X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=splitmatrix.cpp;h=384b09af1bb94be09c5607c8b863b28b16215731;hb=8e67e9de1b200106bea5a468ac02125954656499;hp=9e53c51a8f1b6fbba8e755279e68993fb7f03fa5;hpb=a0f87c2ae6414af28d4e70b1e6830401eac21bef;p=mothur.git

diff --git a/splitmatrix.cpp b/splitmatrix.cpp
index 9e53c51..384b09a 100644
--- a/splitmatrix.cpp
+++ b/splitmatrix.cpp
@@ -10,6 +10,7 @@
 #include "splitmatrix.h"
 #include "phylotree.h"
 #include "distancecommand.h"
+#include "seqsummarycommand.h"
 
 /***********************************************************************/
 
@@ -24,14 +25,17 @@ SplitMatrix::SplitMatrix(string distfile, string name, string tax, float c, stri
 }
 /***********************************************************************/
 
-SplitMatrix::SplitMatrix(string ffile, string name, string tax, float c, string t, int p){
+SplitMatrix::SplitMatrix(string ffile, string name, string tax, float c, float cu, string t, int p, bool cl, string output){
 	m = MothurOut::getInstance();
 	fastafile = ffile;
 	namefile = name;
 	taxFile = tax;
-	cutoff = c;
+	cutoff = c;  //tax level cutoff
+	distCutoff = cu; //for fasta method if you are creating distance matrix you need a cutoff for that
 	method = t;
 	processors = p;
+    classic = cl;
+	outputDir = output;
 }
 
 /***********************************************************************/
@@ -63,6 +67,8 @@ int SplitMatrix::splitDistance(){
         
 		if (large)	{ splitDistanceLarge(); }
 		else		{ splitDistanceRAM();	}
+		
+		return 0;
 			
 	}
 	catch(exception& e) {
@@ -75,7 +81,7 @@ int SplitMatrix::splitDistance(){
 int SplitMatrix::splitClassify(){
 	try {
 		cutoff = int(cutoff);
-		
+				
 		map<string, int> seqGroup;
 		map<string, int>::iterator it;
 		map<string, int>::iterator it2;
@@ -85,16 +91,13 @@ int SplitMatrix::splitClassify(){
 		//build tree from users taxonomy file
 		PhyloTree* phylo = new PhyloTree();
 		
-		ifstream in;
-		openInputFile(taxFile, in);
-			
-		//read in users taxonomy file and add sequences to tree
-		string seqname, tax;
-		while(!in.eof()){
-			in >> seqname >> tax; gobble(in);
-			phylo->addSeqToTree(seqname, tax);
-		}
-		in.close();
+        map<string, string> temp;
+        m->readTax(taxFile, temp);
+        
+        for (map<string, string>::iterator itTemp = temp.begin(); itTemp != temp.end();) {
+            phylo->addSeqToTree(itTemp->first, itTemp->second);
+            temp.erase(itTemp++);
+        }
 		
 		phylo->assignHeirarchyIDs(0);
 
@@ -141,16 +144,16 @@ int SplitMatrix::createDistanceFilesFromTax(map<string, int>& seqGroup, int numG
 		set<string> names;
 				
 		for (int i = 0; i < numGroups; i++) { //remove old temp files, just in case
-			remove((fastafile + "." + toString(i) + ".temp").c_str());
+			m->mothurRemove((fastafile + "." + toString(i) + ".temp"));
 		}
 			
 		ifstream in;
-		openInputFile(fastafile, in);
+		m->openInputFile(fastafile, in);
 	
 		//parse fastafile
 		ofstream outFile;
 		while (!in.eof()) {
-			Sequence query(in); gobble(in);
+			Sequence query(in); m->gobble(in);
 			if (query.getName() != "") {
 		
 				it = seqGroup.find(query.getName());
@@ -159,11 +162,11 @@ int SplitMatrix::createDistanceFilesFromTax(map<string, int>& seqGroup, int numG
 				if (namefile == "") {  names.insert(query.getName()); }
 			
 				if (it != seqGroup.end()) { //not singleton 
-					openOutputFileAppend((fastafile + "." + toString(it->second) + ".temp"), outFile);
+					m->openOutputFileAppend((fastafile + "." + toString(it->second) + ".temp"), outFile);
 					query.printSequence(outFile); 
 					outFile.close();
 					
-					copyGroups.erase(it);
+					copyGroups.erase(query.getName());
 				}
 			}
 		}
@@ -180,36 +183,40 @@ int SplitMatrix::createDistanceFilesFromTax(map<string, int>& seqGroup, int numG
 		//process each distance file
 		for (int i = 0; i < numGroups; i++) { 
 			
-			string options = "fasta=" + (fastafile + "." + toString(i) + ".temp") + ", processors=" + toString(processors);
+			string options = "";
+            if (classic) { options = "fasta=" + (fastafile + "." + toString(i) + ".temp") + ", processors=" + toString(processors) + ", output=lt"; }
+            else { options = "fasta=" + (fastafile + "." + toString(i) + ".temp") + ", processors=" + toString(processors) + ", cutoff=" + toString(distCutoff); }
+			if (outputDir != "") { options += ", outputdir=" + outputDir; }
 			
 			Command* command = new DistanceCommand(options);
+			
 			command->execute();
 			delete command;
 			
-			remove((fastafile + "." + toString(i) + ".temp").c_str());
+			m->mothurRemove((fastafile + "." + toString(i) + ".temp"));
 			
 			//remove old names files just in case
-			remove((namefile + "." + toString(i) + ".temp").c_str());
+			m->mothurRemove((namefile + "." + toString(i) + ".temp"));
 		}
-		
+			
 		singleton = namefile + ".extra.temp";
 		ofstream remainingNames;
-		openOutputFile(singleton, remainingNames);
+		m->openOutputFile(singleton, remainingNames);
 		
 		bool wroteExtra = false;
 
 		ifstream bigNameFile;
-		openInputFile(namefile, bigNameFile);
+		m->openInputFile(namefile, bigNameFile);
 		
 		string name, nameList;
 		while(!bigNameFile.eof()){
-			bigNameFile >> name >> nameList;  gobble(bigNameFile);
+			bigNameFile >> name >> nameList;  m->gobble(bigNameFile);
 			
 			//did this sequence get assigned a group
 			it = seqGroup.find(name);
 			
 			if (it != seqGroup.end()) {  
-				openOutputFileAppend((namefile + "." + toString(it->second) + ".temp"), outFile);
+				m->openOutputFileAppend((namefile + "." + toString(it->second) + ".temp"), outFile);
 				outFile << name << '\t' << nameList << endl;
 				outFile.close();
 			}else{
@@ -219,31 +226,45 @@ int SplitMatrix::createDistanceFilesFromTax(map<string, int>& seqGroup, int numG
 		}
 		bigNameFile.close();
 		
-		remainingNames.close();
-		if (!wroteExtra) { 
-			remove(singleton.c_str());
-			singleton = "none";
-		}
-
 		for(int i=0;i<numGroups;i++){
 			string tempNameFile = namefile + "." + toString(i) + ".temp";
-			string tempDistFile = getRootName(getSimpleName((fastafile + "." + toString(i) + ".temp"))) + "dist";
+			if (outputDir == "") { outputDir = m->hasPath(fastafile); }
+			string tempDistFile = "";
+            if (classic) { tempDistFile =  outputDir + m->getRootName(m->getSimpleName((fastafile + "." + toString(i) + ".temp"))) + "phylip.dist";}
+            else { tempDistFile = outputDir + m->getRootName(m->getSimpleName((fastafile + "." + toString(i) + ".temp"))) + "dist"; }
 
 			//if there are valid distances
 			ifstream fileHandle;
 			fileHandle.open(tempDistFile.c_str());
 			if(fileHandle) 	{	
-				gobble(fileHandle);
-				if (!fileHandle.eof()) {  //check for blank file
+				m->gobble(fileHandle);
+				if (!fileHandle.eof()) {  //check for blank file - this could occur if all dists in group are above cutoff
 					map<string, string> temp;
 					temp[tempDistFile] = tempNameFile;
 					dists.push_back(temp);
+				}else {
+					ifstream in;
+					m->openInputFile(tempNameFile, in);
+				
+					while(!in.eof()) { 
+						in >> name >> nameList;  m->gobble(in);
+						wroteExtra = true;
+						remainingNames << name << '\t' << nameList << endl;
+					}
+					in.close();
+					m->mothurRemove(tempNameFile);
 				}
 			}
 			fileHandle.close();
 		}
 		
-		if (m->control_pressed)	 {  for (int i = 0; i < dists.size(); i++) { remove((dists[i].begin()->first).c_str()); remove((dists[i].begin()->second).c_str()); } dists.clear(); }
+		remainingNames.close();
+		if (!wroteExtra) { 
+			m->mothurRemove(singleton);
+			singleton = "none";
+		}
+
+		if (m->control_pressed)	 {  for (int i = 0; i < dists.size(); i++) { m->mothurRemove((dists[i].begin()->first)); m->mothurRemove((dists[i].begin()->second)); } dists.clear(); }
 		
 		return 0;
 	}
@@ -259,11 +280,11 @@ int SplitMatrix::splitDistanceFileByTax(map<string, int>& seqGroup, int numGroup
 		map<string, int>::iterator it2;
 		
 		ifstream dFile;
-		openInputFile(distFile, dFile);
+		m->openInputFile(distFile, dFile);
 		ofstream outFile;
 		
 		for (int i = 0; i < numGroups; i++) { //remove old temp files, just in case
-			remove((distFile + "." + toString(i) + ".temp").c_str());
+			m->mothurRemove((distFile + "." + toString(i) + ".temp"));
 		}
 		
 		//for buffering the io to improve speed
@@ -280,9 +301,9 @@ int SplitMatrix::splitDistanceFileByTax(map<string, int>& seqGroup, int numGroup
 			string seqA, seqB;
 			float dist;
 			
-			if (m->control_pressed) { dFile.close(); for (int i = 0; i < numGroups; i++) { remove((distFile + "." + toString(i) + ".temp").c_str());	} }
+			if (m->control_pressed) { dFile.close(); for (int i = 0; i < numGroups; i++) { m->mothurRemove((distFile + "." + toString(i) + ".temp"));	} }
 			
-			dFile >> seqA >> seqB >> dist;  gobble(dFile);
+			dFile >> seqA >> seqB >> dist;  m->gobble(dFile);
 			
 			//if both sequences are in the same group then they are within the cutoff
 			it = seqGroup.find(seqA);
@@ -291,7 +312,7 @@ int SplitMatrix::splitDistanceFileByTax(map<string, int>& seqGroup, int numGroup
 			if ((it != seqGroup.end()) && (it2 != seqGroup.end())) { //they are both not singletons 
 				if (it->second == it2->second) { //they are from the same group so add the distance
 					if (numOutputs[it->second] > 30) {
-						openOutputFileAppend((distFile + "." + toString(it->second) + ".temp"), outFile);
+						m->openOutputFileAppend((distFile + "." + toString(it->second) + ".temp"), outFile);
 						outFile << outputs[it->second] << seqA << '\t' << seqB << '\t' << dist << endl;
 						outFile.close();
 						outputs[it->second] = "";
@@ -307,11 +328,11 @@ int SplitMatrix::splitDistanceFileByTax(map<string, int>& seqGroup, int numGroup
 		dFile.close();
 	
 		for (int i = 0; i < numGroups; i++) { //remove old temp files, just in case
-			remove((namefile + "." + toString(i) + ".temp").c_str());
+			m->mothurRemove((namefile + "." + toString(i) + ".temp"));
 			
 			//write out any remaining buffers
 			if (numOutputs[i] > 0) {
-				openOutputFileAppend((distFile + "." + toString(i) + ".temp"), outFile);
+				m->openOutputFileAppend((distFile + "." + toString(i) + ".temp"), outFile);
 				outFile << outputs[i];
 				outFile.close();
 				outputs[i] = "";
@@ -321,23 +342,23 @@ int SplitMatrix::splitDistanceFileByTax(map<string, int>& seqGroup, int numGroup
 		}
 		
 		ifstream bigNameFile;
-		openInputFile(namefile, bigNameFile);
+		m->openInputFile(namefile, bigNameFile);
 		
 		singleton = namefile + ".extra.temp";
 		ofstream remainingNames;
-		openOutputFile(singleton, remainingNames);
+		m->openOutputFile(singleton, remainingNames);
 		
 		bool wroteExtra = false;
 						
 		string name, nameList;
 		while(!bigNameFile.eof()){
-			bigNameFile >> name >> nameList;  gobble(bigNameFile);
+			bigNameFile >> name >> nameList;  m->gobble(bigNameFile);
 			
 			//did this sequence get assigned a group
 			it = seqGroup.find(name);
 			
 			if (it != seqGroup.end()) {  
-				openOutputFileAppend((namefile + "." + toString(it->second) + ".temp"), outFile);
+				m->openOutputFileAppend((namefile + "." + toString(it->second) + ".temp"), outFile);
 				outFile << name << '\t' << nameList << endl;
 				outFile.close();
 			}else{
@@ -358,29 +379,29 @@ int SplitMatrix::splitDistanceFileByTax(map<string, int>& seqGroup, int numGroup
 				dists.push_back(temp);
 			}else{
 				ifstream in;
-				openInputFile(tempNameFile, in);
+				m->openInputFile(tempNameFile, in);
 				
 				while(!in.eof()) { 
-					in >> name >> nameList;  gobble(in);
+					in >> name >> nameList;  m->gobble(in);
 					wroteExtra = true;
 					remainingNames << name << '\t' << nameList << endl;
 				}
 				in.close();
-				remove(tempNameFile.c_str());
+				m->mothurRemove(tempNameFile);
 			}
 		}
 		
 		remainingNames.close();
 		
 		if (!wroteExtra) { 
-			remove(singleton.c_str());
+			m->mothurRemove(singleton);
 			singleton = "none";
 		}
 
 		if (m->control_pressed)	 {  
 			for (int i = 0; i < dists.size(); i++) { 
-				remove((dists[i].begin()->first).c_str());
-				remove((dists[i].begin()->second).c_str());
+				m->mothurRemove((dists[i].begin()->first));
+				m->mothurRemove((dists[i].begin()->second));
 			}
 			dists.clear();
 		}
@@ -407,7 +428,7 @@ int SplitMatrix::splitDistanceLarge(){
 
 		ofstream outFile;
 		ifstream dFile;
-		openInputFile(distFile, dFile);
+		m->openInputFile(distFile, dFile);
 	
 		while(dFile){
 			string seqA, seqB;
@@ -415,7 +436,7 @@ int SplitMatrix::splitDistanceLarge(){
 
 			dFile >> seqA >> seqB >> dist;
 			
-			if (m->control_pressed) {   dFile.close();  for(int i=0;i<numGroups;i++){	if(groups[i].size() > 0){  remove((distFile + "." + toString(i) + ".temp").c_str()); }  } return 0; }
+			if (m->control_pressed) {   dFile.close();  for(int i=0;i<numGroups;i++){	if(groups[i].size() > 0){  m->mothurRemove((distFile + "." + toString(i) + ".temp")); }  } return 0; }
 					
 			if(dist < cutoff){
 				//cout << "in cutoff: " << dist << endl;
@@ -543,7 +564,7 @@ int SplitMatrix::splitDistanceLarge(){
 								delete memblock;
 								
 								fileB.close();
-								remove(fileName2.c_str());
+								m->mothurRemove(fileName2);
 								
 								//write out the merged memory
 								if (numOutputs[groupID] > 60) {
@@ -603,7 +624,7 @@ int SplitMatrix::splitDistanceLarge(){
 								delete memblock;
 								
 								fileB.close();
-								remove(fileName2.c_str());
+								m->mothurRemove(fileName2);
 								
 								//write out the merged memory
 								if (numOutputs[groupID] > 60) {
@@ -621,7 +642,7 @@ int SplitMatrix::splitDistanceLarge(){
 					}
 				}
 			}
-			gobble(dFile);
+			m->gobble(dFile);
 		}
 		dFile.close();
 		
@@ -659,7 +680,7 @@ int SplitMatrix::splitNames(vector<set<string> >& groups){
 		while(bigNameFile){
 			bigNameFile >> name >> nameList;
 			nameMap[name] = nameList;
-			gobble(bigNameFile);
+			m->gobble(bigNameFile);
 		}
 		bigNameFile.close();
 			
@@ -706,8 +727,8 @@ int SplitMatrix::splitNames(vector<set<string> >& groups){
 		
 		if (m->control_pressed)	 {  
 			for (int i = 0; i < dists.size(); i++) { 
-				remove((dists[i].begin()->first).c_str());
-				remove((dists[i].begin()->second).c_str());
+				m->mothurRemove((dists[i].begin()->first));
+				m->mothurRemove((dists[i].begin()->second));
 			}
 			dists.clear();
 		}
@@ -728,7 +749,7 @@ int SplitMatrix::splitDistanceRAM(){
 		int numGroups = 0;
 
 		ifstream dFile;
-		openInputFile(distFile, dFile);
+		m->openInputFile(distFile, dFile);
 
 		while(dFile){
 			string seqA, seqB;
@@ -736,7 +757,7 @@ int SplitMatrix::splitDistanceRAM(){
 
 			dFile >> seqA >> seqB >> dist;
 			
-			if (m->control_pressed) {   dFile.close();  for(int i=0;i<numGroups;i++){	if(groups[i].size() > 0){  remove((distFile + "." + toString(i) + ".temp").c_str()); }  } return 0; }
+			if (m->control_pressed) {   dFile.close();  for(int i=0;i<numGroups;i++){	if(groups[i].size() > 0){  m->mothurRemove((distFile + "." + toString(i) + ".temp")); }  } return 0; }
 					
 			if(dist < cutoff){
 				//cout << "in cutoff: " << dist << endl;
@@ -811,7 +832,7 @@ int SplitMatrix::splitDistanceRAM(){
 					}
 				}
 			}
-			gobble(dFile);
+			m->gobble(dFile);
 		}
 		dFile.close();