X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=deconvolutecommand.cpp;h=3d0c0d51acaac8844b71a3efa40c7925674ad4c5;hb=36a6b02cf7f09d2bc34376b588944a9ca73429c5;hp=33e51ffdff96a570d2a91d85590b3a2111ddcf63;hpb=ca9ac1d80c62f57270b0dcd49410ebe08a8aecd6;p=mothur.git

diff --git a/deconvolutecommand.cpp b/deconvolutecommand.cpp
index 33e51ff..3d0c0d5 100644
--- a/deconvolutecommand.cpp
+++ b/deconvolutecommand.cpp
@@ -8,6 +8,7 @@
  */
 
 #include "deconvolutecommand.h"
+#include "sequence.hpp"
 
 //**********************************************************************************************************************
 vector<string> DeconvoluteCommand::setParameters(){	
@@ -64,6 +65,7 @@ DeconvoluteCommand::DeconvoluteCommand(string option)  {
 		
 		//allow user to run help
 		if(option == "help") { help(); abort = true; calledHelp = true; }
+		else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 		
 		else {
 			vector<string> myArray = setParameters();
@@ -114,7 +116,7 @@ DeconvoluteCommand::DeconvoluteCommand(string option)  {
 				inFastaName = m->getFastaFile(); 
 				if (inFastaName != "") { m->mothurOut("Using " + inFastaName + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 				else { 	m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
-			}	
+			}else { m->setFastaFile(inFastaName); }	
 			
 			//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"){	
@@ -123,8 +125,15 @@ DeconvoluteCommand::DeconvoluteCommand(string option)  {
 			}
 			
 			oldNameMapFName = validParameter.validFile(parameters, "name", true);
-			if (oldNameMapFName == "not open") { abort = true; }
+			if (oldNameMapFName == "not open") { oldNameMapFName = ""; abort = true; }
 			else if (oldNameMapFName == "not found"){	oldNameMapFName = "";	}
+			else { m->setNameFile(oldNameMapFName); }
+			
+			if (oldNameMapFName == "") {
+				vector<string> files; files.push_back(inFastaName);
+				parser.getNameFile(files);
+			}
+			
 		}
 
 	}
@@ -143,17 +152,107 @@ int DeconvoluteCommand::execute() {
 		string outNameFile = outputDir + m->getRootName(m->getSimpleName(inFastaName)) + "names";
 		string outFastaFile = outputDir + m->getRootName(m->getSimpleName(inFastaName)) + "unique" + m->getExtension(inFastaName);
 		
-		FastaMap fastamap;
-	
-		if(oldNameMapFName == "")	{	fastamap.readFastaFile(inFastaName);					}
-		else						{	fastamap.readFastaFile(inFastaName, oldNameMapFName);	}
+		map<string, string> nameMap;
+		map<string, string>::iterator itNames;
+		if (oldNameMapFName != "")  {  
+            m->readNames(oldNameMapFName, nameMap); 
+            if (oldNameMapFName == outNameFile){ outNameFile = outputDir + m->getRootName(m->getSimpleName(inFastaName)) + "unique.names";   }
+        }
 		
 		if (m->control_pressed) { return 0; }
 		
-		fastamap.printCondensedFasta(outFastaFile);
-		fastamap.printNamesFile(outNameFile);
+		ifstream in; 
+		m->openInputFile(inFastaName, in);
+		
+		ofstream outFasta;
+		m->openOutputFile(outFastaFile, outFasta);
+		
+		map<string, string> sequenceStrings; //sequenceString -> list of names.  "atgc...." -> seq1,seq2,seq3.
+		map<string, string>::iterator itStrings;
+		set<string> nameInFastaFile; //for sanity checking
+		set<string>::iterator itname;
+		vector<string> nameFileOrder;
+		int count = 0;
+		while (!in.eof()) {
+			
+			if (m->control_pressed) { in.close(); outFasta.close(); m->mothurRemove(outFastaFile); return 0; }
+			
+			Sequence seq(in);
+			
+			if (seq.getName() != "") {
+				
+				//sanity checks
+				itname = nameInFastaFile.find(seq.getName());
+				if (itname == nameInFastaFile.end()) { nameInFastaFile.insert(seq.getName());  }
+				else { m->mothurOut("[ERROR]: You already have a sequence named " + seq.getName() + " in your fasta file, sequence names must be unique, please correct."); m->mothurOutEndLine(); }
+
+				itStrings = sequenceStrings.find(seq.getAligned());
+				
+				if (itStrings == sequenceStrings.end()) { //this is a new unique sequence
+					//output to unique fasta file
+					seq.printSequence(outFasta);
+					
+					if (oldNameMapFName != "") {
+						itNames = nameMap.find(seq.getName());
+						
+						if (itNames == nameMap.end()) { //namefile and fastafile do not match
+							m->mothurOut("[ERROR]: " + seq.getName() + " is in your fasta file, and not in your namefile, please correct."); m->mothurOutEndLine();
+						}else {
+							sequenceStrings[seq.getAligned()] = itNames->second;
+							nameFileOrder.push_back(seq.getAligned());
+						}
+					}else {	sequenceStrings[seq.getAligned()] = seq.getName();	nameFileOrder.push_back(seq.getAligned()); }
+				}else { //this is a dup
+					if (oldNameMapFName != "") {
+						itNames = nameMap.find(seq.getName());
+						
+						if (itNames == nameMap.end()) { //namefile and fastafile do not match
+							m->mothurOut("[ERROR]: " + seq.getName() + " is in your fasta file, and not in your namefile, please correct."); m->mothurOutEndLine();
+						}else {
+							sequenceStrings[seq.getAligned()] += "," + itNames->second;
+						}
+					}else {	sequenceStrings[seq.getAligned()] += "," + seq.getName();	}
+				}
+				
+				count++;
+			}
+			
+			m->gobble(in);
+			
+			if(count % 1000 == 0)	{ m->mothurOut(toString(count) + "\t" + toString(sequenceStrings.size())); m->mothurOutEndLine();	}
+		}
+		
+		if(count % 1000 != 0)	{ m->mothurOut(toString(count) + "\t" + toString(sequenceStrings.size())); m->mothurOutEndLine();	}
+		
+		in.close();
+		outFasta.close();
+		
+		if (m->control_pressed) { m->mothurRemove(outFastaFile); return 0; }
+		
+		//print new names file
+		ofstream outNames;
+		m->openOutputFile(outNameFile, outNames);
+		
+		for (int i = 0; i < nameFileOrder.size(); i++) {
+		//for (itStrings = sequenceStrings.begin(); itStrings != sequenceStrings.end(); itStrings++) {
+			if (m->control_pressed) { outputTypes.clear(); m->mothurRemove(outFastaFile); outNames.close(); m->mothurRemove(outNameFile); return 0; }
+			
+			itStrings = sequenceStrings.find(nameFileOrder[i]);
+			
+			if (itStrings != sequenceStrings.end()) {
+				//get rep name
+				int pos = (itStrings->second).find_first_of(',');
+			
+				if (pos == string::npos) { // only reps itself
+					outNames << itStrings->second << '\t' << itStrings->second << endl;
+				}else {
+					outNames << (itStrings->second).substr(0, pos) << '\t' << itStrings->second << endl;
+				}
+			}else{ m->mothurOut("[ERROR]: mismatch in namefile print."); m->mothurOutEndLine(); m->control_pressed = true; }
+		}
+		outNames.close();
 		
-		if (m->control_pressed) { outputTypes.clear(); remove(outFastaFile.c_str()); remove(outNameFile.c_str()); return 0; }
+		if (m->control_pressed) { outputTypes.clear(); m->mothurRemove(outFastaFile); m->mothurRemove(outNameFile); return 0; }
 		
 		m->mothurOutEndLine();
 		m->mothurOut("Output File Names: "); m->mothurOutEndLine();