added sets to amova and homova commands. added oligos to make.contigs. added metadat...

[mothur.git] / shhhercommand.cpp

diff --git a/shhhercommand.cpp b/shhhercommand.cpp
index 97629b600031c06c063f405175dd9fa369ae7c93..19ffc899d60ac778ac4eedbeb2034a62b5e04bd7 100644 (file)
--- a/shhhercommand.cpp
+++ b/shhhercommand.cpp
@@ -776,8 +776,8 @@ int ShhherCommand::execute(){
 
                
                if(compositeFASTAFileName != ""){
-                       outputNames.push_back(compositeFASTAFileName);
-                       outputNames.push_back(compositeNamesFileName);
+                       outputNames.push_back(compositeFASTAFileName); outputTypes["fasta"].push_back(compositeFASTAFileName);
+                       outputNames.push_back(compositeNamesFileName); outputTypes["name"].push_back(compositeNamesFileName); 
                }
 
                m->mothurOutEndLine();
@@ -802,7 +802,7 @@ string ShhherCommand::createNamesFile(){
                        duplicateNames[mapSeqToUnique[i]] += seqNameVector[i] + ',';
                }
                
-               string nameFileName = outputDir + m->getRootName(m->getSimpleName(flowFileName) + getOutputFileNameTag("name");
+               string nameFileName = outputDir + m->getRootName(m->getSimpleName(flowFileName)) + getOutputFileNameTag("name");
                
                ofstream nameFile;
                m->openOutputFile(nameFileName, nameFile);
@@ -1039,7 +1039,12 @@ void ShhherCommand::getFlowData(){
         
         float intensity;
         
-        flowFile >> numFlowCells;
+        string numFlowTest;
+        flowFile >> numFlowTest;
+        
+        if (!m->isContainingOnlyDigits(numFlowTest)) { m->mothurOut("[ERROR]: expected a number and got " + numFlowTest + ", quitting. Did you use the flow parameter instead of the file parameter?"); m->mothurOutEndLine(); exit(1); }
+        else { convert(numFlowTest, numFlowCells); }
+        
         int index = 0;//pcluster
         while(!flowFile.eof()){
             
@@ -1376,17 +1381,17 @@ string ShhherCommand::cluster(string distFileName, string namesFileName){
     try {
         
         ReadMatrix* read = new ReadColumnMatrix(distFileName);         
-        read->setCutoff(cutoff);
-        
-        NameAssignment* clusterNameMap = new NameAssignment(namesFileName);
-        clusterNameMap->readMap();
-        read->read(clusterNameMap);
-        
-        ListVector* list = read->getListVector();
-        SparseMatrix* matrix = read->getMatrix();
+               read->setCutoff(cutoff);
+               
+               NameAssignment* clusterNameMap = new NameAssignment(namesFileName);
+               clusterNameMap->readMap();
+               read->read(clusterNameMap);
         
-        delete read; 
-        delete clusterNameMap; 
+               ListVector* list = read->getListVector();
+               SparseDistanceMatrix* matrix = read->getDMatrix();
+               
+               delete read; 
+               delete clusterNameMap; 
         
         RAbundVector* rabund = new RAbundVector(list->getRAbundVector());
         
@@ -1738,7 +1743,7 @@ void ShhherCommand::writeQualities(vector<int> otuCounts){
             }
         }
         qualityFile.close();
-        outputNames.push_back(qualityFileName);
+        outputNames.push_back(qualityFileName); outputTypes["qfile"].push_back(qualityFileName);
         
     }
     catch(exception& e) {
@@ -1783,7 +1788,7 @@ void ShhherCommand::writeSequences(vector<int> otuCounts){
         }
         fastaFile.close();
         
-        outputNames.push_back(fastaFileName);
+        outputNames.push_back(fastaFileName); outputTypes["fasta"].push_back(fastaFileName);
         
         if(compositeFASTAFileName != ""){
             m->appendFiles(fastaFileName, compositeFASTAFileName);
@@ -1820,7 +1825,7 @@ void ShhherCommand::writeNames(vector<int> otuCounts){
             }
         }
         nameFile.close();
-        outputNames.push_back(nameFileName);
+        outputNames.push_back(nameFileName); outputTypes["name"].push_back(nameFileName);
         
         
         if(compositeNamesFileName != ""){
@@ -1839,17 +1844,20 @@ void ShhherCommand::writeGroups(){
     try {
         string thisOutputDir = outputDir;
         if (outputDir == "") {  thisOutputDir += m->hasPath(flowFileName);  }
-        string fileRoot = thisOutputDir + m->getRootName(m->getSimpleName(flowFileName));
-        string groupFileName = fileRoot + getOutputFileNameTag("group");
+        string fileRoot = m->getRootName(m->getSimpleName(flowFileName));
+        int pos = fileRoot.find_first_of('.');
+        string fileGroup = fileRoot;
+        if (pos != string::npos) {  fileGroup = fileRoot.substr(pos+1, (fileRoot.length()-1-(pos+1)));  }
+        string groupFileName = thisOutputDir + fileRoot + getOutputFileNameTag("group");
         ofstream groupFile;
         m->openOutputFile(groupFileName, groupFile);
         
         for(int i=0;i<numSeqs;i++){
             if (m->control_pressed) { break; }
-            groupFile << seqNameVector[i] << '\t' << fileRoot << endl;
+            groupFile << seqNameVector[i] << '\t' << fileGroup << endl;
         }
         groupFile.close();
-        outputNames.push_back(groupFileName);
+        outputNames.push_back(groupFileName); outputTypes["group"].push_back(groupFileName);
         
     }
     catch(exception& e) {
@@ -1909,7 +1917,7 @@ void ShhherCommand::writeClusters(vector<int> otuCounts){
             }
         }
         otuCountsFile.close();
-        outputNames.push_back(otuCountsFileName);
+        outputNames.push_back(otuCountsFileName); outputTypes["counts"].push_back(otuCountsFileName);
         
     }
     catch(exception& e) {
@@ -1923,7 +1931,7 @@ void ShhherCommand::writeClusters(vector<int> otuCounts){
 
 int ShhherCommand::execute(){
        try {
-               if (abort == true) { return 0; }
+               if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
                
                getSingleLookUp();      if (m->control_pressed) { return 0; }
                getJointLookUp();       if (m->control_pressed) { return 0; }
@@ -1940,8 +1948,8 @@ int ShhherCommand::execute(){
 #endif
         
                if(compositeFASTAFileName != ""){
-                       outputNames.push_back(compositeFASTAFileName);
-                       outputNames.push_back(compositeNamesFileName);
+                       outputNames.push_back(compositeFASTAFileName); outputTypes["fasta"].push_back(compositeFASTAFileName);
+                       outputNames.push_back(compositeNamesFileName); outputTypes["name"].push_back(compositeNamesFileName);
                }
 
                m->mothurOutEndLine();
@@ -2026,7 +2034,7 @@ int ShhherCommand::createProcesses(vector<string> filenames){
                //Windows version shared memory, so be careful when passing variables through the shhhFlowsData struct. 
                //Above fork() will clone, so memory is separate, but that's not the case with windows, 
                //////////////////////////////////////////////////////////////////////////////////////////////////////
-               
+               /*
                vector<shhhFlowsData*> pDataArray; 
                DWORD   dwThreadIdArray[processors-1];
                HANDLE  hThreadArray[processors-1]; 
@@ -2057,7 +2065,7 @@ int ShhherCommand::createProcesses(vector<string> filenames){
                        CloseHandle(hThreadArray[i]);
                        delete pDataArray[i];
                }
-               
+               */
         #endif
         
         for (int i=0;i<processIDS.size();i++) { 
@@ -2306,7 +2314,7 @@ int ShhherCommand::driver(vector<string> filenames, string thisCompositeFASTAFil
                 if (m->control_pressed) { break; }
                 
                 vector<int> otuCounts(numOTUs, 0);
-                for(int i=0;i<numSeqs;i++)     {       otuCounts[otuData[i]]++;        }
+                for(int j=0;j<numSeqs;j++)     {       otuCounts[otuData[j]]++;        }
                 
                 calcCentroidsDriver(numOTUs, cumNumSeqs, nSeqsPerOTU, seqIndex, change, centroids, singleTau, mapSeqToUnique, uniqueFlowgrams, flowDataIntI, lengths, numFlowCells, seqNumber);        
                 
@@ -2319,15 +2327,18 @@ int ShhherCommand::driver(vector<string> filenames, string thisCompositeFASTAFil
                 string fastaFileName = thisOutputDir + m->getRootName(m->getSimpleName(flowFileName)) + getOutputFileNameTag("fasta");
                 string nameFileName = thisOutputDir + m->getRootName(m->getSimpleName(flowFileName)) + getOutputFileNameTag("name");
                 string otuCountsFileName = thisOutputDir + m->getRootName(m->getSimpleName(flowFileName)) + getOutputFileNameTag("counts");
-                string fileRoot = thisOutputDir + m->getRootName(m->getSimpleName(flowFileName));
-                string groupFileName = fileRoot + getOutputFileNameTag("group");
+                string fileRoot = m->getRootName(m->getSimpleName(flowFileName));
+                int pos = fileRoot.find_first_of('.');
+                string fileGroup = fileRoot;
+                if (pos != string::npos) {  fileGroup = fileRoot.substr(pos+1, (fileRoot.length()-1-(pos+1)));  }
+                string groupFileName = thisOutputDir + fileRoot + getOutputFileNameTag("group");
 
                 
                 writeQualities(numOTUs, numFlowCells, qualityFileName, otuCounts, nSeqsPerOTU, seqNumber, singleTau, flowDataIntI, uniqueFlowgrams, cumNumSeqs, mapUniqueToSeq, seqNameVector, centroids, aaI); if (m->control_pressed) { break; }
                 writeSequences(thisCompositeFASTAFileName, numOTUs, numFlowCells, fastaFileName, otuCounts, uniqueFlowgrams, seqNameVector, aaI, centroids);if (m->control_pressed) { break; }
                 writeNames(thisCompositeNamesFileName, numOTUs, nameFileName, otuCounts, seqNameVector, aaI, nSeqsPerOTU);                             if (m->control_pressed) { break; }
                 writeClusters(otuCountsFileName, numOTUs, numFlowCells,otuCounts, centroids, uniqueFlowgrams, seqNameVector, aaI, nSeqsPerOTU, lengths, flowDataIntI);                 if (m->control_pressed) { break; }
-                writeGroups(groupFileName, fileRoot, numSeqs, seqNameVector);                                          if (m->control_pressed) { break; }
+                writeGroups(groupFileName, fileGroup, numSeqs, seqNameVector);                                         if (m->control_pressed) { break; }
                 
                 if (large) {
                     if (g > 0) {
@@ -2376,7 +2387,12 @@ int ShhherCommand::getFlowData(string filename, vector<string>& thisSeqNameVecto
                thisFlowDataIntI.clear();
                thisNameMap.clear();
                
-               flowFile >> numFlowCells;
+               string numFlowTest;
+        flowFile >> numFlowTest;
+        
+        if (!m->isContainingOnlyDigits(numFlowTest)) { m->mothurOut("[ERROR]: expected a number and got " + numFlowTest + ", quitting. Did you use the flow parameter instead of the file parameter?"); m->mothurOutEndLine(); exit(1); }
+        else { convert(numFlowTest, numFlowCells); }
+        
         if (m->debug) { m->mothurOut("[DEBUG]: numFlowCells = " + toString(numFlowCells) + ".\n"); }
                int index = 0;//pcluster
                while(!flowFile.eof()){
@@ -2626,7 +2642,7 @@ int ShhherCommand::cluster(string filename, string distFileName, string namesFil
                read->read(clusterNameMap);
         
                ListVector* list = read->getListVector();
-               SparseMatrix* matrix = read->getMatrix();
+               SparseDistanceMatrix* matrix = read->getDMatrix();
                
                delete read; 
                delete clusterNameMap; 
@@ -3250,7 +3266,7 @@ void ShhherCommand::writeQualities(int numOTUs, int numFlowCells, string quality
                        }
                }
                qualityFile.close();
-               outputNames.push_back(qualityFileName);
+               outputNames.push_back(qualityFileName); outputTypes["qfile"].push_back(qualityFileName);
         
        }
        catch(exception& e) {
@@ -3294,7 +3310,7 @@ void ShhherCommand::writeSequences(string thisCompositeFASTAFileName, int numOTU
                }
                fastaFile.close();
         
-               outputNames.push_back(fastaFileName);
+               outputNames.push_back(fastaFileName); outputTypes["fasta"].push_back(fastaFileName);
         
                if(thisCompositeFASTAFileName != ""){
                        m->appendFiles(fastaFileName, thisCompositeFASTAFileName);
@@ -3329,7 +3345,7 @@ void ShhherCommand::writeNames(string thisCompositeNamesFileName, int numOTUs, s
                        }
                }
                nameFile.close();
-               outputNames.push_back(nameFileName);
+               outputNames.push_back(nameFileName); outputTypes["name"].push_back(nameFileName);
                
                
                if(thisCompositeNamesFileName != ""){
@@ -3354,7 +3370,7 @@ void ShhherCommand::writeGroups(string groupFileName, string fileRoot, int numSe
                        groupFile << seqNameVector[i] << '\t' << fileRoot << endl;
                }
                groupFile.close();
-               outputNames.push_back(groupFileName);
+               outputNames.push_back(groupFileName); outputTypes["group"].push_back(groupFileName);
         
        }
        catch(exception& e) {
@@ -3413,7 +3429,7 @@ void ShhherCommand::writeClusters(string otuCountsFileName, int numOTUs, int num
                        }
                }
                otuCountsFile.close();
-               outputNames.push_back(otuCountsFileName);
+               outputNames.push_back(otuCountsFileName); outputTypes["counts"].push_back(otuCountsFileName);
         
        }
        catch(exception& e) {