X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=globaldata.cpp;h=6138e2cc1b8f4fc70b8febc1105fcecca0448306;hb=4f4070ac3fbb5da75f834b48062e9123364b006e;hp=3ddd02e5c24dc03cd8ff7549b6b03a32982562e9;hpb=39ef0543b73cb12dfc64529522b66ad3df7527d0;p=mothur.git

diff --git a/globaldata.cpp b/globaldata.cpp
index 3ddd02e..6138e2c 100644
--- a/globaldata.cpp
+++ b/globaldata.cpp
@@ -1,9 +1,6 @@
 #include "globaldata.hpp"
-#include "sparsematrix.hpp"
 #include "tree.h"
-#include "rabundvector.hpp"
-#include "sabundvector.hpp"
-#include "listvector.hpp"
+#include "sparsematrix.hpp"
 
 /*******************************************************/
 
@@ -20,7 +17,6 @@ GlobalData* GlobalData::getInstance() {
 //This function parses through the option string of the command to remove its parameters
 void GlobalData::parseGlobalData(string commandString, string optionText){
 	try {
-		allLines = 1;
 		commandName = commandString; //save command name to be used by other classes
 		
 		//set all non filename paramters to default
@@ -30,8 +26,12 @@ void GlobalData::parseGlobalData(string commandString, string optionText){
 		if ((commandName == "read.dist") || (commandName == "read.otu") || (commandName == "read.tree")) { 
 			clear();
 			gGroupmap = NULL;
+			gListVector = NULL;
+			gSparseMatrix = NULL;
 			gTree.clear();
-			labels.clear(); lines.clear(); groups.clear();
+			Treenames.clear();
+			labels.clear(); lines.clear(); Groups.clear();
+			allLines = 1;
 		}
 		
 		//saves help request
@@ -39,6 +39,14 @@ void GlobalData::parseGlobalData(string commandString, string optionText){
 			helpRequest = optionText;
 		}
 		
+		if (commandName == "libshuff") {
+			iters = "10000";
+			cutoff = "1.0";
+		}
+		
+		//set default value for cutoff
+		if (commandName == "dist.seqs") {	cutoff = "1.0";		}
+
 		string key, value;		
 		//reads in parameters and values
 		if((optionText != "") && (commandName != "help")){
@@ -51,7 +59,7 @@ void GlobalData::parseGlobalData(string commandString, string optionText){
 				if (key == "list" )		{ listfile = value; inputFileName = value; fileroot = value; format = "list";		}
 				if (key == "rabund" )	{ rabundfile = value; inputFileName = value; fileroot = value; format = "rabund";	}
 				if (key == "sabund" )	{ sabundfile = value; inputFileName = value; fileroot = value; format = "sabund";	} 
-				if (key == "fasta" )	{ fastafile = value; inputFileName = value; fileroot = value; format = "fasta";		} 
+				if (key == "fasta" )	{ fastafile = value; inputFileName = value; fileroot = value; format = "fasta";		}
 				if (key == "tree" )		{ treefile = value; inputFileName = value; fileroot = value; format = "tree";		}
 				if (key == "shared" )	{ sharedfile = value; inputFileName = value; fileroot = value; format = "sharedfile";	}
 				if (key == "name" )		{ namefile = value;		}
@@ -66,22 +74,49 @@ void GlobalData::parseGlobalData(string commandString, string optionText){
 				if (key == "fileroot" )		{ fileroot = value;		}
 				if (key == "abund" )        { abund = value;        }
 				if (key == "random" )		{ randomtree = value;	}
-				if (key == "calc")			{ calc = value;		}
-
+				if (key == "calc")			{ calc = value;			}
+				if (key == "step")			{ step = value;			}
+				if (key == "form")			{ form = value;			}
+				if (key == "sorted")		{ sorted = value;		}
+				if (key == "vertical")		{ vertical = value;		}
+				if (key == "trump")		    { trump = value;		}
+				if (key == "hard")			{ hard = value;		}
+				if (key == "soft")		    { soft = value;		    }
+				if (key == "scale")			{ scale = value;		}
+				if (key == "countends" )	{ countends = value;	}
+				if (key == "processors" )	{ processors = value;	}
+				if (key == "size" )         { size = value;         }
+				if (key == "candidate")		{ candidatefile = value;	}
+				if (key == "search")		{ search = value;		}
+				if (key == "ksize")			{ ksize = value;		}
+				if (key == "align")		    { align = value;		}
+				if (key == "match")			{ match = value;		}
+				if (key == "mismatch")		{ mismatch = value;	    }
+				if (key == "gapopen")		{ gapopen = value;		}
+				if (key == "gapextend" )	{ gapextend = value;	}
+				if (key == "start" )		{ startPos = value;	}
+				if (key == "end" )			{ endPos = value;	}
+				if (key == "maxambig" )		{ maxAmbig = value;	}
+				if (key == "maxhomop" )		{ maxHomoPolymer = value;	}
+				if (key == "minlength" )	{ minLength = value;	}
+				if (key == "maxlength" )	{ maxLength = value;	}
 				
-				if (key == "line") {//stores lines to be used in a set
+				if (key == "line") {//stores lines to be used in a vector
 					lines.clear();
+					labels.clear();
 					line = value;
 					label = "";
-					splitAtDash(value, lines);
-					allLines = 0;
+					if (line != "all") {  splitAtDash(value, lines);  allLines = 0;  }
+					else { allLines = 1;  }
 				}
-				if (key == "label") {//stores labels to be used in a set
+			
+				if (key == "label") {//stores lines to be used in a vector
 					labels.clear();
+					lines.clear();
 					label = value;
 					line = "";
-					splitAtDash(value, labels);
-					allLines = 0;
+					if (label != "all") {  splitAtDash(value, labels);  allLines = 0;  }
+					else { allLines = 1;  }
 				}
 
 				if (key == "groups") {//stores groups to be used in a vector
@@ -102,7 +137,7 @@ void GlobalData::parseGlobalData(string commandString, string optionText){
 			if (key == "sabund" )	{ sabundfile = value; inputFileName = value; fileroot = value; format = "sabund";	}
 			if (key == "fasta" )	{ fastafile = value; inputFileName = value; fileroot = value; format = "fasta";		}
 			if (key == "tree" )		{ treefile = value; inputFileName = value; fileroot = value; format = "tree";		} 
-			if (key == "shared" )	{ sharedfile = value; inputFileName = value; fileroot = value; format = "sharedfile";	} 
+			if (key == "shared" )	{ sharedfile = value; inputFileName = value; fileroot = value; format = "sharedfile";	}
 			if (key == "name" )		{ namefile = value;		}
 			if (key == "order" )	{ orderfile = value;	}
 			if (key == "group" )	{ groupfile = value;	}
@@ -115,23 +150,52 @@ void GlobalData::parseGlobalData(string commandString, string optionText){
 			if (key == "fileroot" )		{ fileroot = value;		}
 			if (key == "abund" )        { abund = value;        }
 			if (key == "random" )		{ randomtree = value;	}
-			if (key == "calc")			{ calc = value;		}
+			if (key == "calc")			{ calc = value;			}
+			if (key == "step")			{ step = value;			}
+			if (key == "form")			{ form = value;			}
+			if (key == "sorted")		{ sorted = value;		}
+			if (key == "vertical")		{ vertical = value;		}
+			if (key == "trump")		    { trump = value;		}
+			if (key == "hard")			{ hard = value;		}
+			if (key == "soft")		    { soft = value;		    }
+			if (key == "scale")			{ scale = value;		}
+			if (key == "countends" )	{ countends = value;	}
+			if (key == "processors" )	{ processors = value;	}
+			if (key == "size" )         { size = value;         }
+			if (key == "candidate")		{ candidatefile = value;	}
+			if (key == "search")		{ search = value;		}
+			if (key == "ksize")			{ ksize = value;		}
+			if (key == "align")		    { align = value;		}
+			if (key == "match")			{ match = value;		}
+			if (key == "mismatch")		{ mismatch = value;	    }
+			if (key == "gapopen")		{ gapopen = value;		}
+			if (key == "gapextend" )	{ gapextend = value;	}
+			if (key == "start" )		{ startPos = value;	}
+			if (key == "end" )			{ endPos = value;	}
+			if (key == "maxambig" )		{ maxAmbig = value;	}
+			if (key == "maxhomop" )		{ maxHomoPolymer = value;	}
+			if (key == "minlength" )	{ minLength = value;	}
+			if (key == "maxlength" )	{ maxLength = value;	}
 
 
 			if (key == "line") {//stores lines to be used in a vector
 				lines.clear();
+				labels.clear();
 				line = value;
 				label = "";
-				splitAtDash(value, lines);
-				allLines = 0;
+				if (line != "all") {  splitAtDash(value, lines);  allLines = 0;  }
+				else { allLines = 1;  }
 			}
+			
 			if (key == "label") {//stores lines to be used in a vector
 				labels.clear();
+				lines.clear();
 				label = value;
 				line = "";
-				splitAtDash(value, labels);
-				allLines = 0;
+				if (label != "all") {  splitAtDash(value, labels);  allLines = 0;  }
+				else { allLines = 1;  }
 			}
+			
 			if (key == "groups") {//stores groups to be used in a vector
 					Groups.clear();
 					groups = value;
@@ -145,7 +209,8 @@ void GlobalData::parseGlobalData(string commandString, string optionText){
 				
 		//input defaults for calculators
 		if (commandName == "collect.single") {
-			if ((calc == "default") || (calc == "")) { calc = "sobs-chao-ace-jack-bootstrap-shannon-npshannon-simpson"; }
+
+			if ((calc == "default") || (calc == "")) { calc = "sobs-chao-ace-jack-shannon-npshannon-simpson"; }
 			Estimators.clear();
 			splitAtDash(calc, Estimators); 
 		}
@@ -155,17 +220,17 @@ void GlobalData::parseGlobalData(string commandString, string optionText){
 			splitAtDash(calc, Estimators); 
 		}
 		if (commandName == "collect.shared") {
-			if ((calc == "default") || (calc == "")) { calc = "sharedsobs-sharedchao-sharedace-sharedjabund-sharedsorensonabund-sharedjclass-sharedsorclass-sharedjest-sharedsorest-sharedthetayc-sharedthetan"; }
+			if ((calc == "default") || (calc == "")) { calc = "sharedsobs-sharedchao-sharedace-jabund-sorabund-jclass-sorclass-jest-sorest-thetayc-thetan"; }
 			Estimators.clear();
 			splitAtDash(calc, Estimators); 
 		}
 		if (commandName == "summary.single") {
-			if ((calc == "default") || (calc == "")) { calc = "sobs-chao-ace-jack-bootstrap-shannon-npshannon-simpson"; }
+			if ((calc == "default") || (calc == "")) { calc = "sobs-chao-ace-jack-shannon-npshannon-simpson"; }
 			Estimators.clear();
 			splitAtDash(calc, Estimators); 
 		}
 		if (commandName == "summary.shared") {
-			if ((calc == "default") || (calc == "")) { calc = "sharedsobs-sharedchao-sharedace-sharedjabund-sharedsorensonabund-sharedjclass-sharedsorclass-sharedjest-sharedsorest-sharedthetayc-sharedthetan"; }
+			if ((calc == "default") || (calc == "")) { calc = "sharedsobs-sharedchao-sharedace-jabund-sorabund-jclass-sorclass-jest-sorest-thetayc-thetan"; }
 			Estimators.clear();
 			splitAtDash(calc, Estimators); 
 		}
@@ -174,6 +239,34 @@ void GlobalData::parseGlobalData(string commandString, string optionText){
 			Estimators.clear();
 			splitAtDash(calc, Estimators); 
 		}
+		if (commandName == "dist.seqs") {
+			if ((calc == "default") || (calc == "")) {  calc = "onegap";	}
+			if (countends == "")	{ countends = "T"; }
+			Estimators.clear();
+			splitAtDash(calc, Estimators); 
+		}
+		if (commandName == "venn") {
+			if ((calc == "default") || (calc == "")) { 
+				if (format == "list") { calc = "sobs"; }
+				else { calc = "sharedsobs"; }
+			}
+			Estimators.clear();
+			splitAtDash(calc, Estimators); 
+		}
+		if ((commandName == "tree.shared") || (commandName == "bootstrap.shared") || (commandName == "dist.shared")) {
+			if ((calc == "default") || (calc == "")) { 
+				calc = "jclass-thetayc";
+			}
+			Estimators.clear();
+			splitAtDash(calc, Estimators); 
+		}
+		
+		if(commandName == "filter.seqs"){
+			if(trump == "" && vertical == "" && hard == "" && soft == ""){
+				trump = '.';
+			}
+
+		}
 
 		//if you have done a read.otu with a groupfile but don't want to use it anymore because you want to do single commands
 		if ((commandName == "collect.single") || (commandName == "rarefaction.single") || (commandName == "summary.single")) {
@@ -217,16 +310,48 @@ string GlobalData::getFreq()			{	return freq;		}
 string GlobalData::getAbund()           {   return abund;       }
 string GlobalData::getRandomTree()		{	return randomtree;	}
 string GlobalData::getGroups()			{	return groups;		}
-void GlobalData::setListFile(string file)	{	listfile = file;	inputFileName = file;}
-void GlobalData::setRabundFile(string file)	{	rabundfile = file;	inputFileName = file;}
-void GlobalData::setSabundFile(string file)	{	sabundfile = file;	inputFileName = file;}
-void GlobalData::setPhylipFile(string file)	{	phylipfile = file;    inputFileName = file;}
-void GlobalData::setColumnFile(string file)	{	columnfile = file;    inputFileName = file;}
+string GlobalData::getStep()			{	return step;		}
+string GlobalData::getForm()			{	return form;		}
+string GlobalData::getSorted()			{	return sorted;		}
+string GlobalData::getVertical()		{   return vertical;    }
+string GlobalData::getTrump()			{   return trump;       }
+string GlobalData::getSoft()			{   return soft;		}
+string GlobalData::getHard()			{   return hard;		}
+string GlobalData::getScale()			{	return scale;		}
+string GlobalData::getCountEnds()		{   return countends;	}
+string GlobalData::getProcessors()		{	return processors;	}
+string GlobalData::getSize()            {   return size;        }
+string GlobalData::getCandidateFile()	{	return candidatefile;}
+string GlobalData::getSearch()			{	return search;		}
+string GlobalData::getKSize()			{	return ksize;		}
+string GlobalData::getAlign()			{	return align;		}
+string GlobalData::getMatch()			{	return match;		}
+string GlobalData::getMismatch()		{	return mismatch;	}
+string GlobalData::getGapopen()			{	return gapopen;		}
+string GlobalData::getGapextend()		{	return gapextend;	}
+string GlobalData::getStartPos()		{	return startPos;	}
+string GlobalData::getEndPos()			{	return endPos;		}
+string GlobalData::getMaxAmbig()		{	return maxAmbig;	}
+string GlobalData::getMaxHomoPolymer()	{	return maxHomoPolymer;	}
+string GlobalData::getMinLength()		{	return minLength;	}
+string GlobalData::getMaxLength()		{	return maxLength;	}
+
+
+void GlobalData::setListFile(string file)		{	listfile = file;	inputFileName = file;					}
+void GlobalData::setRabundFile(string file)		{	rabundfile = file;	inputFileName = file;					}
+void GlobalData::setSabundFile(string file)		{	sabundfile = file;	inputFileName = file;					}
+void GlobalData::setPhylipFile(string file)		{	phylipfile = file;    inputFileName = file;					}
+void GlobalData::setColumnFile(string file)		{	columnfile = file;    inputFileName = file;					}
+void GlobalData::setGroupFile(string file)		{	groupfile = file;											}
+void GlobalData::setSharedFile(string file)		{	sharedfile = file;	inputFileName = file; fileroot = file;	}
 void GlobalData::setNameFile(string file)		{	namefile = file;		}
 void GlobalData::setFormat(string Format)		{	format = Format;		}
 void GlobalData::setRandomTree(string Random)	{	randomtree = Random;	}
 void GlobalData::setGroups(string g)			{	groups = g;				}
 void GlobalData::setCalc(string Calc)			{	calc = Calc;			}
+void GlobalData::setCountEnds(string e)			{   countends = e;			}
+void GlobalData::setProcessors(string p)		{	processors = p;			}
+
 
 /*******************************************************/
 
@@ -235,6 +360,8 @@ GlobalData::GlobalData() {
 	//option definitions should go here...
 	helpRequest = "";
 	clear();
+	gListVector = NULL;		
+	gSparseMatrix = NULL;	
 }
 /*******************************************************/
 
@@ -252,6 +379,7 @@ void GlobalData::clear() {
 	fastafile		=   "";
 	treefile		=	"";
 	sharedfile		=	"";
+	candidatefile	=	"";
 	cutoff			=	"10.00";
 	format			=	"";
 	precision		=	"100";
@@ -265,24 +393,73 @@ void GlobalData::clear() {
 	method			=	"furthest";
 	fileroot		=	"";
 	abund           =   "10";
+	step			=	"0.01";
+	form			=	"integral";
+	sorted			=	"T";  //F means don't sort, T means sort.
+	vertical        =   "F";		
+	trump           =	"";		
+	hard			=   "";		
+	soft            =   "";	
+	scale			=	"log10";
+	countends		=   "T";  //yes
+	processors		=	"1";
+	size            =   "0";
+	search			=	"kmer";
+	ksize			=	"8";
+	align			=	"needleman";
+	match			=	"1.0";
+	mismatch		=	"-1.0";
+	gapopen			=	"-1.0";
+	gapextend		=	"-2.0";
+	startPos		=	"-1";
+	endPos			=	"-1";
+	maxAmbig		=	"-1";
+	maxHomoPolymer	=	"-1";
+	minLength		=	"-1";
+	maxLength		=	"-1";
+	
 }
 
 //*******************************************************/
 
 /******************************************************/
 void GlobalData::reset() {
+	label			=	"";
+	line			=   "";
 	cutoff			=	"10.00";
 	precision		=	"100";
 	iters			=	"1000"; 
-	line			=   "";
-	label			=	"";
 	groups			=	"";
 	jumble			=	"1";	//0 means don't jumble, 1 means jumble.
+	sorted			=	"T";  //F means don't sort, T means sort.
 	randomtree		=	"";  //"" means user will enter some user trees, "outputfile" means they just want the random tree distribution to be outputted to outputfile.
 	freq			=	"100";
 	method			=	"furthest";
 	calc			=	"";
-	abund = "10";
+	abund			=   "10";
+	step			=	"0.01";
+	form			=	"integral";
+	countends		=   "T";
+	processors		=	"1";
+	size            =   "0";
+	search			=	"kmer";
+	ksize			=	"8";
+	align			=	"needleman";
+	match			=	"1.0";
+	mismatch		=	"-1.0";
+	gapopen			=	"-1.0";
+	gapextend		=	"-2.0";
+	vertical        =   "";		
+	trump           =   "";		
+	hard			=   "";		
+	soft            =   "";	
+	startPos		=	"-1";
+	endPos			=	"-1";
+	maxAmbig		=	"-1";
+	maxHomoPolymer	=	"-1";
+	minLength		=	"-1";
+	maxLength		=	"-1";
+	
 }
 /*******************************************************/
 
@@ -294,3 +471,146 @@ GlobalData::~GlobalData() {
 	if(gorder != NULL)			{	delete gorder;		}
 }
 /*******************************************************/
+
+/*******************************************************/
+void GlobalData::parseTreeFile() {
+	//Why is THIS in GlobalData??? - PDS
+	
+	//only takes names from the first tree and assumes that all trees use the same names.
+	try {
+		string filename = treefile;
+		ifstream filehandle;
+		openInputFile(filename, filehandle);
+		int c, comment;
+		comment = 0;
+		
+		//if you are not a nexus file 
+		if ((c = filehandle.peek()) != '#') {  
+			while((c = filehandle.peek()) != ';') { 
+				while ((c = filehandle.peek()) != ';') {
+					// get past comments
+					if(c == '[') {
+						comment = 1;
+					}
+					if(c == ']'){
+						comment = 0;
+					}
+					if((c == '(') && (comment != 1)){ break; }
+					filehandle.get();
+				}
+
+				readTreeString(filehandle); 
+			}
+		//if you are a nexus file
+		}else if ((c = filehandle.peek()) == '#') {
+			string holder = "";
+					
+			// get past comments
+			while(holder != "translate" && holder != "Translate"){	
+				if(holder == "[" || holder == "[!"){
+					comment = 1;
+				}
+				if(holder == "]"){
+					comment = 0;
+				}
+				filehandle >> holder; 
+	
+				//if there is no translate then you must read tree string otherwise use translate to get names
+				if(holder == "tree" && comment != 1){	
+					//pass over the "tree rep.6878900 = "
+					while (((c = filehandle.get()) != '(') && ((c = filehandle.peek()) != EOF) ) {;}
+
+					if (c == EOF ) { break; }
+					filehandle.putback(c);  //put back first ( of tree.
+					readTreeString(filehandle);	
+					break;
+				}
+			}
+			
+			//use nexus translation rather than parsing tree to save time
+			if ((holder == "translate") || (holder == "Translate")) {
+
+				string number, name, h;
+				h = ""; // so it enters the loop the first time
+				while((h != ";") && (number != ";")) { 
+					filehandle >> number;
+					filehandle >> name;
+	
+					//c = , until done with translation then c = ;
+					h = name.substr(name.length()-1, name.length()); 
+					name.erase(name.end()-1);  //erase the comma
+					Treenames.push_back(number);
+				}
+				if (number == ";") { Treenames.pop_back(); }  //in case ';' from translation is on next line instead of next to last name
+			}
+		}
+		
+	}
+	catch(exception& e) {
+		cout << "Standard Error: " << e.what() << " has occurred in the GlobalData class Function parseTreeFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+		exit(1);
+	}
+	catch(...) {
+		cout << "An unknown error has occurred in the GlobalData class function parseTreeFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+		exit(1);
+	}		
+}
+/*******************************************************/
+
+/*******************************************************/
+void GlobalData::readTreeString(ifstream& filehandle)	{
+	try {
+		int c;
+		string name; //k
+		
+		while((c = filehandle.peek()) != ';') { 
+				//if you are a name
+			if ((c != '(') && (c != ')') && (c != ',') && (c != ':') && (c != '\n') && (c != '\t') && (c != 32)) { //32 is space
+				name = "";
+				c = filehandle.get();
+	//		k = c;
+//cout << k << endl;
+				while ((c != '(') && (c != ')') && (c != ',') && (c != ':') && (c != '\n') && (c != 32) && (c != '\t')) {			
+					name += c;
+					c = filehandle.get();
+		//	k = c;
+//cout << " in name while " << k << endl;
+				}
+				
+//cout << "name = " << name << endl;
+				Treenames.push_back(name);
+				filehandle.putback(c);
+//k = c;
+//cout << " after putback" <<  k << endl;
+			} 
+			
+			if (c  == ':') { //read until you reach the end of the branch length
+				while ((c != '(') && (c != ')') && (c != ',') && (c != ';') && (c != '\n') && (c != '\t') && (c != 32)) {
+					c = filehandle.get();
+				//	k = c;
+	//cout << " in branch while " << k << endl;
+				}
+				filehandle.putback(c);
+			}
+			c = filehandle.get();
+			if (c == ';') { break; }
+		//	k = c;
+//cout << k << endl;
+
+		}
+	}
+	catch(exception& e) {
+		cout << "Standard Error: " << e.what() << " has occurred in the GlobalData class Function parseTreeFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+		exit(1);
+	}
+	catch(...) {
+		cout << "An unknown error has occurred in the GlobalData class function parseTreeFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+		exit(1);
+	}		
+}	
+
+/*******************************************************/
+
+/*******************************************************/
+
+