X-Git-Url: https://git.donarmstrong.com/?p=mothur.git;a=blobdiff_plain;f=unweighted.cpp;h=a845f9bc2fa54a5c27b0d74aed20929f1aeb756f;hp=312d549b11a1a4a7a4e7960d5a0689885e9a52d2;hb=cf9987b67aa49777a4c91c2d21f96e58bf17aa82;hpb=edad2fa8defaa6856b3606a215bf64b91340eeb4 diff --git a/unweighted.cpp b/unweighted.cpp index 312d549..a845f9b 100644 --- a/unweighted.cpp +++ b/unweighted.cpp @@ -13,12 +13,13 @@ 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 groups; groups.push_back(globaldata->Groups[r]); groups.push_back(globaldata->Groups[l]); + vector 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 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 > namesOfGroupCombos) { +EstOutput Unweighted::createProcesses(Tree* t, vector< vector > namesOfGroupCombos, CountTable* ct) { try { -#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) - int process = 1; + int process = 1; vector 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 > 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 > 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 > namesOfG m->gobble(in); } in.close(); - remove(s.c_str()); + m->mothurRemove(s); + } +#else + //fill in functions + vector pDataArray; + DWORD dwThreadIdArray[processors-1]; + HANDLE hThreadArray[processors-1]; + vector cts; + vector trees; + + //Create processor worker threads. + for( int i=1; icopy(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); - return results; -#endif + //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"); @@ -166,19 +194,17 @@ EstOutput Unweighted::createProcesses(Tree* t, vector< vector > namesOfG } } /**************************************************************************************************/ -EstOutput Unweighted::driver(Tree* t, vector< vector > namesOfGroupCombos, int start, int num) { +EstOutput Unweighted::driver(Tree* t, vector< vector > 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++) { - //cout << namesOfGroupCombos[h][0] << '\t' << namesOfGroupCombos[h][1] << endl; + if (m->control_pressed) { return results; } double UniqueBL=0.0000; //a branch length is unique if it's chidren are from the same group @@ -198,13 +224,14 @@ EstOutput Unweighted::driver(Tree* t, vector< vector > 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;igetNumNodes();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 @@ -214,21 +241,21 @@ EstOutput Unweighted::driver(Tree* t, vector< vector > namesOfGroupCombo map::iterator itGroup = t->tree[i].pcount.find(namesOfGroupCombos[h][j]); if (itGroup != t->tree[i].pcount.end()) { pcountSize++; if (pcountSize > 1) { break; } } } - // + + //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()); } - } - cout << UniqueBL << '\t' << totalBL << endl; + //cout << UniqueBL << '\t' << totalBL << endl; UW = (UniqueBL / totalBL); if (isnan(UW) || isinf(UW)) { UW = 0; } @@ -237,12 +264,7 @@ EstOutput Unweighted::driver(Tree* t, vector< vector > 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; } @@ -255,13 +277,13 @@ EstOutput Unweighted::driver(Tree* t, vector< vector > 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; @@ -269,7 +291,7 @@ EstOutput Unweighted::getValues(Tree* t, string groupA, string groupB, int p, st for (int r=0; r groups; groups.push_back(globaldata->Groups[r]); groups.push_back(globaldata->Groups[l]); + vector groups; groups.push_back((m->getGroups())[r]); groups.push_back((m->getGroups())[l]); namesOfGroupCombos.push_back(groups); } } @@ -278,44 +300,33 @@ EstOutput Unweighted::getValues(Tree* t, string groupA, string groupB, int p, st vector 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) { @@ -325,14 +336,14 @@ EstOutput Unweighted::getValues(Tree* t, string groupA, string groupB, int p, st } /**************************************************************************************************/ -EstOutput Unweighted::createProcesses(Tree* t, vector< vector > namesOfGroupCombos, bool usingGroups) { +EstOutput Unweighted::createProcesses(Tree* t, vector< vector > namesOfGroupCombos, bool usingGroups, CountTable* ct) { try { -#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) - int process = 1; + int process = 1; vector 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(); @@ -342,7 +353,7 @@ EstOutput Unweighted::createProcesses(Tree* t, vector< vector > 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); } @@ -362,7 +373,7 @@ EstOutput Unweighted::createProcesses(Tree* t, vector< vector > 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++) { @@ -394,11 +405,53 @@ EstOutput Unweighted::createProcesses(Tree* t, vector< vector > 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 pDataArray; + DWORD dwThreadIdArray[processors-1]; + HANDLE hThreadArray[processors-1]; + vector cts; + vector trees; + + //Create processor worker threads. + for( int i=1; icopy(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"); @@ -406,14 +459,14 @@ EstOutput Unweighted::createProcesses(Tree* t, vector< vector > namesOfG } } /**************************************************************************************************/ -EstOutput Unweighted::driver(Tree* t, vector< vector > namesOfGroupCombos, int start, int num, bool usingGroups) { +EstOutput Unweighted::driver(Tree* t, vector< vector > 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++) { @@ -442,6 +495,7 @@ EstOutput Unweighted::driver(Tree* t, vector< vector > namesOfGroupCombo 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;igetNumNodes();i++){ @@ -477,7 +531,7 @@ EstOutput Unweighted::driver(Tree* t, vector< vector > namesOfGroupCombo if (isnan(UW) || isinf(UW)) { UW = 0; } results[count] = UW; - } + } count++; } @@ -495,49 +549,54 @@ EstOutput Unweighted::driver(Tree* t, vector< vector > namesOfGroupCombo int Unweighted::getRoot(Tree* t, int v, vector grouping) { try { //you are a leaf so get your parent - int index = t->tree[index].getParent(); - - //my parent is a potential root - rootForGrouping[grouping].insert(index); + int index = t->tree[v].getParent(); - //while you aren't at root - while(t->tree[index].getParent() != -1){ - - 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(); + if (includeRoot) { + rootForGrouping[grouping].clear(); + }else { - int sib = lc; - if (lc == index) { sib = rc; } + //my parent is a potential root + rootForGrouping[grouping].insert(index); - map::iterator itGroup; - int pcountSize = 0; - for (int j = 0; j < grouping.size(); j++) { - map::iterator itGroup = t->tree[sib].pcount.find(grouping[j]); - if (itGroup != t->tree[sib].pcount.end()) { pcountSize++; if (pcountSize > 1) { break; } } + //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::iterator itGroup; + int pcountSize = 0; + for (int j = 0; j < grouping.size(); j++) { + map::iterator itGroup = t->tree[sib].pcount.find(grouping[j]); + if (itGroup != t->tree[sib].pcount.end()) { pcountSize++; if (pcountSize > 1) { break; } } + } + + //if yes, I am not the root + if (pcountSize != 0) { + rootForGrouping[grouping].clear(); + rootForGrouping[grouping].insert(parent); + } + + index = parent; } - //if yes, I am not the root - if (pcountSize != 0) { - rootForGrouping[grouping].clear(); + //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; } - - 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; } return 0;