From 814675616c10a40911886afb9dfe216f0245e3a2 Mon Sep 17 00:00:00 2001
From: westcott <westcott>
Date: Mon, 14 Jun 2010 14:30:17 +0000
Subject: [PATCH] added fasta method to cluster.split

---
 clustersplitcommand.cpp | 10 +++++-----
 splitmatrix.cpp         | 34 +++++++++++++++++++++++-----------
 splitmatrix.h           |  4 ++--
 3 files changed, 30 insertions(+), 18 deletions(-)

diff --git a/clustersplitcommand.cpp b/clustersplitcommand.cpp
index f6bf105..8d21d1a 100644
--- a/clustersplitcommand.cpp
+++ b/clustersplitcommand.cpp
@@ -113,7 +113,7 @@ ClusterSplitCommand::ClusterSplitCommand(string option)  {
 			fastafile = validParameter.validFile(parameters, "fasta", true);
 			if (fastafile == "not open") { abort = true; }	
 			else if (fastafile == "not found") { fastafile = ""; }
-			else { splitmethod = "fasta";  }
+			else { distfile = fastafile;  splitmethod = "fasta";  }
 			
 			taxFile = validParameter.validFile(parameters, "taxonomy", true);
 			if (taxFile == "not open") { abort = true; }	
@@ -271,10 +271,10 @@ int ClusterSplitCommand::execute(){
 		
 		//split matrix into non-overlapping groups
 		SplitMatrix* split;
-		if (splitmethod == "distance")			{	split = new SplitMatrix(distfile, namefile, taxFile, cutoff, splitmethod, large);					}
-		else if (splitmethod == "classify")		{	split = new SplitMatrix(distfile, namefile, taxFile, taxLevelCutoff, splitmethod, large);			}
-		else if (splitmethod == "fasta")		{	split = new SplitMatrix(fastafile, namefile, taxFile, taxLevelCutoff, splitmethod, processors);		}
-		else { m->mothurOut("Not a valid splitting method.  Valid splitting algorithms are distance, classify or fasta."); m->mothurOutEndLine(); return 0; }
+		if (splitmethod == "distance")			{	split = new SplitMatrix(distfile, namefile, taxFile, cutoff, splitmethod, large);							}
+		else if (splitmethod == "classify")		{	split = new SplitMatrix(distfile, namefile, taxFile, taxLevelCutoff, splitmethod, large);					}
+		else if (splitmethod == "fasta")		{	split = new SplitMatrix(fastafile, namefile, taxFile, taxLevelCutoff, splitmethod, processors, outputDir);	}
+		else { m->mothurOut("Not a valid splitting method.  Valid splitting algorithms are distance, classify or fasta."); m->mothurOutEndLine(); return 0;		}
 		
 		split->split();
 		
diff --git a/splitmatrix.cpp b/splitmatrix.cpp
index 9e53c51..6e40e97 100644
--- a/splitmatrix.cpp
+++ b/splitmatrix.cpp
@@ -24,7 +24,7 @@ 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, string t, int p, string output){
 	m = MothurOut::getInstance();
 	fastafile = ffile;
 	namefile = name;
@@ -32,6 +32,7 @@ SplitMatrix::SplitMatrix(string ffile, string name, string tax, float c, string
 	cutoff = c;
 	method = t;
 	processors = p;
+	outputDir = output;
 }
 
 /***********************************************************************/
@@ -163,7 +164,7 @@ int SplitMatrix::createDistanceFilesFromTax(map<string, int>& seqGroup, int numG
 					query.printSequence(outFile); 
 					outFile.close();
 					
-					copyGroups.erase(it);
+					copyGroups.erase(query.getName());
 				}
 			}
 		}
@@ -180,7 +181,7 @@ 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 = "fasta=" + (fastafile + "." + toString(i) + ".temp") + ", processors=" + toString(processors) + ", cutoff=" + toString(cutoff);
 			
 			Command* command = new DistanceCommand(options);
 			command->execute();
@@ -219,30 +220,41 @@ 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";
+			string tempDistFile = outputDir + getRootName(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
+				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;
+					openInputFile(tempNameFile, in);
+				
+					while(!in.eof()) { 
+						in >> name >> nameList;  gobble(in);
+						wroteExtra = true;
+						remainingNames << name << '\t' << nameList << endl;
+					}
+					in.close();
+					remove(tempNameFile.c_str());
 				}
 			}
 			fileHandle.close();
 		}
 		
+		remainingNames.close();
+		if (!wroteExtra) { 
+			remove(singleton.c_str());
+			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(); }
 		
 		return 0;
diff --git a/splitmatrix.h b/splitmatrix.h
index b98bdfb..acaa8c5 100644
--- a/splitmatrix.h
+++ b/splitmatrix.h
@@ -20,7 +20,7 @@ class SplitMatrix  {
 	public:
 
 		SplitMatrix(string, string, string, float, string, bool); //column formatted distance file, namesfile, cutoff, method, large
-		SplitMatrix(string, string, string, float, string, int); //fastafile, namefile, taxFile, cutoff, method, processors
+		SplitMatrix(string, string, string, float, string, int, string); //fastafile, namefile, taxFile, cutoff, method, processors, outputDir
 		
 		~SplitMatrix();
 		int split();
@@ -30,7 +30,7 @@ class SplitMatrix  {
 	private:
 		MothurOut* m;
 
-		string distFile, namefile, singleton, method, taxFile, fastafile;
+		string distFile, namefile, singleton, method, taxFile, fastafile, outputDir;
 		vector< map< string, string> > dists;
 		float cutoff;
 		bool large;
-- 
2.39.2