X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=unweighted.cpp;h=431323f066665bd07edce8da80c32dadb1600cf2;hb=2bb9267aa4b4ecdf8488b06605cc9f3f36fa4332;hp=8103688fb04d34542ee2174728d3399055a5d452;hpb=4745a956b3116a719f52f341d2a2db84df4817da;p=mothur.git diff --git a/unweighted.cpp b/unweighted.cpp index 8103688..431323f 100644 --- a/unweighted.cpp +++ b/unweighted.cpp @@ -13,12 +13,11 @@ EstOutput Unweighted::getValues(Tree* t, int p, string o) { try { - globaldata = GlobalData::getInstance(); processors = p; outputDir = o; //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 +25,7 @@ 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); } } @@ -35,15 +34,15 @@ 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 < (tmap->getNamesOfGroups()).size(); i++) { + if ((tmap->getNamesOfGroups())[i] != "xxx") { + groups.push_back((tmap->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); } @@ -66,7 +65,6 @@ EstOutput Unweighted::getValues(Tree* t, int p, string o) { lines.push_back(linePair(startPos, numPairsPerProcessor)); } - data = createProcesses(t, namesOfGroupCombos); lines.clear(); } @@ -87,7 +85,6 @@ EstOutput Unweighted::createProcesses(Tree* t, vector< vector > namesOfG try { #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) int process = 1; - int num = 0; vector processIDS; EstOutput results; @@ -116,7 +113,11 @@ EstOutput Unweighted::createProcesses(Tree* t, vector< vector > 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); @@ -150,7 +151,7 @@ EstOutput Unweighted::createProcesses(Tree* t, vector< vector > namesOfG m->gobble(in); } in.close(); - remove(s.c_str()); + m->mothurRemove(s); } m->mothurOut("DONE."); m->mothurOutEndLine(); m->mothurOutEndLine(); @@ -166,57 +167,83 @@ EstOutput Unweighted::createProcesses(Tree* t, vector< vector > namesOfG /**************************************************************************************************/ EstOutput Unweighted::driver(Tree* t, vector< vector > namesOfGroupCombos, int start, int num) { 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;igetNumNodes();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::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;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 + + int pcountSize = 0; + for (int j = 0; j < namesOfGroupCombos[h].size(); j++) { + map::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()); + } } - } + //cout << UniqueBL << '\t' << totalBL << endl; + UW = (UniqueBL / totalBL); - 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) { float tempOut = (count / (float)total); if (isnan(tempOut) || isinf(tempOut)) { tempOut = 0.0; } m->mothurOut("Percentage complete: " + toString((int(tempOut) * 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(); } } //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(); } + //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; } @@ -229,13 +256,11 @@ 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; - //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; @@ -243,7 +268,7 @@ 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); } } @@ -252,15 +277,15 @@ 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 < (tmap->getNamesOfGroups()).size(); i++) { + if ((tmap->getNamesOfGroups())[i] != "xxx") { + groups.push_back((tmap->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); } @@ -281,7 +306,7 @@ EstOutput Unweighted::getValues(Tree* t, string groupA, string groupB, int p, st } lines.push_back(linePair(startPos, numPairsPerProcessor)); } - + data = createProcesses(t, namesOfGroupCombos, true); lines.clear(); @@ -303,7 +328,6 @@ EstOutput Unweighted::createProcesses(Tree* t, vector< vector > namesOfG try { #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) int process = 1; - int num = 0; vector processIDS; EstOutput results; @@ -330,7 +354,11 @@ EstOutput Unweighted::createProcesses(Tree* t, vector< vector > 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, usingGroups); @@ -353,18 +381,19 @@ EstOutput Unweighted::createProcesses(Tree* t, vector< vector > namesOfG 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(); - remove(s.c_str()); + m->mothurRemove(s); } return results; @@ -382,10 +411,8 @@ EstOutput Unweighted::driver(Tree* t, vector< vector > namesOfGroupCombo EstOutput results; results.resize(num); int count = 0; - int total = num; - int twentyPercent = (total * 0.20); - Tree* copyTree = new Tree; + Tree* copyTree = new Tree(tmap); for (int h = start; h < (start+num); h++) { @@ -400,36 +427,59 @@ EstOutput Unweighted::driver(Tree* t, vector< vector > namesOfGroupCombo 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;igetNumNodes();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 (copyTree->groupNodeInfo[namesOfGroupCombos[h][g]].size() != 0) { nodeBelonging = copyTree->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 + //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{ - int pcountSize = 0; - for (int j = 0; j < namesOfGroupCombos[h].size(); j++) { - map::iterator itGroup = t->tree[i].pcount.find(namesOfGroupCombos[h][j]); - if (itGroup != t->tree[i].pcount.end()) { pcountSize++; if (pcountSize > 1) { break; } } - } + //if including the root this clears rootForGrouping[namesOfGroupCombos[h]] + getRoot(copyTree, nodeBelonging, namesOfGroupCombos[h]); - if (pcountSize == 0) { } - else if ((t->tree[i].getBranchLength() != -1) && (pcountSize == 1)) { UniqueBL += abs(t->tree[i].getBranchLength()); } + for(int i=0;igetNumNodes();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::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()); + } - if ((t->tree[i].getBranchLength() != -1) && (pcountSize != 0)) { - totalBL += abs(t->tree[i].getBranchLength()); } + //cout << UniqueBL << '\t' << totalBL << endl; + UW = (UniqueBL / totalBL); + + if (isnan(UW) || isinf(UW)) { UW = 0; } + + results[count] = UW; } - - UW = (UniqueBL / totalBL); - - if (isnan(UW) || isinf(UW)) { UW = 0; } - - results[count] = UW; count++; - + } delete copyTree; @@ -442,5 +492,66 @@ 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[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" + //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; + } + + //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; + } + catch(exception& e) { + m->errorOut(e, "Unweighted", "getRoot"); + exit(1); + } +} +/**************************************************************************************************/