/**************************************************************************************************/
-EstOutput Unweighted::getValues(Tree* t) {
+EstOutput Unweighted::getValues(Tree* t, int p, string o) {
try {
- globaldata = GlobalData::getInstance();
-
- 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;
- map<string, int>::iterator it; //iterator to traverse pgroups
- map<string, int> copyIpcount;
-
-
+ 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((m->getGroups())[r]); groups.push_back((m->getGroups())[l]);
+ namesOfGroupCombos.push_back(groups);
}
}
-
- //numComp+1 for AB, AC, BC, ABC
- data.resize(numComp+1,0);
-
- 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;
- copyIpcount.clear();
-
- //groups in this combo
- groups.push_back(globaldata->Groups[a]); groups.push_back(globaldata->Groups[l]);
-
- for(int i=t->getNumLeaves();i<t->getNumNodes();i++){
- int lc = t->tree[i].getLChild(); //lc = vector index of left child
- int rc = t->tree[i].getRChild(); //rc = vector index of right child
-
- /**********************************************************************/
- //This section adds in all lengths that are non leaf
-
- copyIpcount = t->tree[i].pcount;
- for (it = copyIpcount.begin(); it != copyIpcount.end(); it++) {
- if (inUsersGroups(it->first, groups) != true) { copyIpcount.erase(it->first); }
- }
-
- //if i's children are from the same group then i's pcount size will be 1
- //if copyIpcount.size() = 0 they are from a branch that is entirely from a group the user doesn't want
- if (copyIpcount.size() == 0) { }
- else if ((t->tree[i].getBranchLength() != -1) && (copyIpcount.size() == 1)) { UniqueBL += t->tree[i].getBranchLength(); }
-
- //add i's BL to total if it is from the groups the user wants
- if ((t->tree[i].getBranchLength() != -1) && (copyIpcount.size() != 0)) {
- totalBL += t->tree[i].getBranchLength();
- }
-
- /**********************************************************************/
- //This section adds in all lengths that are leaf
-
- //if i's chidren are leaves
- if (t->tree[rc].getRChild() == -1) {
- //if rc is a valid group and rc has a BL
- if ((inUsersGroups(t->tree[rc].getGroup(), groups) == true) && (t->tree[rc].getBranchLength() != -1)) {
- UniqueBL += t->tree[rc].getBranchLength();
- totalBL += t->tree[rc].getBranchLength();
- }
- }
-
- if (t->tree[lc].getLChild() == -1) {
- //if lc is a valid group and lc has a BL
- if ((inUsersGroups(t->tree[lc].getGroup(), groups) == true) && (t->tree[lc].getBranchLength() != -1)) {
- UniqueBL += t->tree[lc].getBranchLength();
- totalBL += t->tree[lc].getBranchLength();
- }
+ 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);
+ }
+ }
+
+ 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);
+
+ lines.clear();
+
+ return data;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "Unweighted", "getValues");
+ exit(1);
+ }
+}
+/**************************************************************************************************/
+
+EstOutput Unweighted::createProcesses(Tree* t, vector< vector<string> > namesOfGroupCombos, CountTable* ct) {
+ try {
+ int process = 1;
+ vector<int> processIDS;
- UW = (UniqueBL / totalBL);
-
- if (isnan(UW) || isinf(UW)) { UW = 0; }
-
- data[count] = UW;
- count++;
- groups.clear();
+ 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();
+
+ 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, ct);
+
+ if (m->control_pressed) { exit(0); }
+
+ //m->mothurOut("Merging results."); m->mothurOutEndLine();
+
+ //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, ct);
- if (numComp != 1) {
- if (numGroups == 0) {
- //get score for all users groups
- for (int i = 0; i < tmap->namesOfGroups.size(); i++) {
- groups.push_back(tmap->namesOfGroups[i]);
- }
- }else {
- for (int i = 0; i < globaldata->Groups.size(); i++) {
- groups.push_back(globaldata->Groups[i]);
+ //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; }
+
+ double w;
+ for (int j = 0; j < num; j++) {
+ in >> w;
+ results.push_back(w);
}
+ m->gobble(in);
}
+ in.close();
+ 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]);
+ }
- 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;
- copyIpcount.clear();
+ results = driver(t, namesOfGroupCombos, lines[0].start, lines[0].num, 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];
+ }
+
+#endif
+ return results;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "Unweighted", "createProcesses");
+ exit(1);
+ }
+}
+/**************************************************************************************************/
+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;
+
+ for (int h = start; h < (start+num); h++) {
- for(int i=t->getNumLeaves();i<t->getNumNodes();i++){
+ if (m->control_pressed) { return results; }
- int lc = t->tree[i].getLChild(); //lc = vector index of left child
- int rc = t->tree[i].getRChild(); //rc = vector index of right child
-
- /**********************************************************************/
- //This section adds in all lengths that are non leaf
-
- copyIpcount = t->tree[i].pcount;
- for (it = copyIpcount.begin(); it != copyIpcount.end(); it++) {
- if (inUsersGroups(it->first, groups) != true) { copyIpcount.erase(it->first); }
- }
-
- //if i's children are from the same group then i's pcount size will be 1
- //if copyIpcount.size() = 0 they are from a branch that is entirely from a group the user doesn't want
- if (copyIpcount.size() == 0) { }
- else if ((t->tree[i].getBranchLength() != -1) && (copyIpcount.size() == 1)) { UniqueBL += t->tree[i].getBranchLength(); }
-
- //add i's BL to total if it is from the groups the user wants
- if ((t->tree[i].getBranchLength() != -1) && (copyIpcount.size() != 0)) {
- totalBL += t->tree[i].getBranchLength();
- }
-
- /**********************************************************************/
- //This section adds in all lengths that are leaf
+ 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 (t->groupNodeInfo[namesOfGroupCombos[h][g]].size() != 0) { nodeBelonging = t->groupNodeInfo[namesOfGroupCombos[h][g]][0]; break; }
+ }
- //if i's chidren are leaves
- if (t->tree[rc].getRChild() == -1) {
- //if rc is a valid group and rc has a BL
- if ((inUsersGroups(t->tree[rc].getGroup(), groups) == true) && (t->tree[rc].getBranchLength() != -1)) {
- UniqueBL += t->tree[rc].getBranchLength();
- totalBL += t->tree[rc].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[lc].getLChild() == -1) {
- //if lc is a valid group and lc has a BL
- if ((inUsersGroups(t->tree[lc].getGroup(), groups) == true) && (t->tree[lc].getBranchLength() != -1)) {
- UniqueBL += t->tree[lc].getBranchLength();
- totalBL += t->tree[lc].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; }
- data[count] = UW;
- }
+ results[count] = UW;
+ }
+ count++;
- return data;
-
+ }
+
+ return results;
}
catch(exception& e) {
- cout << "Standard Error: " << e.what() << " has occurred in the Unweighted class Function getValues. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
- exit(1);
- }
- catch(...) {
- cout << "An unknown error has occurred in the Unweighted class function getValues. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+ m->errorOut(e, "Unweighted", "driver");
exit(1);
}
-
}
-
/**************************************************************************************************/
-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();
-
- 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;
- map<string, int>::iterator it; //iterator to traverse pgroups
- map<string, int> copyIpcount;
- copyTree = new Tree;
-
+ 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((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);
}
}
+
+ 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();
+
+ 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 {
+ 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;
- copyIpcount.clear();
+ 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();
+
+ 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);
- //copy random tree passed in
- copyTree->getCopy(t);
-
- //groups in this combo
- groups.push_back(globaldata->Groups[a]); groups.push_back(globaldata->Groups[l]);
+ if (m->control_pressed) { exit(0); }
- //swap labels in the groups you want to compare
- copyTree->assembleRandomUnifracTree(groups[0], groups[1]);
+ //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();
- //copyTree->createNewickFile("random"+groupA+toString(count));
-
- for(int i=copyTree->getNumLeaves();i<copyTree->getNumNodes();i++){
+ 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);
+ }
+ }
- int lc = copyTree->tree[i].getLChild(); //lc = vector index of left child
- int rc = copyTree->tree[i].getRChild(); //rc = vector index of right child
-
- /**********************************************************************/
- //This section adds in all lengths that are non leaf
-
- copyIpcount = copyTree->tree[i].pcount;
- for (it = copyIpcount.begin(); it != copyIpcount.end(); it++) {
- if (inUsersGroups(it->first, groups) != true) { copyIpcount.erase(it->first); }
- }
-
- //if i's children are from the same group then i's pcount size will be 1
- //if copyIpcount.size() = 0 they are from a branch that is entirely from a group the user doesn't want
- if (copyIpcount.size() == 0) { }
- else if ((copyTree->tree[i].getBranchLength() != -1) && (copyIpcount.size() == 1)) { UniqueBL += copyTree->tree[i].getBranchLength(); }
-
- //add i's BL to total if it is from the groups the user wants
- if ((copyTree->tree[i].getBranchLength() != -1) && (copyIpcount.size() != 0)) {
- totalBL += copyTree->tree[i].getBranchLength();
- }
-
- /**********************************************************************/
- //This section adds in all lengths that are leaf
-
- //if i's chidren are leaves
- if (copyTree->tree[rc].getRChild() == -1) {
- //if rc is a valid group and rc has a BL
- if ((inUsersGroups(copyTree->tree[rc].getGroup(), groups) == true) && (copyTree->tree[rc].getBranchLength() != -1)) {
- UniqueBL += copyTree->tree[rc].getBranchLength();
- totalBL += copyTree->tree[rc].getBranchLength();
- }
- }
-
- if (copyTree->tree[lc].getLChild() == -1) {
- //if lc is a valid group and lc has a BL
- if ((inUsersGroups(copyTree->tree[lc].getGroup(), groups) == true) && (copyTree->tree[lc].getBranchLength() != -1)) {
- UniqueBL += copyTree->tree[lc].getBranchLength();
- totalBL += copyTree->tree[lc].getBranchLength();
- }
- }
-
- /**********************************************************************/
- }
+ results = driver(t, namesOfGroupCombos, lines[0].start, lines[0].num, usingGroups, ct);
- UW = (UniqueBL / totalBL);
-
- if (isnan(UW) || isinf(UW)) { UW = 0; }
-
- data[count] = UW;
- count++;
- groups.clear();
- }
+ //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; }
- if (numComp != 1) {
- if (numGroups == 0) {
- //get score for all users groups
- for (int i = 0; i < tmap->namesOfGroups.size(); i++) {
- groups.push_back(tmap->namesOfGroups[i]);
- }
- }else {
- for (int i = 0; i < globaldata->Groups.size(); i++) {
- groups.push_back(globaldata->Groups[i]);
+ //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; }
+
+ double w;
+ for (int j = 0; j < num; j++) {
+ in >> w;
+
+ results.push_back(w);
}
+ m->gobble(in);
}
+ in.close();
+ 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]);
+ }
- 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;
- copyIpcount.clear();
+ 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");
+ 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 all the groups you want to compare
- copyTree->assembleRandomUnifracTree(groups);
-
- for(int i=copyTree->getNumLeaves();i<copyTree->getNumNodes();i++){
-
- int lc = copyTree->tree[i].getLChild(); //lc = vector index of left child
- int rc = copyTree->tree[i].getRChild(); //rc = vector index of right child
-
- /**********************************************************************/
- //This section adds in all lengths that are non leaf
+ //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; }
+ }
- copyIpcount = copyTree->tree[i].pcount;
- for (it = copyIpcount.begin(); it != copyIpcount.end(); it++) {
- if (inUsersGroups(it->first, groups) != true) { copyIpcount.erase(it->first); }
+ //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 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
+ //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 = 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) && (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; }
+
+ results[count] = UW;
+ }
+ count++;
- //if i's children are from the same group then i's pcount size will be 1
- //if copyIpcount.size() = 0 they are from a branch that is entirely from a group the user doesn't want
- if (copyIpcount.size() == 0) { }
- else if ((copyTree->tree[i].getBranchLength() != -1) && (copyIpcount.size() == 1)) { UniqueBL += copyTree->tree[i].getBranchLength(); }
-
- //add i's BL to total if it is from the groups the user wants
- if ((copyTree->tree[i].getBranchLength() != -1) && (copyIpcount.size() != 0)) {
- totalBL += copyTree->tree[i].getBranchLength();
- }
+ }
+
+ delete copyTree;
+
+ 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();
+
+ if (includeRoot) {
+ rootForGrouping[grouping].clear();
+ }else {
- /**********************************************************************/
- //This section adds in all lengths that are leaf
+ //my parent is a potential root
+ rootForGrouping[grouping].insert(index);
- //if i's chidren are leaves
- if (copyTree->tree[rc].getRChild() == -1) {
- //if rc is a valid group and rc has a BL
- if ((inUsersGroups(copyTree->tree[rc].getGroup(), groups) == true) && (copyTree->tree[rc].getBranchLength() != -1)) {
- UniqueBL += copyTree->tree[rc].getBranchLength();
- totalBL += copyTree->tree[rc].getBranchLength();
- }
+ //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"
+ //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 < 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 (copyTree->tree[lc].getLChild() == -1) {
- //if lc is a valid group and lc has a BL
- if ((inUsersGroups(copyTree->tree[lc].getGroup(), groups) == true) && (copyTree->tree[lc].getBranchLength() != -1)) {
- UniqueBL += copyTree->tree[lc].getBranchLength();
- totalBL += copyTree->tree[lc].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;
}
-
- return data;
-
+
+ return 0;
}
catch(exception& e) {
- cout << "Standard Error: " << e.what() << " has occurred in the Unweighted class Function getValues. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
- exit(1);
- }
- catch(...) {
- cout << "An unknown error has occurred in the Unweighted class function getValues. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+ m->errorOut(e, "Unweighted", "getRoot");
exit(1);
}
}
-
-
+/**************************************************************************************************/
projects / mothur.git / blobdiff