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

[mothur.git] / mgclustercommand.cpp

diff --git a/mgclustercommand.cpp b/mgclustercommand.cpp
index e5681efb220d1635382ea6b077a36ea4e0f50226..dd98b3764884f5c50009d8e9e7422a71681d94c1 100644 (file)
--- a/mgclustercommand.cpp
+++ b/mgclustercommand.cpp
@@ -14,6 +14,7 @@ vector<string> MGClusterCommand::setParameters(){
        try {
                CommandParameter pblast("blast", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pblast);
                CommandParameter pname("name", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pname);
+        CommandParameter pcount("count", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pcount);
                CommandParameter plength("length", "Number", "", "5", "", "", "",false,false); parameters.push_back(plength);
                CommandParameter ppenalty("penalty", "Number", "", "0.10", "", "", "",false,false); parameters.push_back(ppenalty);
                CommandParameter pcutoff("cutoff", "Number", "", "0.70", "", "", "",false,false); parameters.push_back(pcutoff);
@@ -164,12 +165,19 @@ MGClusterCommand::MGClusterCommand(string option) {
                        if (namefile == "not open") { abort = true; }   
                        else if (namefile == "not found") { namefile = ""; }
                        else { m->setNameFile(namefile); }
+            
+            countfile = validParameter.validFile(parameters, "count", true);
+                       if (countfile == "not open") { abort = true; }  
+                       else if (countfile == "not found") { countfile = ""; }
+            else { m->setCountTableFile(countfile); }
+            
+            if (countfile != "" && namefile != "") { m->mothurOut("[ERROR]: Cannot have both a name file and count file. Please use one or the other."); m->mothurOutEndLine(); abort = true; }
                        
                        if ((blastfile == "")) { m->mothurOut("When executing a mgcluster command you must provide a blastfile."); m->mothurOutEndLine(); abort = true; }
                        
                        //check for optional parameter and set defaults
                        string temp;
-                       temp = validParameter.validFile(parameters, "precision", false);                if (temp == "not found") { temp = "100"; }
+            temp = validParameter.validFile(parameters, "precision", false);           if (temp == "not found") { temp = "100"; }
                        precisionLength = temp.length();
                        m->mothurConvert(temp, precision); 
                        
@@ -199,7 +207,7 @@ MGClusterCommand::MGClusterCommand(string option) {
                        hclusterWanted = m->isTrue(temp); 
                        
                        temp = validParameter.validFile(parameters, "hard", false);                     if (temp == "not found") { temp = "T"; }
-                       hard = m->isTrue(temp);
+                       hard = m->isTrue(temp);            
                }
 
        }
@@ -211,7 +219,6 @@ MGClusterCommand::MGClusterCommand(string option) {
 //**********************************************************************************************************************
 int MGClusterCommand::execute(){
        try {
-               
                if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
                
                //read names file
@@ -225,7 +232,7 @@ int MGClusterCommand::execute(){
                time_t start;
                float previousDist = 0.00000;
                float rndPreviousDist = 0.00000; 
-               
+        
                //read blastfile - creates sparsematrices for the distances and overlaps as well as a listvector
                //must remember to delete those objects here since readBlast does not
                read = new ReadBlast(blastfile, cutoff, penalty, length, minWanted, hclusterWanted);
@@ -233,16 +240,20 @@ int MGClusterCommand::execute(){
         
         list = new ListVector(nameMap->getListVector());
         RAbundVector* rabund = NULL;
-        if(large) {
-            map<string, int> nameMapCounts = m->readNames(namefile);
-            RAbundVector* rabund = createRabund(list, nameMapCounts);
+        
+        if(countfile != "") {
+            //map<string, int> nameMapCounts = m->readNames(namefile);
+            ct = new CountTable();
+            ct->readTable(countfile);
+            rabund = new RAbundVector();
+            createRabund(ct, list, rabund);
         }else {
-            RAbundVector* rabund = new RAbundVector(list->getRAbundVector());
+            rabund = new RAbundVector(list->getRAbundVector());
         }
         
                 
                //list = new ListVector(nameMap->getListVector());
-               //RAbundVector* rabund = new RAbundVector(list->getRAbundVector());
+               //rabund = new RAbundVector(list->getRAbundVector());
                
                if (m->control_pressed) { outputTypes.clear(); delete nameMap; delete read; delete list; delete rabund; return 0; }
                
@@ -259,13 +270,15 @@ int MGClusterCommand::execute(){
         string rabundFileName = fileroot+ tag + "." + getOutputFileNameTag("rabund");
         string listFileName = fileroot+ tag + "." + getOutputFileNameTag("list");
         
-               m->openOutputFile(sabundFileName,       sabundFile);
-               m->openOutputFile(rabundFileName,       rabundFile);
+        if (countfile == "") {
+            m->openOutputFile(sabundFileName,  sabundFile);
+            m->openOutputFile(rabundFileName,  rabundFile);
+        }
                m->openOutputFile(listFileName, listFile);
                
                if (m->control_pressed) { 
                        delete nameMap; delete read; delete list; delete rabund; 
-                       listFile.close(); rabundFile.close(); sabundFile.close(); m->mothurRemove((fileroot+ tag + ".list")); m->mothurRemove((fileroot+ tag + ".rabund")); m->mothurRemove((fileroot+ tag + ".sabund"));
+                       listFile.close(); if (countfile == "") { rabundFile.close(); sabundFile.close();  m->mothurRemove((fileroot+ tag + ".rabund")); m->mothurRemove((fileroot+ tag + ".sabund")); } m->mothurRemove((fileroot+ tag + ".list"));
                        outputTypes.clear();
                        return 0; 
                }
@@ -274,7 +287,7 @@ int MGClusterCommand::execute(){
                
                if (!hclusterWanted) {
                        //get distmatrix and overlap
-                       SparseMatrix* distMatrix = read->getDistMatrix();
+                       SparseDistanceMatrix* distMatrix = read->getDistMatrix();
                        overlapMatrix = read->getOverlapMatrix(); //already sorted by read 
                        delete read;
                
@@ -288,7 +301,7 @@ int MGClusterCommand::execute(){
                        
                        if (m->control_pressed) { 
                                delete nameMap; delete distMatrix; delete list; delete rabund; delete cluster;
-                               listFile.close(); rabundFile.close(); sabundFile.close(); m->mothurRemove((fileroot+ tag + ".list")); m->mothurRemove((fileroot+ tag + ".rabund")); m->mothurRemove((fileroot+ tag + ".sabund"));
+                               listFile.close(); if (countfile == "") { rabundFile.close(); sabundFile.close();  m->mothurRemove((fileroot+ tag + ".rabund")); m->mothurRemove((fileroot+ tag + ".sabund")); } m->mothurRemove((fileroot+ tag + ".list"));
                                outputTypes.clear();
                                return 0; 
                        }
@@ -300,7 +313,7 @@ int MGClusterCommand::execute(){
                                
                                if (m->control_pressed) { 
                                        delete nameMap; delete distMatrix; delete list; delete rabund; delete cluster;
-                                       listFile.close(); rabundFile.close(); sabundFile.close(); m->mothurRemove((fileroot+ tag + ".list")); m->mothurRemove((fileroot+ tag + ".rabund")); m->mothurRemove((fileroot+ tag + ".sabund"));
+                                       listFile.close(); if (countfile == "") { rabundFile.close(); sabundFile.close();  m->mothurRemove((fileroot+ tag + ".rabund")); m->mothurRemove((fileroot+ tag + ".sabund")); } m->mothurRemove((fileroot+ tag + ".list"));
                                        outputTypes.clear();
                                        return 0; 
                                }
@@ -323,7 +336,7 @@ int MGClusterCommand::execute(){
                                                
                                                if (m->control_pressed) { 
                                                        delete nameMap; delete distMatrix; delete list; delete rabund; delete cluster; delete temp;
-                                                       listFile.close(); rabundFile.close(); sabundFile.close(); m->mothurRemove((fileroot+ tag + ".list")); m->mothurRemove((fileroot+ tag + ".rabund")); m->mothurRemove((fileroot+ tag + ".sabund"));
+                                                       listFile.close(); if (countfile == "") { rabundFile.close(); sabundFile.close();  m->mothurRemove((fileroot+ tag + ".rabund")); m->mothurRemove((fileroot+ tag + ".sabund")); } m->mothurRemove((fileroot+ tag + ".list"));
                                                        outputTypes.clear();
                                                        return 0; 
                                                }
@@ -354,7 +367,7 @@ int MGClusterCommand::execute(){
                                        
                                        if (m->control_pressed) { 
                                                        delete nameMap; delete distMatrix; delete list; delete rabund; delete cluster; delete temp;
-                                                       listFile.close(); rabundFile.close(); sabundFile.close(); m->mothurRemove((fileroot+ tag + ".list")); m->mothurRemove((fileroot+ tag + ".rabund")); m->mothurRemove((fileroot+ tag + ".sabund"));
+                                                       listFile.close(); if (countfile == "") { rabundFile.close(); sabundFile.close();  m->mothurRemove((fileroot+ tag + ".rabund")); m->mothurRemove((fileroot+ tag + ".sabund")); } m->mothurRemove((fileroot+ tag + ".list"));
                                                        outputTypes.clear();
                                                        return 0; 
                                        }
@@ -384,7 +397,7 @@ int MGClusterCommand::execute(){
                        
                        if (m->control_pressed) { 
                                delete nameMap;  delete list; delete rabund; 
-                               listFile.close(); rabundFile.close(); sabundFile.close(); m->mothurRemove((fileroot+ tag + ".list")); m->mothurRemove((fileroot+ tag + ".rabund")); m->mothurRemove((fileroot+ tag + ".sabund"));
+                               listFile.close(); if (countfile == "") { rabundFile.close(); sabundFile.close();  m->mothurRemove((fileroot+ tag + ".rabund")); m->mothurRemove((fileroot+ tag + ".sabund")); } m->mothurRemove((fileroot+ tag + ".list"));
                                outputTypes.clear();
                                return 0; 
                        }
@@ -401,7 +414,7 @@ int MGClusterCommand::execute(){
                        
                        if (m->control_pressed) { 
                                delete nameMap;  delete list; delete rabund; delete hcluster;
-                               listFile.close(); rabundFile.close(); sabundFile.close(); m->mothurRemove((fileroot+ tag + ".list")); m->mothurRemove((fileroot+ tag + ".rabund")); m->mothurRemove((fileroot+ tag + ".sabund"));
+                               listFile.close(); if (countfile == "") { rabundFile.close(); sabundFile.close();  m->mothurRemove((fileroot+ tag + ".rabund")); m->mothurRemove((fileroot+ tag + ".sabund")); } m->mothurRemove((fileroot+ tag + ".list"));
                                outputTypes.clear();
                                return 0; 
                        }
@@ -417,7 +430,7 @@ int MGClusterCommand::execute(){
                                
                                if (m->control_pressed) { 
                                        delete nameMap;  delete list; delete rabund; delete hcluster;
-                                       listFile.close(); rabundFile.close(); sabundFile.close(); m->mothurRemove((fileroot+ tag + ".list")); m->mothurRemove((fileroot+ tag + ".rabund")); m->mothurRemove((fileroot+ tag + ".sabund"));
+                                       listFile.close(); if (countfile == "") { rabundFile.close(); sabundFile.close();  m->mothurRemove((fileroot+ tag + ".rabund")); m->mothurRemove((fileroot+ tag + ".sabund")); } m->mothurRemove((fileroot+ tag + ".list"));
                                        m->mothurRemove(distFile);
                                        m->mothurRemove(overlapFile);
                                        outputTypes.clear();
@@ -432,7 +445,7 @@ int MGClusterCommand::execute(){
                                                
                                                if (m->control_pressed) { 
                                                        delete nameMap;  delete list; delete rabund; delete hcluster;
-                                                       listFile.close(); rabundFile.close(); sabundFile.close(); m->mothurRemove((fileroot+ tag + ".list")); m->mothurRemove((fileroot+ tag + ".rabund")); m->mothurRemove((fileroot+ tag + ".sabund"));
+                                                       listFile.close(); if (countfile == "") { rabundFile.close(); sabundFile.close();  m->mothurRemove((fileroot+ tag + ".rabund")); m->mothurRemove((fileroot+ tag + ".sabund")); } m->mothurRemove((fileroot+ tag + ".list"));
                                                        m->mothurRemove(distFile);
                                                        m->mothurRemove(overlapFile);
                                                        outputTypes.clear();
@@ -456,7 +469,7 @@ int MGClusterCommand::execute(){
                                                                
                                                                if (m->control_pressed) { 
                                                                        delete nameMap;  delete list; delete rabund; delete hcluster; delete temp;
-                                                                       listFile.close(); rabundFile.close(); sabundFile.close(); m->mothurRemove((fileroot+ tag + ".list")); m->mothurRemove((fileroot+ tag + ".rabund")); m->mothurRemove((fileroot+ tag + ".sabund"));
+                                                                       listFile.close(); if (countfile == "") { rabundFile.close(); sabundFile.close();  m->mothurRemove((fileroot+ tag + ".rabund")); m->mothurRemove((fileroot+ tag + ".sabund")); } m->mothurRemove((fileroot+ tag + ".list"));
                                                                        m->mothurRemove(distFile);
                                                                        m->mothurRemove(overlapFile);
                                                                        outputTypes.clear();
@@ -492,7 +505,7 @@ int MGClusterCommand::execute(){
                                        
                                        if (m->control_pressed) { 
                                                        delete nameMap; delete list; delete rabund; delete hcluster; delete temp;
-                                                       listFile.close(); rabundFile.close(); sabundFile.close(); m->mothurRemove((fileroot+ tag + ".list")); m->mothurRemove((fileroot+ tag + ".rabund")); m->mothurRemove((fileroot+ tag + ".sabund"));
+                                                       listFile.close(); if (countfile == "") { rabundFile.close(); sabundFile.close();  m->mothurRemove((fileroot+ tag + ".rabund")); m->mothurRemove((fileroot+ tag + ".sabund")); } m->mothurRemove((fileroot+ tag + ".list"));
                                                        m->mothurRemove(distFile);
                                                        m->mothurRemove(overlapFile);
                                                        outputTypes.clear();
@@ -513,15 +526,16 @@ int MGClusterCommand::execute(){
                        m->mothurRemove(overlapFile);
                }
                
-               delete list; 
+               delete list;
                delete rabund;
                listFile.close();
-               sabundFile.close();
-               rabundFile.close();
-       
+        if (countfile == "") {
+            sabundFile.close();
+            rabundFile.close();
+        }
                if (m->control_pressed) { 
                        delete nameMap; 
-                       listFile.close(); rabundFile.close(); sabundFile.close(); m->mothurRemove((fileroot+ tag + ".list")); m->mothurRemove((fileroot+ tag + ".rabund")); m->mothurRemove((fileroot+ tag + ".sabund"));
+                       listFile.close(); if (countfile == "") { rabundFile.close(); sabundFile.close();  m->mothurRemove((fileroot+ tag + ".rabund")); m->mothurRemove((fileroot+ tag + ".sabund")); } m->mothurRemove((fileroot+ tag + ".list"));
                        outputTypes.clear();
                        return 0; 
                }
@@ -529,8 +543,10 @@ int MGClusterCommand::execute(){
                m->mothurOutEndLine();
                m->mothurOut("Output File Names: "); m->mothurOutEndLine();
                m->mothurOut(listFileName); m->mothurOutEndLine();      outputNames.push_back(listFileName); outputTypes["list"].push_back(listFileName);
-               m->mothurOut(rabundFileName); m->mothurOutEndLine();    outputNames.push_back(rabundFileName); outputTypes["rabund"].push_back(rabundFileName);
-               m->mothurOut(sabundFileName); m->mothurOutEndLine();    outputNames.push_back(sabundFileName); outputTypes["sabund"].push_back(sabundFileName);
+               if (countfile == "") {
+            m->mothurOut(rabundFileName); m->mothurOutEndLine();       outputNames.push_back(rabundFileName); outputTypes["rabund"].push_back(rabundFileName);
+            m->mothurOut(sabundFileName); m->mothurOutEndLine();       outputNames.push_back(sabundFileName); outputTypes["sabund"].push_back(sabundFileName);
+        }
                m->mothurOutEndLine();
                
                if (saveCutoff != cutoff) { 
@@ -573,12 +589,14 @@ int MGClusterCommand::execute(){
 void MGClusterCommand::printData(ListVector* mergedList){
        try {
                mergedList->print(listFile);
-               mergedList->getRAbundVector().print(rabundFile);
-               
-               SAbundVector sabund = mergedList->getSAbundVector();
+        SAbundVector sabund = mergedList->getSAbundVector();
+        
+        if (countfile == "") {
+            mergedList->getRAbundVector().print(rabundFile);
+            sabund.print(sabundFile);
+        }
 
                sabund.print(cout);
-               sabund.print(sabundFile);
        }
        catch(exception& e) {
                m->errorOut(e, "MGClusterCommand", "printData");
@@ -709,11 +727,25 @@ void MGClusterCommand::sortHclusterFiles(string unsortedDist, string unsortedOve
 
 //**********************************************************************************************************************
 
-RAbundVector MGClusterCommand::createRabund(ListVector list, map<string, int> nameMapCounts){
+void MGClusterCommand::createRabund(CountTable*& ct, ListVector*& list, RAbundVector*& rabund){
     try {
-        RAbundVector rav;
-        for(int i = 0; i < list->getNumBins(); i++) {   rav.push_back(nameMapCounts[i]);    }
-        return rav;
+        //vector<string> names = ct.getNamesOfSeqs();
+
+        //for ( int i; i < ct.getNumGroups(); i++ ) {    rav.push_back( ct.getNumSeqs(names[i]) );    }
+        //return rav;
+        
+        for(int i = 0; i < list->getNumBins(); i++) { 
+           vector<string> binNames;
+           string bin = list->get(i);
+           m->splitAtComma(bin, binNames);
+           int total = 0;
+           for (int j = 0; j < binNames.size(); j++) { 
+               total += ct->getNumSeqs(binNames[j]);
+           }
+           rabund->push_back(total);   
+       }
+        
+        
     }
     catch(exception& e) {
                m->errorOut(e, "MGClusterCommand", "createRabund");