; lookup[i] = NULL; }
+						
+			//get next line to process
+			lookup = input->getSharedRAbundVectors();				
+		}
+		
+		if (m->control_pressed) { globaldata->Groups.clear(); delete read;  out.close(); remove(outputFileName.c_str());  return 0; }
+
+		//output error messages about any remaining user labels
+		set<string>::iterator it;
+		bool needToRun = false;
+		for (it = userLabels.begin(); it != userLabels.end(); it++) {  
+			m->mothurOut("Your file does not include the label " + *it); 
+			if (processedLabels.count(lastLabel) != 1) {
+				m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
+				needToRun = true;
+			}else {
+				m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
+			}
+		}
+	
+		//run last label if you need to
+		if (needToRun == true)  {
+			for (int i = 0; i < lookup.size(); i++) { if (lookup[i] != NULL) { delete lookup[i]; } }  
+			lookup = input->getSharedRAbundVectors(lastLabel);
+			
+			m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
+			
+			getRelAbundance(lookup, out);
+			
+			for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
+		}
+	
+		//reset groups parameter
+		globaldata->Groups.clear();  
+		delete input; globaldata->ginput = NULL;
+		delete read;
+		out.close();
+		
+		if (m->control_pressed) { remove(outputFileName.c_str()); return 0;}
+		
+		m->mothurOutEndLine();
+		m->mothurOut("Output File Names: "); m->mothurOutEndLine();
+		m->mothurOut(outputFileName); m->mothurOutEndLine();
+		m->mothurOutEndLine();
+		
+		return 0;
+	}
+	catch(exception& e) {
+		m->errorOut(e, "GetRelAbundCommand", "execute");
+		exit(1);
+	}
+}
+//**********************************************************************************************************************
+
+int GetRelAbundCommand::getRelAbundance(vector<SharedRAbundVector*> thisLookUp, ofstream& out){
+	try {
+	
+	
+	}
+	catch(exception& e) {
+		m->errorOut(e, "GetRelAbundCommand", "getRelAbundance");
+		exit(1);
+	}
+}
+//**********************************************************************************************************************
+
+
diff --git a/getrelabundcommand.h b/getrelabundcommand.h
new file mode 100644
index 0000000..aa123a4
--- /dev/null
+++ b/getrelabundcommand.h
@@ -0,0 +1,44 @@
+#ifndef GETRELABUNDCOMMAND_H
+#define GETRELABUNDCOMMAND_H
+
+/*
+ *  getrelabundcommand.h
+ *  Mothur
+ *
+ *  Created by westcott on 6/21/10.
+ *  Copyright 2010 Schloss Lab. All rights reserved.
+ *
+ */
+ 
+#include "command.hpp"
+#include "inputdata.h"
+#include "readotu.h"
+#include "sharedrabundvector.h"
+
+class GlobalData;
+
+class GetRelAbundCommand : public Command {
+
+public:
+	GetRelAbundCommand(string);
+	~GetRelAbundCommand();
+	int execute();
+	void help();
+	
+private:
+	GlobalData* globaldata;
+	ReadOTUFile* read;
+	InputData* input;
+	vector<SharedRAbundVector*> lookup;
+	
+	bool abort, allLines;
+	set<string> labels; //holds labels to be used
+	string groups, label, outputDir, scale;
+	vector<string> Groups;
+	
+	int getRelAbundance(vector<SharedRAbundVector*>, ofstream&);
+
+};
+
+#endif
+
diff --git a/makefile b/makefile
index 8fefa14..08dee23 100644
--- a/makefile
+++ b/makefile
@@ -33,7 +33,7 @@ ifeq  ($(strip $(USEREADLINE)),yes)
       -L../readline-6.0
 endif
 
-USEMPI ?= yes
+USEMPI ?= no
 
 ifeq  ($(strip $(USEMPI)),yes)
 	CC = mpic++
@@ -228,6 +228,7 @@ mothur : \
 		./hclustercommand.o\
 		./hcluster.o\
 		./getlistcountcommand.o\
+		./getrelabundcommand.o\
 		./readcluster.o\
 		./ccode.o\
 		./taxonomyequalizer.o\
@@ -439,6 +440,7 @@ mothur : \
 		./hclustercommand.o\
 		./hcluster.o\
 		./getlistcountcommand.o\
+		./getrelabundcommand.o\
 		./readcluster.o\
 		./ccode.o\
 		./taxonomyequalizer.o\
@@ -653,6 +655,7 @@ clean :
 		./hclustercommand.o\
 		./hcluster.o\
 		./getlistcountcommand.o\
+		./getrelabundcommand.o\
 		./readcluster.o\
 		./ccode.o\
 		./taxonomyequalizer.o\
@@ -1726,4 +1729,8 @@ install : mothur
 ./degapseqscommand.o : degapseqscommand.cpp
 	$(CC) $(CC_OPTIONS) degapseqscommand.cpp -c $(INCLUDE) -o ./degapseqscommand.o
 
+# Item # 211 -- getrelabundcommand --
+./getrelabundcommand.o : getrelabundcommand.cpp
+	$(CC) $(CC_OPTIONS) getrelabundcommand.cpp -c $(INCLUDE) -o ./getrelabundcommand.o
+
 ##### END RUN ####