X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=unweighted.cpp;h=08e83ec5ba4ccbbbefe36163c25d1075fb05a883;hb=fca3f55d5ded10c3dc77856f3cc4a1c53b02bb6f;hp=0ac7147d7d8b0b288789267c18c43e1867b43e93;hpb=59c2827ee568dcc9293faef40e5d67499bc68bc3;p=mothur.git diff --git a/unweighted.cpp b/unweighted.cpp index 0ac7147..08e83ec 100644 --- a/unweighted.cpp +++ b/unweighted.cpp @@ -66,7 +66,6 @@ EstOutput Unweighted::getValues(Tree* t, int p, string o) { lines.push_back(linePair(startPos, numPairsPerProcessor)); } - data = createProcesses(t, namesOfGroupCombos); lines.clear(); } @@ -87,7 +86,6 @@ 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 +114,11 @@ 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); @@ -166,90 +168,83 @@ 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 numLeaves = t->getNumLeaves(); 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 double totalBL = 0.00; //all branch lengths double UW = 0.00; //Unweighted Value = UniqueBL / totalBL; - map tempTotals; //maps node to total Branch Length - map nodePcountSize; //maps node to pcountSize - map::iterator itCount; - - 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; } } - } - - nodePcountSize[i] = pcountSize; + //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 << i << '\t' << t->tree[i].getName() << " br = " << abs(t->tree[i].getBranchLength()) << '\t'; - if (pcountSize == 0) { } - else if ((t->tree[i].getBranchLength() != -1) && (pcountSize == 1)) { UniqueBL += abs(t->tree[i].getBranchLength()); } + //if including the root this clears rootForGrouping[namesOfGroupCombos[h]] + getRoot(t, nodeBelonging, namesOfGroupCombos[h]); - //if you are a leaf from a users group add to total - if (i < numLeaves) { - if ((t->tree[i].getBranchLength() != -1) && pcountSize != 0) { - //cout << "added to total" << endl; - totalBL += 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 = t->tree[i].pcount.find(namesOfGroupCombos[h][j]); + if (itGroup != t->tree[i].pcount.end()) { pcountSize++; if (pcountSize > 1) { break; } } } - tempTotals[i] = 0.0; //we don't care about you, or we have already added you - }else{ //if you are not a leaf - //do both your chidren have have descendants from the users groups? - int lc = t->tree[i].getLChild(); - int rc = t->tree[i].getRChild(); - //if yes, add your childrens tempTotals - if ((nodePcountSize[lc] != 0) && (nodePcountSize[rc] != 0)) { - totalBL += tempTotals[lc] + tempTotals[rc]; - //cout << "added to total " << tempTotals[lc] << '\t' << tempTotals[rc] << endl; - if (t->tree[i].getBranchLength() != -1) { - tempTotals[i] = abs(t->tree[i].getBranchLength()); - }else { - tempTotals[i] = 0.0; - } - }else if ((nodePcountSize[lc] == 0) && (nodePcountSize[rc] == 0)) { tempTotals[i] = 0.0; //we don't care about you - }else { //if no, your tempTotal is your childrens temp totals + your branch length - tempTotals[i] = tempTotals[lc] + tempTotals[rc] + 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 << "temptotal = "<< tempTotals[i] << endl; } - - } //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; } @@ -266,7 +261,6 @@ EstOutput Unweighted::getValues(Tree* t, string groupA, string groupB, int p, st processors = p; outputDir = o; - //if the users enters no groups then give them the score of all groups int numGroups = globaldata->Groups.size(); @@ -314,7 +308,7 @@ 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(); @@ -336,7 +330,6 @@ 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; @@ -363,7 +356,11 @@ 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); @@ -386,12 +383,13 @@ 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); @@ -415,9 +413,6 @@ EstOutput Unweighted::driver(Tree* t, vector< vector > namesOfGroupCombo EstOutput results; results.resize(num); int count = 0; - int total = num; - int twentyPercent = (total * 0.20); - int numLeaves = t->getNumLeaves(); Tree* copyTree = new Tree; @@ -434,64 +429,59 @@ 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; - map tempTotals; //maps node to total Branch Length - map nodePcountSize; //maps node to pcountSize - - 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 - - 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; } } - } + //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{ - nodePcountSize[i] = pcountSize; - - if (pcountSize == 0) { } - else if ((copyTree->tree[i].getBranchLength() != -1) && (pcountSize == 1)) { UniqueBL += abs(copyTree->tree[i].getBranchLength()); } + //if including the root this clears rootForGrouping[namesOfGroupCombos[h]] + getRoot(copyTree, nodeBelonging, namesOfGroupCombos[h]); - //if you are a leaf from a users group add to total - if (i < numLeaves) { - if ((copyTree->tree[i].getBranchLength() != -1) && pcountSize != 0) { - totalBL += abs(copyTree->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; } } } - tempTotals[i] = 0.0; //we don't care about you, or we have already added you - }else{ //if you are not a leaf - //do both your chidren have have descendants from the users groups? - int lc = copyTree->tree[i].getLChild(); - int rc = copyTree->tree[i].getRChild(); - //if yes, add your childrens tempTotals - if ((nodePcountSize[lc] != 0) && (nodePcountSize[rc] != 0)) { - totalBL += tempTotals[lc] + tempTotals[rc]; - - if (copyTree->tree[i].getBranchLength() != -1) { - tempTotals[i] = abs(copyTree->tree[i].getBranchLength()); - }else { - tempTotals[i] = 0.0; - } - }else if ((nodePcountSize[lc] == 0) && (nodePcountSize[rc] == 0)) { tempTotals[i] = 0.0; //we don't care about you - }else { //if no, your tempTotal is your childrens temp totals + your branch length - tempTotals[i] = tempTotals[lc] + tempTotals[rc] + abs(copyTree->tree[i].getBranchLength()); + //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; } - - UW = (UniqueBL / totalBL); - - if (isnan(UW) || isinf(UW)) { UW = 0; } - - results[count] = UW; count++; - + } delete copyTree; @@ -504,5 +494,65 @@ 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(); + + 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){ + + 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); + } +} +/**************************************************************************************************/