added sparseDistanceMatrix class. Modified cluster commands to use the new sparse...

[mothur.git] / chimeraperseuscommand.cpp

diff --git a/chimeraperseuscommand.cpp b/chimeraperseuscommand.cpp
index 8eaf536040a5703bbd7d9f4b93329f79cf6c6830..e3691e8b942c2f34da91cec14691e717d5c16c1a 100644 (file)
--- a/chimeraperseuscommand.cpp
+++ b/chimeraperseuscommand.cpp
@@ -58,6 +58,27 @@ string ChimeraPerseusCommand::getHelpString(){
        }
 }
 //**********************************************************************************************************************
+string ChimeraPerseusCommand::getOutputFileNameTag(string type, string inputName=""){  
+       try {
+        string outputFileName = "";
+               map<string, vector<string> >::iterator it;
+        
+        //is this a type this command creates
+        it = outputTypes.find(type);
+        if (it == outputTypes.end()) {  m->mothurOut("[ERROR]: this command doesn't create a " + type + " output file.\n"); }
+        else {
+            if (type == "chimera") {  outputFileName =  "perseus.chimeras"; }
+            else if (type == "accnos") {  outputFileName =  "perseus.accnos"; }
+            else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true;  }
+        }
+        return outputFileName;
+       }
+       catch(exception& e) {
+               m->errorOut(e, "ChimeraPerseusCommand", "getOutputFileNameTag");
+               exit(1);
+       }
+}
+//**********************************************************************************************************************
 ChimeraPerseusCommand::ChimeraPerseusCommand(){        
        try {
                abort = true; calledHelp = true;
@@ -376,8 +397,9 @@ int ChimeraPerseusCommand::execute(){
                        
                        int start = time(NULL); 
                        if (outputDir == "") { outputDir = m->hasPath(fastaFileNames[s]);  }//if user entered a file with a path then preserve it                               
-                       string outputFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + "perseus.chimera";
-                       string accnosFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s]))  + "perseus.accnos";
+                       string outputFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + getOutputFileNameTag("chimera");
+                       string accnosFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + getOutputFileNameTag("accnos");
+
                        //string newFasta = m->getRootName(fastaFileNames[s]) + "temp";
                        
                        //you provided a groupfile
@@ -466,14 +488,15 @@ string ChimeraPerseusCommand::getNamesFile(string& inputFile){
                string inputString = "fasta=" + inputFile;
                m->mothurOut("/******************************************/"); m->mothurOutEndLine(); 
                m->mothurOut("Running command: unique.seqs(" + inputString + ")"); m->mothurOutEndLine(); 
-               
+               m->mothurCalling = true;
+        
                Command* uniqueCommand = new DeconvoluteCommand(inputString);
                uniqueCommand->execute();
                
                map<string, vector<string> > filenames = uniqueCommand->getOutputFiles();
                
                delete uniqueCommand;
-               
+               m->mothurCalling = false;
                m->mothurOut("/******************************************/"); m->mothurOutEndLine(); 
                
                nameFile = filenames["name"][0];
@@ -533,6 +556,7 @@ vector<seqData> ChimeraPerseusCommand::loadSequences(SequenceParser& parser, str
                
                vector<seqData> sequences;
                bool error = false;
+        alignLength = 0;
                
                for (int i = 0; i < thisGroupsSeqs.size(); i++) {
                
@@ -543,6 +567,7 @@ vector<seqData> ChimeraPerseusCommand::loadSequences(SequenceParser& parser, str
                        else {
                                int num = m->getNumNames(it->second);
                                sequences.push_back(seqData(thisGroupsSeqs[i].getName(), thisGroupsSeqs[i].getUnaligned(), num));
+                if (thisGroupsSeqs[i].getUnaligned().length() > alignLength) { alignLength = thisGroupsSeqs[i].getUnaligned().length(); }
                        }
                }
                
@@ -570,7 +595,8 @@ vector<seqData> ChimeraPerseusCommand::readFiles(string inputFile, string name){
                bool error = false;
                ifstream in;
                m->openInputFile(inputFile, in);
-               
+               alignLength = 0;
+        
                while (!in.eof()) {
                        
                        if (m->control_pressed) { in.close(); return sequences; }
@@ -581,6 +607,7 @@ vector<seqData> ChimeraPerseusCommand::readFiles(string inputFile, string name){
                        if (it == nameMap.end()) { error = true; m->mothurOut("[ERROR]: " + temp.getName() + " is in your fasta file and not in your namefile, please correct."); m->mothurOutEndLine(); }
                        else {
                                sequences.push_back(seqData(temp.getName(), temp.getUnaligned(), it->second));
+                if (temp.getUnaligned().length() > alignLength) { alignLength = temp.getUnaligned().length(); }
                        }
                }
                in.close();
@@ -625,7 +652,7 @@ int ChimeraPerseusCommand::driver(string chimeraFileName, vector<seqData>& seque
                }
                
                int numSeqs = sequences.size();
-               int alignLength = sequences[0].sequence.size();
+               //int alignLength = sequences[0].sequence.size();
                
                ofstream chimeraFile;
                ofstream accnosFile;
@@ -641,7 +668,7 @@ int ChimeraPerseusCommand::driver(string chimeraFileName, vector<seqData>& seque
                
                for(int i=0;i<numSeqs;i++){     
                        if (m->control_pressed) { chimeraFile.close(); m->mothurRemove(chimeraFileName); accnosFile.close(); m->mothurRemove(accnosFileName); return 0; }
-                       
+    
                        vector<bool> restricted = chimeras;
                        
                        vector<vector<int> > leftDiffs(numSeqs);
@@ -662,7 +689,9 @@ int ChimeraPerseusCommand::driver(string chimeraFileName, vector<seqData>& seque
                        
                        string dummyA, dummyB;
                        
-                       if(comparisons >= 2){   
+            if (sequences[i].sequence.size() < 3) { 
+                chimeraFile << i << '\t' << sequences[i].seqName << "\t0\t0\tNull\t0\t0\t0\tNull\tNull\t0.0\t0.0\t0.0\t0\t0\t0\t0.0\t0.0\tgood" << endl;
+            }else if(comparisons >= 2){        
                                minMismatchToChimera = myPerseus.getChimera(sequences, leftDiffs, rightDiffs, leftParentBi, rightParentBi, breakPointBi, singleLeft, bestLeft, singleRight, bestRight, restricted);
                                if (m->control_pressed) { chimeraFile.close(); m->mothurRemove(chimeraFileName); accnosFile.close(); m->mothurRemove(accnosFileName); return 0; }
 
@@ -762,7 +791,7 @@ int ChimeraPerseusCommand::createProcessesGroups(SequenceParser& parser, string
                        lines.push_back(linePair(startIndex, endIndex));
                }
                
-#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)          
+#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)         
                
                //loop through and create all the processes you want
                while (process != processors) {