fixed bug with dist.shared subsampling. added mode parameter to dist.shared so...

[mothur.git] / normalizesharedcommand.cpp

diff --git a/normalizesharedcommand.cpp b/normalizesharedcommand.cpp
index 51c755cc6bdc91d767e56c79771bfe1d81fef2c4..bbf9a8335a3c2847f89b8cf136670c03a73e9725 100644 (file)
--- a/normalizesharedcommand.cpp
+++ b/normalizesharedcommand.cpp
@@ -166,7 +166,7 @@ NormalizeSharedCommand::NormalizeSharedCommand(string option) {
                        else { 
                                pickedGroups = true;
                                m->splitAtDash(groups, Groups);
-                               m->Groups = Groups;
+                               m->setGroups(Groups);
                        }
                        
                        method = validParameter.validFile(parameters, "method", false);                         if (method == "not found") { method = "totalgroup"; }
@@ -176,7 +176,7 @@ NormalizeSharedCommand::NormalizeSharedCommand(string option) {
                        if (temp == "not found") {  
                                norm = 0;  //once you have read, set norm to smallest group number
                        }else { 
-                               convert(temp, norm);
+                               m->mothurConvert(temp, norm);
                                if (norm < 0) { m->mothurOut("norm must be positive."); m->mothurOutEndLine(); abort=true; }
                        }
                        
@@ -206,18 +206,20 @@ int NormalizeSharedCommand::execute(){
                        
                        //look for groups whose numseqs is below norm and remove them, warning the user
                        if (norm != 0) { 
-                               m->Groups.clear();
+                               m->clearGroups();
+                               vector<string> mGroups;
                                vector<SharedRAbundVector*> temp;
                                for (int i = 0; i < lookup.size(); i++) {
                                        if (lookup[i]->getNumSeqs() < norm) { 
                                                m->mothurOut(lookup[i]->getGroup() + " contains " + toString(lookup[i]->getNumSeqs()) + ". Eliminating."); m->mothurOutEndLine();
                                                delete lookup[i];
                                        }else { 
-                                               m->Groups.push_back(lookup[i]->getGroup()); 
+                                               mGroups.push_back(lookup[i]->getGroup()); 
                                                temp.push_back(lookup[i]);
                                        }
                                } 
                                lookup = temp;
+                               m->setGroups(mGroups);
                        }
                        
                        //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
@@ -240,7 +242,7 @@ int NormalizeSharedCommand::execute(){
                        //as long as you are not at the end of the file or done wih the lines you want
                        while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
                                
-                               if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]);        } outputTypes.clear();  for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  } m->Groups.clear();   return 0; }
+                               if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]);        } outputTypes.clear();  for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  } m->clearGroups();   return 0; }
                                
                                if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){                  
                                        
@@ -271,13 +273,13 @@ int NormalizeSharedCommand::execute(){
                                //prevent memory leak
                                for (int i = 0; i < lookup.size(); i++) {  delete lookup[i]; lookup[i] = NULL; }
                                
-                               if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]);        } outputTypes.clear(); m->Groups.clear();  return 0; }
+                               if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]);        } outputTypes.clear(); m->clearGroups();  return 0; }
                                
                                //get next line to process
                                lookup = input->getSharedRAbundVectors();                               
                        }
                        
-                       if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]);        } outputTypes.clear(); m->Groups.clear();   return 0; }
+                       if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]);        } outputTypes.clear(); m->clearGroups();   return 0; }
                        
                        //output error messages about any remaining user labels
                        set<string>::iterator it;
@@ -314,18 +316,20 @@ int NormalizeSharedCommand::execute(){
                        
                        //look for groups whose numseqs is below norm and remove them, warning the user
                        if (norm != 0) { 
-                               m->Groups.clear();
+                               m->clearGroups();
+                               vector<string> mGroups;
                                vector<SharedRAbundFloatVector*> temp;
                                for (int i = 0; i < lookupFloat.size(); i++) {
                                        if (lookupFloat[i]->getNumSeqs() < norm) { 
                                                m->mothurOut(lookupFloat[i]->getGroup() + " contains " + toString(lookupFloat[i]->getNumSeqs()) + ". Eliminating."); m->mothurOutEndLine();
                                                delete lookupFloat[i];
                                        }else { 
-                                               m->Groups.push_back(lookupFloat[i]->getGroup()); 
+                                               mGroups.push_back(lookupFloat[i]->getGroup()); 
                                                temp.push_back(lookupFloat[i]);
                                        }
                                } 
                                lookupFloat = temp;
+                               m->setGroups(mGroups);
                        }
                        
                        //set norm to smallest group number
@@ -343,7 +347,7 @@ int NormalizeSharedCommand::execute(){
                        //as long as you are not at the end of the file or done wih the lines you want
                        while((lookupFloat[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
                                
-                               if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]);        } outputTypes.clear();  for (int i = 0; i < lookupFloat.size(); i++) {  delete lookupFloat[i];  } m->Groups.clear();  return 0; }
+                               if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]);        } outputTypes.clear();  for (int i = 0; i < lookupFloat.size(); i++) {  delete lookupFloat[i];  } m->clearGroups();  return 0; }
                                
                                if(allLines == 1 || labels.count(lookupFloat[0]->getLabel()) == 1){                     
                                        
@@ -376,13 +380,13 @@ int NormalizeSharedCommand::execute(){
                                //prevent memory leak
                                for (int i = 0; i < lookupFloat.size(); i++) {  delete lookupFloat[i]; lookupFloat[i] = NULL; }
                                
-                               if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]);        } outputTypes.clear(); m->Groups.clear();   return 0; }
+                               if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]);        } outputTypes.clear(); m->clearGroups();   return 0; }
                                
                                //get next line to process
                                lookupFloat = input->getSharedRAbundFloatVectors();                             
                        }
                        
-                       if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]);        } outputTypes.clear(); m->Groups.clear();   return 0; }
+                       if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]);        } outputTypes.clear(); m->clearGroups();   return 0; }
                        
                        //output error messages about any remaining user labels
                        set<string>::iterator it;
@@ -411,7 +415,7 @@ int NormalizeSharedCommand::execute(){
                        
                }
                //reset groups parameter
-               m->Groups.clear();  
+               m->clearGroups();  
                delete input;
                
                if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]);        } outputTypes.clear(); return 0;}
@@ -628,6 +632,7 @@ int NormalizeSharedCommand::eliminateZeroOTUS(vector<SharedRAbundVector*>& thisl
                
                //for each bin
                vector<string> newBinLabels;
+               string snumBins = toString(thislookup[0]->getNumBins());
                for (int i = 0; i < thislookup[0]->getNumBins(); i++) {
                        if (m->control_pressed) { for (int j = 0; j < newLookup.size(); j++) {  delete newLookup[j];  } return 0; }
                
@@ -643,7 +648,13 @@ int NormalizeSharedCommand::eliminateZeroOTUS(vector<SharedRAbundVector*>& thisl
                                        newLookup[j]->push_back(thislookup[j]->getAbundance(i), thislookup[j]->getGroup());
                                }
                                //if there is a bin label use it otherwise make one
-                               string binLabel = "Otu" + toString(i+1);
+                               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; 
                                if (i < m->currentBinLabels.size()) {  binLabel = m->currentBinLabels[i]; }
                                
                                newBinLabels.push_back(binLabel);
@@ -677,6 +688,7 @@ int NormalizeSharedCommand::eliminateZeroOTUS(vector<SharedRAbundFloatVector*>&
                
                //for each bin
                vector<string> newBinLabels;
+               string snumBins = toString(thislookup[0]->getNumBins());
                for (int i = 0; i < thislookup[0]->getNumBins(); i++) {
                        if (m->control_pressed) { for (int j = 0; j < newLookup.size(); j++) {  delete newLookup[j];  } return 0; }
                        
@@ -692,7 +704,13 @@ int NormalizeSharedCommand::eliminateZeroOTUS(vector<SharedRAbundFloatVector*>&
                                        newLookup[j]->push_back(thislookup[j]->getAbundance(i), thislookup[j]->getGroup());
                                }
                                //if there is a bin label use it otherwise make one
-                               string binLabel = "Otu" + toString(i+1);
+                               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; 
                                if (i < m->currentBinLabels.size()) {  binLabel = m->currentBinLabels[i]; }
                                
                                newBinLabels.push_back(binLabel);