X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=splitmatrix.cpp;h=384b09af1bb94be09c5607c8b863b28b16215731;hb=2ecee16fec29d4c525f740ec19b27962ca09c050;hp=dcc5b50ef8b396d9186c755c3ba949f9c789cf64;hpb=8bc3e5b38c2317a1715f53be22fa96455868c281;p=mothur.git

diff --git a/splitmatrix.cpp b/splitmatrix.cpp
index dcc5b50..384b09a 100644
--- a/splitmatrix.cpp
+++ b/splitmatrix.cpp
@@ -10,6 +10,7 @@
 #include "splitmatrix.h"
 #include "phylotree.h"
 #include "distancecommand.h"
+#include "seqsummarycommand.h"
 
 /***********************************************************************/
 
@@ -24,7 +25,7 @@ SplitMatrix::SplitMatrix(string distfile, string name, string tax, float c, stri
 }
 /***********************************************************************/
 
-SplitMatrix::SplitMatrix(string ffile, string name, string tax, float c, float cu, string t, int p, string output){
+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;
@@ -33,6 +34,7 @@ SplitMatrix::SplitMatrix(string ffile, string name, string tax, float c, float c
 	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;
 }
 
@@ -65,6 +67,8 @@ int SplitMatrix::splitDistance(){
         
 		if (large)	{ splitDistanceLarge(); }
 		else		{ splitDistanceRAM();	}
+		
+		return 0;
 			
 	}
 	catch(exception& e) {
@@ -77,7 +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;
@@ -87,16 +91,13 @@ int SplitMatrix::splitClassify(){
 		//build tree from users taxonomy file
 		PhyloTree* phylo = new PhyloTree();
 		
-		ifstream in;
-		m->openInputFile(taxFile, in);
-			
-		//read in users taxonomy file and add sequences to tree
-		string seqname, tax;
-		while(!in.eof()){
-			in >> seqname >> tax; m->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);
 
@@ -143,7 +144,7 @@ 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;
@@ -182,18 +183,22 @@ 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) + ", cutoff=" + toString(distCutoff);
+			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;
 		m->openOutputFile(singleton, remainingNames);
@@ -224,7 +229,9 @@ int SplitMatrix::createDistanceFilesFromTax(map<string, int>& seqGroup, int numG
 		for(int i=0;i<numGroups;i++){
 			string tempNameFile = namefile + "." + toString(i) + ".temp";
 			if (outputDir == "") { outputDir = m->hasPath(fastafile); }
-			string tempDistFile = outputDir + m->getRootName(m->getSimpleName((fastafile + "." + toString(i) + ".temp"))) + "dist";
+			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;
@@ -245,7 +252,7 @@ int SplitMatrix::createDistanceFilesFromTax(map<string, int>& seqGroup, int numG
 						remainingNames << name << '\t' << nameList << endl;
 					}
 					in.close();
-					remove(tempNameFile.c_str());
+					m->mothurRemove(tempNameFile);
 				}
 			}
 			fileHandle.close();
@@ -253,11 +260,11 @@ int SplitMatrix::createDistanceFilesFromTax(map<string, int>& seqGroup, int numG
 		
 		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()); } dists.clear(); }
+		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;
 	}
@@ -277,7 +284,7 @@ int SplitMatrix::splitDistanceFileByTax(map<string, int>& seqGroup, int numGroup
 		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
@@ -294,7 +301,7 @@ 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;  m->gobble(dFile);
 			
@@ -321,7 +328,7 @@ 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) {
@@ -380,21 +387,21 @@ int SplitMatrix::splitDistanceFileByTax(map<string, int>& seqGroup, int numGroup
 					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();
 		}
@@ -429,7 +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;
@@ -557,7 +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) {
@@ -617,7 +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) {
@@ -720,8 +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();
 		}
@@ -750,7 +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;