added forward and reverse barcodes to trim.seqs to process illumina seqs

[mothur.git] / classifyseqscommand.cpp

diff --git a/classifyseqscommand.cpp b/classifyseqscommand.cpp
index 328cd58f1f32bc265de857de4ede5aa557c2cd94..b6dc24fe5751867c6b79f71fb50e3b7b74ad75a6 100644 (file)
--- a/classifyseqscommand.cpp
+++ b/classifyseqscommand.cpp
@@ -457,13 +457,15 @@ ClassifySeqsCommand::ClassifySeqsCommand(string option)  {
                                search = "kmer";
                        }
                        
-                       if (namefileNames.size() == 0){
-                               vector<string> files; files.push_back(fastaFileNames[fastaFileNames.size()-1]); 
-                               parser.getNameFile(files);
-                       }
-                       
-               }
-               
+            if (!abort) {
+                if (namefileNames.size() == 0){
+                    if (fastaFileNames.size() != 0) {
+                        vector<string> files; files.push_back(fastaFileNames[fastaFileNames.size()-1]); 
+                        parser.getNameFile(files);
+                    }
+                }
+            }
+        }
        }
        catch(exception& e) {
                m->errorOut(e, "ClassifySeqsCommand", "ClassifySeqsCommand");
@@ -482,7 +484,7 @@ ClassifySeqsCommand::~ClassifySeqsCommand(){
 int ClassifySeqsCommand::execute(){
        try {
                if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
-               
+        
                if(method == "bayesian"){       classify = new Bayesian(taxonomyFileName, templateFileName, search, kmerSize, cutoff, iters, rand(), flip);             }
                else if(method == "knn"){       classify = new Knn(taxonomyFileName, templateFileName, search, kmerSize, gapOpen, gapExtend, match, misMatch, numWanted, rand());                               }
                else {
@@ -500,11 +502,17 @@ int ClassifySeqsCommand::execute(){
                        string baseTName = taxonomyFileName;
                        if (taxonomyFileName == "saved") {baseTName = rdb->getSavedTaxonomy();  }
                        
-                       string RippedTaxName = m->getRootName(m->getSimpleName(baseTName));
-                       RippedTaxName = m->getExtension(RippedTaxName.substr(0, RippedTaxName.length()-1));
-                       if (RippedTaxName[0] == '.') { RippedTaxName = RippedTaxName.substr(1, RippedTaxName.length()); }
-                       RippedTaxName +=  "."; 
-               
+            //set rippedTaxName to 
+                       string RippedTaxName = "";
+            bool foundDot = false;
+            for (int i = baseTName.length()-1; i >= 0; i--) {
+                cout << baseTName[i] << endl;
+                if (foundDot && (baseTName[i] != '.')) {  RippedTaxName = baseTName[i] + RippedTaxName; }
+                else if (foundDot && (baseTName[i] == '.')) {  break; }
+                else if (!foundDot && (baseTName[i] == '.')) {  foundDot = true; }
+            }
+            if (RippedTaxName != "") { RippedTaxName +=  "."; }   
+          
                        if (outputDir == "") { outputDir += m->hasPath(fastaFileNames[s]); }
                        string newTaxonomyFile = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + RippedTaxName + "taxonomy";
                        string newaccnosFile = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + RippedTaxName + "flip.accnos";
@@ -615,7 +623,7 @@ int ClassifySeqsCommand::execute(){
 #else
                
                        vector<unsigned long long> positions; 
-#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
                        positions = m->divideFile(fastaFileNames[s], processors);
                        for (int i = 0; i < (positions.size()-1); i++) {        lines.push_back(new linePair(positions[i], positions[(i+1)]));  }
 #else
@@ -623,6 +631,7 @@ int ClassifySeqsCommand::execute(){
                                lines.push_back(new linePair(0, 1000));
                        }else {
                                positions = m->setFilePosFasta(fastaFileNames[s], numFastaSeqs); 
+                if (positions.size() < processors) { processors = positions.size(); }
                                
                                //figure out how many sequences you have to process
                                int numSeqsPerProcessor = numFastaSeqs / processors;
@@ -821,7 +830,7 @@ int ClassifySeqsCommand::createProcesses(string taxFileName, string tempTaxFile,
                int num = 0;
                processIDS.clear();
                
-#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
                int process = 1;
                
                //loop through and create all the processes you want
@@ -881,7 +890,7 @@ int ClassifySeqsCommand::createProcesses(string taxFileName, string tempTaxFile,
                        string extension = "";
                        if (i != 0) { extension = toString(i) + ".temp"; processIDS.push_back(i); }
                        
-                       classifyData* tempclass = new classifyData((accnos + extension), probs, method, templateFileName, taxonomyFileName, (taxFileName + extension), (tempTaxFile + extension), filename, search, kmerSize, iters, numWanted, m, lines[i]->start, lines[i]->end, match, misMatch, gapOpen, gapExtend, cutoff, i, flipThreshold);
+                       classifyData* tempclass = new classifyData((accnos + extension), probs, method, templateFileName, taxonomyFileName, (taxFileName + extension), (tempTaxFile + extension), filename, search, kmerSize, iters, numWanted, m, lines[i]->start, lines[i]->end, match, misMatch, gapOpen, gapExtend, cutoff, i, flip);
                        pDataArray.push_back(tempclass);
                        
                        //MySeqSumThreadFunction is in header. It must be global or static to work with the threads.
@@ -905,16 +914,35 @@ int ClassifySeqsCommand::createProcesses(string taxFileName, string tempTaxFile,
                }
                
        #endif  
-               
+        vector<string> nonBlankAccnosFiles;
+               if (!(m->isBlank(accnos))) { nonBlankAccnosFiles.push_back(accnos); }
+               else { m->mothurRemove(accnos); } //remove so other files can be renamed to it
+        
                for(int i=0;i<processIDS.size();i++){
                        appendTaxFiles((taxFileName + toString(processIDS[i]) + ".temp"), taxFileName);
                        appendTaxFiles((tempTaxFile + toString(processIDS[i]) + ".temp"), tempTaxFile);
-                       appendTaxFiles((accnos + toString(processIDS[i]) + ".temp"), accnos);
+            if (!(m->isBlank(accnos + toString(processIDS[i]) + ".temp"))) {
+                               nonBlankAccnosFiles.push_back(accnos + toString(processIDS[i]) + ".temp");
+                       }else { m->mothurRemove((accnos + toString(processIDS[i]) + ".temp"));  }
+
                        m->mothurRemove((m->getFullPathName(taxFileName) + toString(processIDS[i]) + ".temp"));
                        m->mothurRemove((m->getFullPathName(tempTaxFile) + toString(processIDS[i]) + ".temp"));
-                       m->mothurRemove((m->getFullPathName(accnos) + toString(processIDS[i]) + ".temp"));
                }
                
+        //append accnos files
+               if (nonBlankAccnosFiles.size() != 0) { 
+                       rename(nonBlankAccnosFiles[0].c_str(), accnos.c_str());
+                       
+                       for (int h=1; h < nonBlankAccnosFiles.size(); h++) {
+                               appendTaxFiles(nonBlankAccnosFiles[h], accnos);
+                               m->mothurRemove(nonBlankAccnosFiles[h]);
+                       }
+               }else { //recreate the accnosfile if needed
+                       ofstream out;
+                       m->openOutputFile(accnos, out);
+                       out.close();
+               }
+
                return num;
                
        }
@@ -1002,7 +1030,7 @@ int ClassifySeqsCommand::driver(linePair* filePos, string taxFName, string tempT
                        }
                        delete candidateSeq;
                        
-                       #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+                       #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
                                unsigned long long pos = inFASTA.tellg();
                                if ((pos == -1) || (pos >= filePos->end)) { break; }
                        #else