]> git.donarmstrong.com Git - mothur.git/blobdiff - unweighted.cpp
working on pam
[mothur.git] / unweighted.cpp
index 08e83ec5ba4ccbbbefe36163c25d1075fb05a883..a845f9bc2fa54a5c27b0d74aed20929f1aeb756f 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;
@@ -26,7 +27,7 @@ EstOutput Unweighted::getValues(Tree* t, int p, string o) {
                for (int r=0; r<numGroups; r++) { 
                        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);
                        }
                }
@@ -35,44 +36,34 @@ 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);
                        }
                }
+        
+        lines.clear();
+        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, ct);
 
-               #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
-                       if(processors == 1){
-                               data = driver(t, namesOfGroupCombos, 0, namesOfGroupCombos.size());
-                       }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);
-                               lines.clear();
-                       }
-               #else
-                       data = driver(t, namesOfGroupCombos, 0, namesOfGroupCombos.size());
-               #endif
-               
+        lines.clear();
+        
                return data;
        }
        catch(exception& e) {
@@ -82,13 +73,14 @@ 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)
-               int process = 1;
+        int process = 1;
                vector<int> processIDS;
                
                EstOutput results;
+#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
+               
                
                //loop through and create all the processes you want
                while (process != processors) {
@@ -99,11 +91,11 @@ 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;
@@ -121,7 +113,7 @@ EstOutput Unweighted::createProcesses(Tree* t, vector< vector<string> > namesOfG
                        }
                }
                
-               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++) { 
@@ -152,13 +144,49 @@ EstOutput Unweighted::createProcesses(Tree* t, vector< vector<string> > namesOfG
                                m->gobble(in);
                        }
                        in.close();
-                       remove(s.c_str());
+                       m->mothurRemove(s);
+               }
+#else
+               //fill in functions
+        vector<unweightedData*> pDataArray;
+               DWORD   dwThreadIdArray[processors-1];
+               HANDLE  hThreadArray[processors-1];
+        vector<CountTable*> cts;
+        vector<Tree*> trees;
+               
+               //Create processor worker threads.
+               for( int i=1; i<processors; i++ ){
+            CountTable* copyCount = new CountTable();
+            copyCount->copy(ct);
+            Tree* copyTree = new Tree(copyCount);
+            copyTree->getCopy(t);
+            
+            cts.push_back(copyCount);
+            trees.push_back(copyTree);
+            
+            unweightedData* tempweighted = new unweightedData(m, lines[i].start, lines[i].num, namesOfGroupCombos, copyTree, copyCount, includeRoot);
+                       pDataArray.push_back(tempweighted);
+                       processIDS.push_back(i);
+            
+                       hThreadArray[i-1] = CreateThread(NULL, 0, MyUnWeightedThreadFunction, pDataArray[i-1], 0, &dwThreadIdArray[i-1]);
                }
                
-               m->mothurOut("DONE."); m->mothurOutEndLine(); m->mothurOutEndLine();
+               results = driver(t, namesOfGroupCombos, lines[0].start, lines[0].num, ct);
+               
+               //Wait until all threads have terminated.
+               WaitForMultipleObjects(processors-1, hThreadArray, TRUE, INFINITE);
                
-               return results;
-#endif         
+               //Close all thread handles and free memory allocations.
+               for(int i=0; i < pDataArray.size(); i++){
+            for (int j = 0; j < pDataArray[i]->results.size(); j++) {  results.push_back(pDataArray[i]->results[j]);  }
+                       delete cts[i];
+            delete trees[i];
+                       CloseHandle(hThreadArray[i]);
+                       delete pDataArray[i];
+               }
+
+#endif 
+        return results;
        }
        catch(exception& e) {
                m->errorOut(e, "Unweighted", "createProcesses");
@@ -166,7 +194,7 @@ 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 {
        
         
@@ -174,10 +202,7 @@ EstOutput Unweighted::driver(Tree* t, vector< vector<string> > namesOfGroupCombo
                
                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; }
@@ -199,14 +224,14 @@ EstOutput Unweighted::driver(Tree* t, vector< vector<string> > namesOfGroupCombo
                                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
@@ -239,12 +264,7 @@ EstOutput Unweighted::driver(Tree* t, vector< vector<string> > namesOfGroupCombo
                        }
                        count++;
 
-                       //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();       }
-               }
-               
-               //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; 
        }
@@ -257,12 +277,13 @@ EstOutput Unweighted::driver(Tree* t, vector< vector<string> > namesOfGroupCombo
 
 EstOutput Unweighted::getValues(Tree* t, string groupA, string groupB, 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;
@@ -270,7 +291,7 @@ EstOutput Unweighted::getValues(Tree* t, string groupA, string groupB, int p, st
                for (int r=0; r<numGroups; r++) { 
                        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);
                        }
                }
@@ -279,44 +300,33 @@ EstOutput Unweighted::getValues(Tree* t, string groupA, string groupB, int p, st
                        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);
                        }
                }
+     
+        lines.clear();
+        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();
 
-               #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
-                       if(processors == 1){
-                               data = driver(t, namesOfGroupCombos, 0, namesOfGroupCombos.size(), true);
-                       }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);
-                               
-                               lines.clear();
-                       }
-               #else
-                       data = driver(t, namesOfGroupCombos, 0, namesOfGroupCombos.size(), true);
-               #endif
-       
                return data;
        }
        catch(exception& e) {
@@ -326,14 +336,14 @@ EstOutput Unweighted::getValues(Tree* t, string groupA, string groupB, int p, st
 }
 /**************************************************************************************************/
 
-EstOutput Unweighted::createProcesses(Tree* t, vector< vector<string> > namesOfGroupCombos, bool usingGroups) {
+EstOutput Unweighted::createProcesses(Tree* t, vector< vector<string> > namesOfGroupCombos, bool usingGroups, CountTable* ct) {
        try {
-#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
-               int process = 1;
+        int process = 1;
                vector<int> processIDS;
                
                EstOutput results;
-               
+#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
+
                //loop through and create all the processes you want
                while (process != processors) {
                        int pid = fork();
@@ -343,7 +353,7 @@ 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, usingGroups);
+                               myresults = driver(t, namesOfGroupCombos, lines[process].start, lines[process].num, usingGroups, ct);
                                
                                if (m->control_pressed) { exit(0); }
                                
@@ -363,7 +373,7 @@ EstOutput Unweighted::createProcesses(Tree* t, vector< vector<string> > namesOfG
                        }
                }
                
-               results = driver(t, namesOfGroupCombos, lines[0].start, lines[0].num, usingGroups);
+               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++) { 
@@ -395,11 +405,53 @@ EstOutput Unweighted::createProcesses(Tree* t, vector< vector<string> > namesOfG
                                m->gobble(in);
                        }
                        in.close();
-                       remove(s.c_str());
+                       m->mothurRemove(s);
+               }
+#else
+       //for some reason it doesn't seem to be calculating hte random trees scores.  all scores are the same even though copytree appears to be randomized.
+        
+        /*
+        //fill in functions
+        vector<unweightedData*> pDataArray;
+               DWORD   dwThreadIdArray[processors-1];
+               HANDLE  hThreadArray[processors-1];
+        vector<CountTable*> cts;
+        vector<Tree*> trees;
+               
+               //Create processor worker threads.
+               for( int i=1; i<processors; i++ ){
+            CountTable* copyCount = new CountTable();
+            copyCount->copy(ct);
+            Tree* copyTree = new Tree(copyCount);
+            copyTree->getCopy(t);
+            
+            cts.push_back(copyCount);
+            trees.push_back(copyTree);
+            
+            unweightedData* tempweighted = new unweightedData(m, lines[i].start, lines[i].num, namesOfGroupCombos, copyTree, copyCount, includeRoot);
+                       pDataArray.push_back(tempweighted);
+                       processIDS.push_back(i);
+            
+                       hThreadArray[i-1] = CreateThread(NULL, 0, MyUnWeightedRandomThreadFunction, pDataArray[i-1], 0, &dwThreadIdArray[i-1]);
                }
                
-               return results;
-#endif         
+               results = driver(t, namesOfGroupCombos, lines[0].start, lines[0].num, usingGroups, ct);
+               
+               //Wait until all threads have terminated.
+               WaitForMultipleObjects(processors-1, hThreadArray, TRUE, INFINITE);
+               
+               //Close all thread handles and free memory allocations.
+               for(int i=0; i < pDataArray.size(); i++){
+            for (int j = 0; j < pDataArray[i]->results.size(); j++) {  results.push_back(pDataArray[i]->results[j]);  }
+                       delete cts[i];
+            delete trees[i];
+                       CloseHandle(hThreadArray[i]);
+                       delete pDataArray[i];
+               }       */
+        
+        results = driver(t, namesOfGroupCombos, 0, namesOfGroupCombos.size(), usingGroups, ct);
+#endif
+        return results;
        }
        catch(exception& e) {
                m->errorOut(e, "Unweighted", "createProcesses");
@@ -407,14 +459,14 @@ EstOutput Unweighted::createProcesses(Tree* t, vector< vector<string> > namesOfG
        }
 }
 /**************************************************************************************************/
-EstOutput Unweighted::driver(Tree* t, vector< vector<string> > namesOfGroupCombos, int start, int num, bool usingGroups) { 
+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;
+               Tree* copyTree = new Tree(ct);
                
                for (int h = start; h < (start+num); h++) {
                
@@ -479,7 +531,7 @@ EstOutput Unweighted::driver(Tree* t, vector< vector<string> > namesOfGroupCombo
                                if (isnan(UW) || isinf(UW)) { UW = 0; }
                                
                                results[count] = UW;
-                       }
+            }
                        count++;
                        
                }
@@ -497,17 +549,18 @@ EstOutput Unweighted::driver(Tree* t, vector< vector<string> > namesOfGroupCombo
 int Unweighted::getRoot(Tree* t, int v, vector<string> grouping) { 
        try {
                //you are a leaf so get your parent
-               int index = t->tree[index].getParent();
+               int index = t->tree[v].getParent();
                
                if (includeRoot) { 
                        rootForGrouping[grouping].clear();
                }else {
+                       
                        //my parent is a potential root
                        rootForGrouping[grouping].insert(index);
                        
                        //while you aren't at root
                        while(t->tree[index].getParent() != -1){
-                               
+                               //cout << index << endl;        
                                if (m->control_pressed) {  return 0; }
                                
                                //am I the root for this grouping? if so I want to stop "early"