working on pam

[mothur.git] / classifyotucommand.cpp

diff --git a/classifyotucommand.cpp b/classifyotucommand.cpp
index 47b18410e6e2d61d2e58ec1b2ab170fbce34285f..160928f3aa46906900ac5262f198112fb70fda36 100644 (file)
--- a/classifyotucommand.cpp
+++ b/classifyotucommand.cpp
@@ -42,7 +42,7 @@ vector<string> ClassifyOtuCommand::setParameters(){
 string ClassifyOtuCommand::getHelpString(){    
        try {
                string helpString = "";
-               helpString += "The classify.otu command parameters are list, taxonomy, reftaxonomy, name, group, count, cutoff, label, basis and probs.  The taxonomy and list parameters are required unless you have a valid current file.\n";
+               helpString += "The classify.otu command parameters are list, taxonomy, reftaxonomy, name, group, count, persample, cutoff, label, basis and probs.  The taxonomy and list parameters are required unless you have a valid current file.\n";
                helpString += "The reftaxonomy parameter allows you give the name of the reference taxonomy file used when you classified your sequences. Providing it will keep the rankIDs in the summary file static.\n";
                helpString += "The name parameter allows you add a names file with your taxonomy file.\n";
                helpString += "The group parameter allows you provide a group file to use in creating the summary file breakdown.\n";
@@ -53,6 +53,7 @@ string ClassifyOtuCommand::getHelpString(){
                helpString += "Now for basis=otu could give Clostridiales       3       7       6       1       2, where 7 is the number of otus that classified to Clostridiales.\n";
                helpString += "6 is the number of otus containing sequences from groupA, 1 is the number of otus containing sequences from groupB, and 2 is the number of otus containing sequences from groupC.\n";
                helpString += "The label parameter allows you to select what distance levels you would like a output files created for, and is separated by dashes.\n";
+        helpString += "The persample parameter allows you to find a consensus taxonomy for each group. Default=f\n";
                helpString += "The default value for label is all labels in your inputfile.\n";
                helpString += "The cutoff parameter allows you to specify a consensus confidence threshold for your taxonomy.  The default is 51, meaning 51%. Cutoff cannot be below 51.\n";
                helpString += "The probs parameter shuts off the outputting of the consensus confidence results. The default is true, meaning you want the confidence to be shown.\n";
@@ -292,7 +293,7 @@ int ClassifyOtuCommand::execute(){
                if (namefile != "")     {       m->readNames(namefile, nameMap, true);  }
         if (groupfile != "")    {   groupMap = new GroupMap(groupfile);  groupMap->readMap();  groups = groupMap->getNamesOfGroups(); }
         else { groupMap = NULL;  }
-        if (countfile != "") {  ct = new CountTable(); ct->readTable(countfile);  if (ct->hasGroupInfo()) { groups = ct->getNamesOfGroups(); } }
+        if (countfile != "") {  ct = new CountTable(); ct->readTable(countfile, true, false);  if (ct->hasGroupInfo()) { groups = ct->getNamesOfGroups(); } }
         else {  ct = NULL;    }
         
                //read taxonomy file and save in map for easy access in building bin trees
@@ -585,6 +586,7 @@ int ClassifyOtuCommand::process(ListVector* processList) {
         
                //for each bin in the list vector
         string snumBins = toString(processList->getNumBins());
+        vector<string> binLabels = processList->getLabels();
                for (int i = 0; i < processList->getNumBins(); i++) {
                        
                        if (m->control_pressed) { break; }
@@ -597,17 +599,8 @@ int ClassifyOtuCommand::process(ListVector* processList) {
                        names = findConsensusTaxonomy(thisNames, size, conTax);
                
                        if (m->control_pressed) { break; }
-                       
-                       //output to new names file
-            string binLabel = "Otu";
-            string sbinNumber = toString(i+1);
-            if (sbinNumber.length() < snumBins.length()) { 
-                int diff = snumBins.length() - sbinNumber.length();
-                for (int h = 0; h < diff; h++) { binLabel += "0"; }
-            }
-            binLabel += sbinNumber;
 
-                       out << binLabel << '\t' << size << '\t' << conTax << endl;
+                       out << binLabels[i] << '\t' << size << '\t' << conTax << endl;
                        
                        string noConfidenceConTax = conTax;
                        m->removeConfidences(noConfidenceConTax);
@@ -615,9 +608,10 @@ int ClassifyOtuCommand::process(ListVector* processList) {
                        //add this bins taxonomy to summary
                        if (basis == "sequence") {
                                for(int j = 0; j < names.size(); j++) {  
-                    int numReps = 1;
-                    if (countfile != "") {  numReps = ct->getNumSeqs(names[j]); }
-                    for(int k = 0; k < numReps; k++) {  taxaSum->addSeqToTree(names[j], noConfidenceConTax);  }
+                    //int numReps = 1;
+                    //if (countfile != "") {  numReps = ct->getNumSeqs(names[j]); }
+                    //for(int k = 0; k < numReps; k++) {  taxaSum->addSeqToTree(names[j], noConfidenceConTax);  }
+                    taxaSum->addSeqToTree(names[j], noConfidenceConTax);
                 }
                        }else { //otu
                 map<string, bool> containsGroup; 
@@ -681,16 +675,8 @@ int ClassifyOtuCommand::process(ListVector* processList) {
                     
                     if (m->control_pressed) { break; }
                     
-                    //output to new names file
-                    string binLabel = "Otu";
-                    string sbinNumber = toString(i+1);
-                    if (sbinNumber.length() < snumBins.length()) { 
-                        int diff = snumBins.length() - sbinNumber.length();
-                        for (int h = 0; h < diff; h++) { binLabel += "0"; }
-                    }
-                    binLabel += sbinNumber;
                     
-                    (*outs[groupIndex[itParsed->first]]) << binLabel << '\t' << size << '\t' << conTax << endl;
+                    (*outs[groupIndex[itParsed->first]]) << binLabels[i] << '\t' << size << '\t' << conTax << endl;
                     
                     string noConfidenceConTax = conTax;
                     m->removeConfidences(noConfidenceConTax);