]> git.donarmstrong.com Git - mothur.git/blobdiff - unweighted.cpp
added adjustDots function to pcr.seqs, keeps sequences aligned in case where primers...
[mothur.git] / unweighted.cpp
index ba88049d7e9914231b326861c3abbccdf2af01b2..56694e32e3a977b08df087dfc8deeabb2524564c 100644 (file)
 
 EstOutput Unweighted::getValues(Tree* t, int p, string o) {
        try {
-               globaldata = GlobalData::getInstance();
                processors = p;
                outputDir = o;
-                       
+        
+        CountTable* ct = t->getCountTable();
+        
                //if the users enters no groups then give them the score of all groups
-               int numGroups = globaldata->Groups.size();
+               int numGroups = m->getNumGroups();
                
                //calculate number of comparsions
                int numComp = 0;
                vector< vector<string> > namesOfGroupCombos;
                for (int r=0; r<numGroups; r++) { 
-                       for (int l = r+1; l < numGroups; l++) {
+                       for (int l = 0; l < r; l++) {
                                numComp++;
-                               vector<string> groups; groups.push_back(globaldata->Groups[r]); groups.push_back(globaldata->Groups[l]);
+                               vector<string> groups; groups.push_back((m->getGroups())[r]); groups.push_back((m->getGroups())[l]);
                                namesOfGroupCombos.push_back(groups);
                        }
                }
@@ -35,23 +36,23 @@ EstOutput Unweighted::getValues(Tree* t, int p, string o) {
                        vector<string> groups;
                        if (numGroups == 0) {
                                //get score for all users groups
-                               for (int i = 0; i < tmap->namesOfGroups.size(); i++) {
-                                       if (tmap->namesOfGroups[i] != "xxx") {
-                                               groups.push_back(tmap->namesOfGroups[i]);
+                               for (int i = 0; i < (ct->getNamesOfGroups()).size(); i++) {
+                                       if ((ct->getNamesOfGroups())[i] != "xxx") {
+                                               groups.push_back((ct->getNamesOfGroups())[i]);
                                        }
                                }
                                namesOfGroupCombos.push_back(groups);
                        }else {
-                               for (int i = 0; i < globaldata->Groups.size(); i++) {
-                                       groups.push_back(globaldata->Groups[i]);
+                               for (int i = 0; i < m->getNumGroups(); i++) {
+                                       groups.push_back((m->getGroups())[i]);
                                }
                                namesOfGroupCombos.push_back(groups);
                        }
                }
 
-               #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+               #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
                        if(processors == 1){
-                               data = driver(t, namesOfGroupCombos, 0, namesOfGroupCombos.size());
+                               data = driver(t, namesOfGroupCombos, 0, namesOfGroupCombos.size(), ct);
                        }else{
                                int numPairs = namesOfGroupCombos.size();
                                
@@ -59,18 +60,18 @@ EstOutput Unweighted::getValues(Tree* t, int p, string o) {
                                
                                for (int i = 0; i < processors; i++) {
                                        int startPos = i * numPairsPerProcessor;
+                       
                                        if(i == processors - 1){
                                                numPairsPerProcessor = numPairs - i * numPairsPerProcessor;
                                        }
-                                       lines.push_back(new linePair(startPos, numPairsPerProcessor));
+               
+                                       lines.push_back(linePair(startPos, numPairsPerProcessor));
                                }
-
-                               data = createProcesses(t, namesOfGroupCombos);
-                               
-                               for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
+                               data = createProcesses(t, namesOfGroupCombos, ct);
+                               lines.clear();
                        }
                #else
-                       data = driver(t, namesOfGroupCombos, 0, namesOfGroupCombos.size());
+                       data = driver(t, namesOfGroupCombos, 0, namesOfGroupCombos.size(), ct);
                #endif
                
                return data;
@@ -82,11 +83,10 @@ EstOutput Unweighted::getValues(Tree* t, int p, string o) {
 }
 /**************************************************************************************************/
 
-EstOutput Unweighted::createProcesses(Tree* t, vector< vector<string> > namesOfGroupCombos) {
+EstOutput Unweighted::createProcesses(Tree* t, vector< vector<string> > namesOfGroupCombos, CountTable* ct) {
        try {
-#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
                int process = 1;
-               int num = 0;
                vector<int> processIDS;
                
                EstOutput results;
@@ -100,11 +100,11 @@ EstOutput Unweighted::createProcesses(Tree* t, vector< vector<string> > namesOfG
                                process++;
                        }else if (pid == 0){
                                EstOutput myresults;
-                               myresults = driver(t, namesOfGroupCombos, lines[process]->start, lines[process]->num);
+                               myresults = driver(t, namesOfGroupCombos, lines[process].start, lines[process].num, ct);
                                
                                if (m->control_pressed) { exit(0); }
                                
-                               m->mothurOut("Merging results."); m->mothurOutEndLine();
+                               //m->mothurOut("Merging results."); m->mothurOutEndLine();
                                
                                //pass numSeqs to parent
                                ofstream out;
@@ -115,10 +115,14 @@ EstOutput Unweighted::createProcesses(Tree* t, vector< vector<string> > namesOfG
                                out.close();
                                
                                exit(0);
-                       }else { m->mothurOut("unable to spawn the necessary processes."); m->mothurOutEndLine(); exit(0); }
+                       }else { 
+                               m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine(); 
+                               for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
+                               exit(0); 
+                       }
                }
                
-               results = driver(t, namesOfGroupCombos, lines[0]->start, lines[0]->num);
+               results = driver(t, namesOfGroupCombos, lines[0].start, lines[0].num, ct);
                
                //force parent to wait until all the processes are done
                for (int i=0;i<(processors-1);i++) { 
@@ -149,10 +153,10 @@ EstOutput Unweighted::createProcesses(Tree* t, vector< vector<string> > namesOfG
                                m->gobble(in);
                        }
                        in.close();
-                       remove(s.c_str());
+                       m->mothurRemove(s);
                }
                
-               m->mothurOut("DONE."); m->mothurOutEndLine(); m->mothurOutEndLine();
+               //m->mothurOut("DONE."); m->mothurOutEndLine(); m->mothurOutEndLine();
                
                return results;
 #endif         
@@ -163,57 +167,85 @@ EstOutput Unweighted::createProcesses(Tree* t, vector< vector<string> > namesOfG
        }
 }
 /**************************************************************************************************/
-EstOutput Unweighted::driver(Tree* t, vector< vector<string> > namesOfGroupCombos, int start, int num) { 
+EstOutput Unweighted::driver(Tree* t, vector< vector<string> > namesOfGroupCombos, int start, int num, CountTable* ct) { 
  try {
-               
+       
+        
                EstOutput results; results.resize(num);
                
                int count = 0;
                int total = num;
                int twentyPercent = (total * 0.20);
-
+               if (twentyPercent == 0) { twentyPercent = 1; }
+        
+                       
                for (int h = start; h < (start+num); h++) {
-               
+                               
                        if (m->control_pressed) { return results; }
                
                        double UniqueBL=0.0000;  //a branch length is unique if it's chidren are from the same group
                        double totalBL = 0.00;  //all branch lengths
                        double UW = 0.00;               //Unweighted Value = UniqueBL / totalBL;
-                               
-                       for(int i=0;i<t->getNumNodes();i++){
+                                               
+                       //find a node that belongs to one of the groups in this combo
+                       int nodeBelonging = -1;
+                       for (int g = 0; g < namesOfGroupCombos[h].size(); g++) {
+                               if (t->groupNodeInfo[namesOfGroupCombos[h][g]].size() != 0) { nodeBelonging = t->groupNodeInfo[namesOfGroupCombos[h][g]][0]; break; }
+                       }
                        
-                               if (m->control_pressed) {  return data; }
-                               
-                               //pcountSize = 0, they are from a branch that is entirely from a group the user doesn't want
-                               //pcountSize = 2, not unique to one group
-                               //pcountSize = 1, unique to one group
-                               
-                               int pcountSize = 0;
-                               for (int j = 0; j < namesOfGroupCombos[h].size(); j++) {
-                                       map<string, int>::iterator itGroup = t->tree[i].pcount.find(namesOfGroupCombos[h][j]);
-                                       if (itGroup != t->tree[i].pcount.end()) { pcountSize++; if (pcountSize > 1) { break; } } 
-                               }
-                               
-                               if (pcountSize == 0) { }
-                               else if ((t->tree[i].getBranchLength() != -1) && (pcountSize == 1)) {  UniqueBL += abs(t->tree[i].getBranchLength());   }
+                       //sanity check
+                       if (nodeBelonging == -1) {
+                               m->mothurOut("[WARNING]: cannot find a nodes in the tree from grouping "); 
+                               for (int g = 0; g < namesOfGroupCombos[h].size()-1; g++) { m->mothurOut(namesOfGroupCombos[h][g] + "-"); }
+                               m->mothurOut(namesOfGroupCombos[h][namesOfGroupCombos[h].size()-1]);
+                               m->mothurOut(", skipping."); m->mothurOutEndLine(); results[count] = UW;
+                       }else{
+                               //cout << "trying to get root" << endl; 
+                               //if including the root this clears rootForGrouping[namesOfGroupCombos[h]]
+                               getRoot(t, nodeBelonging, namesOfGroupCombos[h]);
+                               //cout << "here" << endl;       
+                               for(int i=0;i<t->getNumNodes();i++){
+                                       
+                                       if (m->control_pressed) {  return data; }
+                                       //cout << i << endl;    
+                                       //pcountSize = 0, they are from a branch that is entirely from a group the user doesn't want
+                                       //pcountSize = 2, not unique to one group
+                                       //pcountSize = 1, unique to one group
+                                       
+                                       int pcountSize = 0;
+                                       for (int j = 0; j < namesOfGroupCombos[h].size(); j++) {
+                                               map<string, int>::iterator itGroup = t->tree[i].pcount.find(namesOfGroupCombos[h][j]);
+                                               if (itGroup != t->tree[i].pcount.end()) { pcountSize++; if (pcountSize > 1) { break; } } 
+                                       }
                                        
-                               if ((t->tree[i].getBranchLength() != -1) && (pcountSize != 0)) {  
-                                       totalBL += abs(t->tree[i].getBranchLength()); 
+                                       
+                                       //unique calc
+                                       if (pcountSize == 0) { }
+                                       else if ((t->tree[i].getBranchLength() != -1) && (pcountSize == 1) && (rootForGrouping[namesOfGroupCombos[h]].count(i) == 0)) { //you have a unique branch length and you are not the root 
+                                               UniqueBL += abs(t->tree[i].getBranchLength()); 
+                                       }
+                                               
+                                       //total calc
+                                       if (pcountSize == 0) { }
+                                       else if ((t->tree[i].getBranchLength() != -1) && (pcountSize != 0) && (rootForGrouping[namesOfGroupCombos[h]].count(i) == 0)) { //you have a branch length and you are not the root 
+                                               totalBL += abs(t->tree[i].getBranchLength()); 
+                                       }
                                }
-                       }
-               
-                       UW = (UniqueBL / totalBL);  
+       //cout << UniqueBL << '\t' << totalBL << endl;          
+                               UW = (UniqueBL / totalBL);  
        
-                       if (isnan(UW) || isinf(UW)) { UW = 0; }
+                               if (isnan(UW) || isinf(UW)) { UW = 0; }
        
-                       results[count] = UW;
+                               results[count] = UW;
+                       }
                        count++;
 
                        //report progress
-                       if((count) % twentyPercent == 0){       m->mothurOut("Percentage complete: " + toString(int((count / (float)total) * 100.0))); m->mothurOutEndLine();           }
+                       //if((count % twentyPercent) == 0) {    float tempOut = (count / (float)total); if (isnan(tempOut) || isinf(tempOut)) { tempOut = 0.0; } m->mothurOut("Percentage complete: " + toString((int(tempOut) * 100.0))); m->mothurOutEndLine();       }
                }
                
-               m->mothurOut("Percentage complete: 100"); m->mothurOutEndLine();
+               //report progress
+               //if((count % twentyPercent) != 0) {    float tempOut = (count / (float)total); if (isnan(tempOut) || isinf(tempOut)) { tempOut = 0.0; } m->mothurOut("Percentage complete: " + toString((int(tempOut) * 100.0))); m->mothurOutEndLine();       }
                
                return results; 
        }
@@ -224,50 +256,200 @@ EstOutput Unweighted::driver(Tree* t, vector< vector<string> > namesOfGroupCombo
 }
 /**************************************************************************************************/
 
-EstOutput Unweighted::getValues(Tree* t, string groupA, string groupB) { 
+EstOutput Unweighted::getValues(Tree* t, string groupA, string groupB, int p, string o) { 
  try {
-       globaldata = GlobalData::getInstance();
+               processors = p;
+               outputDir = o;
                
-               vector<string> groups;
-               double UniqueBL;  //a branch length is unique if it's chidren are from the same group
-               double totalBL; //all branch lengths
-               double UW;              //Unweighted Value = UniqueBL / totalBL;
-               copyTree = new Tree;
-
+        CountTable* ct = t->getCountTable();
+     
                //if the users enters no groups then give them the score of all groups
-               int numGroups = globaldata->Groups.size();
-
+               int numGroups = m->getNumGroups();
+               
                //calculate number of comparsions
                int numComp = 0;
+               vector< vector<string> > namesOfGroupCombos;
                for (int r=0; r<numGroups; r++) { 
-                       for (int l = r+1; l < numGroups; l++) {
+                       for (int l = 0; l < r; l++) {
                                numComp++;
+                               vector<string> groups; groups.push_back((m->getGroups())[r]); groups.push_back((m->getGroups())[l]);
+                               namesOfGroupCombos.push_back(groups);
                        }
                }
+               
+               if (numComp != 1) {
+                       vector<string> groups;
+                       if (numGroups == 0) {
+                               //get score for all users groups
+                               for (int i = 0; i < (ct->getNamesOfGroups()).size(); i++) {
+                                       if ((ct->getNamesOfGroups())[i] != "xxx") {
+                                               groups.push_back((ct->getNamesOfGroups())[i]);
+                                       }
+                               }
+                               namesOfGroupCombos.push_back(groups);
+                       }else {
+                               for (int i = 0; i < m->getNumGroups(); i++) {
+                                       groups.push_back((m->getGroups())[i]);
+                               }
+                               namesOfGroupCombos.push_back(groups);
+                       }
+               }
+
+               #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
+                       if(processors == 1){
+                               data = driver(t, namesOfGroupCombos, 0, namesOfGroupCombos.size(), true, ct);
+                       }else{
+                               int numPairs = namesOfGroupCombos.size();
+                               
+                               int numPairsPerProcessor = numPairs / processors;
+                               
+                               for (int i = 0; i < processors; i++) {
+                                       int startPos = i * numPairsPerProcessor;
+                                       if(i == processors - 1){
+                                               numPairsPerProcessor = numPairs - i * numPairsPerProcessor;
+                                       }
+                                       lines.push_back(linePair(startPos, numPairsPerProcessor));
+                               }
+                                       
+                               data = createProcesses(t, namesOfGroupCombos, true, ct);
+                               
+                               lines.clear();
+                       }
+               #else
+                       data = driver(t, namesOfGroupCombos, 0, namesOfGroupCombos.size(), true, ct);
+               #endif
+       
+               return data;
+       }
+       catch(exception& e) {
+               m->errorOut(e, "Unweighted", "getValues");
+               exit(1);
+       }
+}
+/**************************************************************************************************/
 
-               //numComp+1 for AB, AC, BC, ABC
-               data.resize(numComp+1,0);
+EstOutput Unweighted::createProcesses(Tree* t, vector< vector<string> > namesOfGroupCombos, bool usingGroups, CountTable* ct) {
+       try {
+#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
+               int process = 1;
+               vector<int> processIDS;
                
-               int count = 0;
-               for (int a=0; a<numGroups; a++) { 
-                       for (int l = a+1; l < numGroups; l++) {
-                               UniqueBL=0.0000;  //a branch length is unique if it's chidren are from the same group
-                               totalBL = 0.00; //all branch lengths
-                               UW = 0.00;              //Unweighted Value = UniqueBL / totalBL;
+               EstOutput results;
+               
+               //loop through and create all the processes you want
+               while (process != processors) {
+                       int pid = fork();
+                       
+                       if (pid > 0) {
+                               processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
+                               process++;
+                       }else if (pid == 0){
+                               EstOutput myresults;
+                               myresults = driver(t, namesOfGroupCombos, lines[process].start, lines[process].num, usingGroups, ct);
+                               
+                               if (m->control_pressed) { exit(0); }
+                               
+                               //pass numSeqs to parent
+                               ofstream out;
+                               string tempFile = outputDir + toString(getpid()) + ".unweighted.results.temp";
+                               m->openOutputFile(tempFile, out);
+                               out << myresults.size() << endl;
+                               for (int i = 0; i < myresults.size(); i++) {  out << myresults[i] << '\t';  } out << endl;
+                               out.close();
+                               
+                               exit(0);
+                       }else { 
+                               m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine(); 
+                               for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
+                               exit(0); 
+                       }
+               }
+               
+               results = driver(t, namesOfGroupCombos, lines[0].start, lines[0].num, usingGroups, ct);
+               
+               //force parent to wait until all the processes are done
+               for (int i=0;i<(processors-1);i++) { 
+                       int temp = processIDS[i];
+                       wait(&temp);
+               }
+               
+               if (m->control_pressed) { return results; }
+               
+               //get data created by processes
+               for (int i=0;i<(processors-1);i++) { 
+                       ifstream in;
+                       string s = outputDir + toString(processIDS[i]) + ".unweighted.results.temp";
+                       m->openInputFile(s, in);
+                       
+                       //get quantiles
+                       if (!in.eof()) {
+                               int num;
+                               in >> num; m->gobble(in);
+                                       
+                               if (m->control_pressed) { break; }
                                
-                               //copy random tree passed in
-                               copyTree->getCopy(t);
-                                                               
-                               //groups in this combo
-                               groups.push_back(globaldata->Groups[a]); groups.push_back(globaldata->Groups[l]);
+                               double w; 
+                               for (int j = 0; j < num; j++) {
+                                       in >> w;
+                                       
+                                       results.push_back(w);
+                               }
+                               m->gobble(in);
+                       }
+                       in.close();
+                       m->mothurRemove(s);
+               }
+               
+               return results;
+#endif         
+       }
+       catch(exception& e) {
+               m->errorOut(e, "Unweighted", "createProcesses");
+               exit(1);
+       }
+}
+/**************************************************************************************************/
+EstOutput Unweighted::driver(Tree* t, vector< vector<string> > namesOfGroupCombos, int start, int num, bool usingGroups, CountTable* ct) { 
+ try {
+               
+               EstOutput results; results.resize(num);
+               
+               int count = 0;
+               
+               Tree* copyTree = new Tree(ct);
+               
+               for (int h = start; h < (start+num); h++) {
+               
+                       if (m->control_pressed) { return results; }
+               
+                       //copy random tree passed in
+                       copyTree->getCopy(t);
                                
-                               //swap labels in the groups you want to compare
-                               copyTree->assembleRandomUnifracTree(groups[0], groups[1]);
+                       //swap labels in the groups you want to compare
+                       copyTree->assembleRandomUnifracTree(namesOfGroupCombos[h]);
+                       
+                       double UniqueBL=0.0000;  //a branch length is unique if it's chidren are from the same group
+                       double totalBL = 0.00;  //all branch lengths
+                       double UW = 0.00;               //Unweighted Value = UniqueBL / totalBL;
+                       //find a node that belongs to one of the groups in this combo
+                       int nodeBelonging = -1;
+                       for (int g = 0; g < namesOfGroupCombos[h].size(); g++) {
+                               if (copyTree->groupNodeInfo[namesOfGroupCombos[h][g]].size() != 0) { nodeBelonging = copyTree->groupNodeInfo[namesOfGroupCombos[h][g]][0]; break; }
+                       }
+                       
+                       //sanity check
+                       if (nodeBelonging == -1) {
+                               m->mothurOut("[WARNING]: cannot find a nodes in the tree from grouping "); 
+                               for (int g = 0; g < namesOfGroupCombos[h].size()-1; g++) { m->mothurOut(namesOfGroupCombos[h][g] + "-"); }
+                               m->mothurOut(namesOfGroupCombos[h][namesOfGroupCombos[h].size()-1]);
+                               m->mothurOut(", skipping."); m->mothurOutEndLine(); results[count] = UW;
+                       }else{
                                
-                               if (m->control_pressed) { delete copyTree; return data; }
+                               //if including the root this clears rootForGrouping[namesOfGroupCombos[h]]
+                               getRoot(copyTree, nodeBelonging, namesOfGroupCombos[h]);
                                
                                for(int i=0;i<copyTree->getNumNodes();i++){
-                       
+                                       
                                        if (m->control_pressed) {  return data; }
                                        
                                        //pcountSize = 0, they are from a branch that is entirely from a group the user doesn't want
@@ -275,97 +457,105 @@ EstOutput Unweighted::getValues(Tree* t, string groupA, string groupB) {
                                        //pcountSize = 1, unique to one group
                                        
                                        int pcountSize = 0;
-                                       for (int j = 0; j < groups.size(); j++) {
-                                               map<string, int>::iterator itGroup = copyTree->tree[i].pcount.find(groups[j]);
-                                               if (itGroup != copyTree->tree[i].pcount.end()) { pcountSize++; } 
+                                       for (int j = 0; j < namesOfGroupCombos[h].size(); j++) {
+                                               map<string, int>::iterator itGroup = copyTree->tree[i].pcount.find(namesOfGroupCombos[h][j]);
+                                               if (itGroup != copyTree->tree[i].pcount.end()) { pcountSize++; if (pcountSize > 1) { break; } 
                                        }
                                        
+                                       //unique calc
                                        if (pcountSize == 0) { }
-                                       else if ((copyTree->tree[i].getBranchLength() != -1) && (pcountSize == 1)) {  UniqueBL += abs(copyTree->tree[i].getBranchLength());     }
-                                               
-                                       if ((copyTree->tree[i].getBranchLength() != -1) && (pcountSize != 0)) {  
+                                       else if ((copyTree->tree[i].getBranchLength() != -1) && (pcountSize == 1) && (rootForGrouping[namesOfGroupCombos[h]].count(i) == 0)) { //you have a unique branch length and you are not the root 
+                                               UniqueBL += abs(copyTree->tree[i].getBranchLength()); 
+                                       }
+                                       
+                                       //total calc
+                                       if (pcountSize == 0) { }
+                                       else if ((copyTree->tree[i].getBranchLength() != -1) && (pcountSize != 0) && (rootForGrouping[namesOfGroupCombos[h]].count(i) == 0)) { //you have a branch length and you are not the root 
                                                totalBL += abs(copyTree->tree[i].getBranchLength()); 
                                        }
+                                       
                                }
-
-                               
+                               //cout << UniqueBL << '\t' << totalBL << endl;          
                                UW = (UniqueBL / totalBL);  
-       
+                               
                                if (isnan(UW) || isinf(UW)) { UW = 0; }
-       
-                               data[count] = UW;
-                               count++;
-                               groups.clear();
+                               
+                               results[count] = UW;
                        }
+                       count++;
+                       
                }
                
+               delete copyTree;
                
-               if (numComp != 1) {
-                       if (numGroups == 0) {
-                               //get score for all users groups
-                               for (int i = 0; i < tmap->namesOfGroups.size(); i++) {
-                                       if (tmap->namesOfGroups[i] != "xxx") {
-                                               groups.push_back(tmap->namesOfGroups[i]);
-                                       }
-                               }
-                       }else {
-                               for (int i = 0; i < globaldata->Groups.size(); i++) {
-                                       groups.push_back(globaldata->Groups[i]);
-                               }
-                       }
-               
-                       UniqueBL=0.0000;  //a branch length is unique if it's chidren are from the same group
-                       totalBL = 0.00; //all branch lengths
-                       UW = 0.00;              //Unweighted Value = UniqueBL / totalBL;
+               return results; 
+       }
+       catch(exception& e) {
+               m->errorOut(e, "Unweighted", "driver");
+               exit(1);
+       }
+}
+/**************************************************************************************************/
+int Unweighted::getRoot(Tree* t, int v, vector<string> grouping) { 
+       try {
+               //you are a leaf so get your parent
+               int index = t->tree[v].getParent();
                
-                       //copy random tree passed in
-                       copyTree->getCopy(t);
-                               
-                       //swap labels in all the groups you want to compare
-                       copyTree->assembleRandomUnifracTree(groups);
+               if (includeRoot) { 
+                       rootForGrouping[grouping].clear();
+               }else {
                        
-                       if (m->control_pressed) { delete copyTree; return data; }
-
-                       for(int i=0;i<copyTree->getNumNodes();i++){
+                       //my parent is a potential root
+                       rootForGrouping[grouping].insert(index);
                        
-                               if (m->control_pressed) {  return data; }
+                       //while you aren't at root
+                       while(t->tree[index].getParent() != -1){
+                               //cout << index << endl;        
+                               if (m->control_pressed) {  return 0; }
                                
-                               //pcountSize = 0, they are from a branch that is entirely from a group the user doesn't want
-                               //pcountSize = 2, not unique to one group
-                               //pcountSize = 1, unique to one group
+                               //am I the root for this grouping? if so I want to stop "early"
+                               //does my sibling have descendants from the users groups? 
+                               //if so I am not the root
+                               int parent = t->tree[index].getParent();
+                               int lc = t->tree[parent].getLChild();
+                               int rc = t->tree[parent].getRChild();
                                
+                               int sib = lc;
+                               if (lc == index) { sib = rc; }
+                               
+                               map<string, int>::iterator itGroup;
                                int pcountSize = 0;
-                               for (int j = 0; j < groups.size(); j++) {
-                                       map<string, int>::iterator itGroup = copyTree->tree[i].pcount.find(groups[j]);
-                                       if (itGroup != copyTree->tree[i].pcount.end()) { pcountSize++; if (pcountSize > 1) { break; } } 
+                               for (int j = 0; j < grouping.size(); j++) {
+                                       map<string, int>::iterator itGroup = t->tree[sib].pcount.find(grouping[j]);
+                                       if (itGroup != t->tree[sib].pcount.end()) { pcountSize++; if (pcountSize > 1) { break; } } 
                                }
                                
-                               if (pcountSize == 0) { }
-                               else if ((copyTree->tree[i].getBranchLength() != -1) && (pcountSize == 1)) {  UniqueBL += abs(copyTree->tree[i].getBranchLength());     }
-                                       
-                               if ((copyTree->tree[i].getBranchLength() != -1) && (pcountSize != 0)) {  
-                                       totalBL += abs(copyTree->tree[i].getBranchLength()); 
+                               //if yes, I am not the root
+                               if (pcountSize != 0) {
+                                       rootForGrouping[grouping].clear();
+                                       rootForGrouping[grouping].insert(parent);
                                }
+                               
+                               index = parent; 
+                       }
+                       
+                       //get all nodes above the root to add so we don't add their u values above
+                       index = *(rootForGrouping[grouping].begin());
+                       while(t->tree[index].getParent() != -1){
+                               int parent = t->tree[index].getParent();
+                               rootForGrouping[grouping].insert(parent);
+                               //cout << parent << " in root" << endl;
+                               index = parent;
                        }
-               
-                       UW = (UniqueBL / totalBL);  
-       
-                       if (isnan(UW) || isinf(UW)) { UW = 0; }
-       
-                       data[count] = UW;
                }
                
-               delete copyTree;
-               
-               return data;
-       
+               return 0;
        }
        catch(exception& e) {
-               m->errorOut(e, "Unweighted", "getValues");
+               m->errorOut(e, "Unweighted", "getRoot");
                exit(1);
        }
 }
-
 /**************************************************************************************************/