5 * Created by Sarah Westcott on 2/9/09.
6 * Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
10 #include "unweighted.h"
12 /**************************************************************************************************/
14 EstOutput Unweighted::getValues(Tree* t, int p, string o) {
16 globaldata = GlobalData::getInstance();
20 //if the users enters no groups then give them the score of all groups
21 int numGroups = globaldata->Groups.size();
23 //calculate number of comparsions
25 vector< vector<string> > namesOfGroupCombos;
26 for (int r=0; r<numGroups; r++) {
27 for (int l = 0; l < r; l++) {
29 vector<string> groups; groups.push_back(globaldata->Groups[r]); groups.push_back(globaldata->Groups[l]);
30 namesOfGroupCombos.push_back(groups);
35 vector<string> groups;
37 //get score for all users groups
38 for (int i = 0; i < tmap->namesOfGroups.size(); i++) {
39 if (tmap->namesOfGroups[i] != "xxx") {
40 groups.push_back(tmap->namesOfGroups[i]);
43 namesOfGroupCombos.push_back(groups);
45 for (int i = 0; i < globaldata->Groups.size(); i++) {
46 groups.push_back(globaldata->Groups[i]);
48 namesOfGroupCombos.push_back(groups);
52 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
54 data = driver(t, namesOfGroupCombos, 0, namesOfGroupCombos.size());
56 int numPairs = namesOfGroupCombos.size();
58 int numPairsPerProcessor = numPairs / processors;
60 for (int i = 0; i < processors; i++) {
61 int startPos = i * numPairsPerProcessor;
63 if(i == processors - 1){
64 numPairsPerProcessor = numPairs - i * numPairsPerProcessor;
67 lines.push_back(linePair(startPos, numPairsPerProcessor));
69 data = createProcesses(t, namesOfGroupCombos);
73 data = driver(t, namesOfGroupCombos, 0, namesOfGroupCombos.size());
79 m->errorOut(e, "Unweighted", "getValues");
83 /**************************************************************************************************/
85 EstOutput Unweighted::createProcesses(Tree* t, vector< vector<string> > namesOfGroupCombos) {
87 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
89 vector<int> processIDS;
93 //loop through and create all the processes you want
94 while (process != processors) {
98 processIDS.push_back(pid); //create map from line number to pid so you can append files in correct order later
102 myresults = driver(t, namesOfGroupCombos, lines[process].start, lines[process].num);
104 if (m->control_pressed) { exit(0); }
106 m->mothurOut("Merging results."); m->mothurOutEndLine();
108 //pass numSeqs to parent
110 string tempFile = outputDir + toString(getpid()) + ".unweighted.results.temp";
111 m->openOutputFile(tempFile, out);
112 out << myresults.size() << endl;
113 for (int i = 0; i < myresults.size(); i++) { out << myresults[i] << '\t'; } out << endl;
118 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
119 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
124 results = driver(t, namesOfGroupCombos, lines[0].start, lines[0].num);
126 //force parent to wait until all the processes are done
127 for (int i=0;i<(processors-1);i++) {
128 int temp = processIDS[i];
132 if (m->control_pressed) { return results; }
134 //get data created by processes
135 for (int i=0;i<(processors-1);i++) {
137 string s = outputDir + toString(processIDS[i]) + ".unweighted.results.temp";
138 m->openInputFile(s, in);
143 in >> num; m->gobble(in);
145 if (m->control_pressed) { break; }
148 for (int j = 0; j < num; j++) {
150 results.push_back(w);
158 m->mothurOut("DONE."); m->mothurOutEndLine(); m->mothurOutEndLine();
163 catch(exception& e) {
164 m->errorOut(e, "Unweighted", "createProcesses");
168 /**************************************************************************************************/
169 EstOutput Unweighted::driver(Tree* t, vector< vector<string> > namesOfGroupCombos, int start, int num) {
172 EstOutput results; results.resize(num);
175 int numLeaves = t->getNumLeaves();
177 int twentyPercent = (total * 0.20);
178 if (twentyPercent == 0) { twentyPercent = 1; }
181 for (int h = start; h < (start+num); h++) {
182 //cout << namesOfGroupCombos[h][0] << '\t' << namesOfGroupCombos[h][1] << endl;
183 if (m->control_pressed) { return results; }
185 double UniqueBL=0.0000; //a branch length is unique if it's chidren are from the same group
186 double totalBL = 0.00; //all branch lengths
187 double UW = 0.00; //Unweighted Value = UniqueBL / totalBL;
188 map<int, double> tempTotals; //maps node to total Branch Length
189 map<int, int> nodePcountSize; //maps node to pcountSize
190 map<int, int>::iterator itCount;
192 for(int i=0;i<t->getNumNodes();i++){
194 if (m->control_pressed) { return data; }
196 //pcountSize = 0, they are from a branch that is entirely from a group the user doesn't want
197 //pcountSize = 2, not unique to one group
198 //pcountSize = 1, unique to one group
201 for (int j = 0; j < namesOfGroupCombos[h].size(); j++) {
202 map<string, int>::iterator itGroup = t->tree[i].pcount.find(namesOfGroupCombos[h][j]);
203 if (itGroup != t->tree[i].pcount.end()) { pcountSize++; if (pcountSize > 1) { break; } }
206 nodePcountSize[i] = pcountSize;
208 //cout << i << '\t' << t->tree[i].getName() << " br = " << abs(t->tree[i].getBranchLength()) << '\t';
209 if (pcountSize == 0) { }
210 else if ((t->tree[i].getBranchLength() != -1) && (pcountSize == 1)) { UniqueBL += abs(t->tree[i].getBranchLength()); }
212 //if you are a leaf from a users group add to total
214 if ((t->tree[i].getBranchLength() != -1) && pcountSize != 0) {
215 //cout << "added to total" << endl;
216 totalBL += abs(t->tree[i].getBranchLength());
218 tempTotals[i] = 0.0; //we don't care about you, or we have already added you
219 }else{ //if you are not a leaf
220 //do both your chidren have have descendants from the users groups?
221 int lc = t->tree[i].getLChild();
222 int rc = t->tree[i].getRChild();
224 //if yes, add your childrens tempTotals
225 if ((nodePcountSize[lc] != 0) && (nodePcountSize[rc] != 0)) {
226 totalBL += tempTotals[lc] + tempTotals[rc];
227 //cout << "added to total " << tempTotals[lc] << '\t' << tempTotals[rc] << endl;
228 if (t->tree[i].getBranchLength() != -1) {
229 tempTotals[i] = abs(t->tree[i].getBranchLength());
233 }else if ((nodePcountSize[lc] == 0) && (nodePcountSize[rc] == 0)) { tempTotals[i] = 0.0; //we don't care about you
234 }else { //if no, your tempTotal is your childrens temp totals + your branch length
235 tempTotals[i] = tempTotals[lc] + tempTotals[rc] + abs(t->tree[i].getBranchLength());
237 //cout << "temptotal = "<< tempTotals[i] << endl;
241 //cout << UniqueBL << '\t' << totalBL << endl;
242 UW = (UniqueBL / totalBL);
244 if (isnan(UW) || isinf(UW)) { UW = 0; }
250 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(); }
254 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(); }
258 catch(exception& e) {
259 m->errorOut(e, "Unweighted", "driver");
263 /**************************************************************************************************/
265 EstOutput Unweighted::getValues(Tree* t, string groupA, string groupB, int p, string o) {
267 globaldata = GlobalData::getInstance();
272 //if the users enters no groups then give them the score of all groups
273 int numGroups = globaldata->Groups.size();
275 //calculate number of comparsions
277 vector< vector<string> > namesOfGroupCombos;
278 for (int r=0; r<numGroups; r++) {
279 for (int l = 0; l < r; l++) {
281 vector<string> groups; groups.push_back(globaldata->Groups[r]); groups.push_back(globaldata->Groups[l]);
282 namesOfGroupCombos.push_back(groups);
287 vector<string> groups;
288 if (numGroups == 0) {
289 //get score for all users groups
290 for (int i = 0; i < tmap->namesOfGroups.size(); i++) {
291 if (tmap->namesOfGroups[i] != "xxx") {
292 groups.push_back(tmap->namesOfGroups[i]);
295 namesOfGroupCombos.push_back(groups);
297 for (int i = 0; i < globaldata->Groups.size(); i++) {
298 groups.push_back(globaldata->Groups[i]);
300 namesOfGroupCombos.push_back(groups);
304 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
306 data = driver(t, namesOfGroupCombos, 0, namesOfGroupCombos.size(), true);
308 int numPairs = namesOfGroupCombos.size();
310 int numPairsPerProcessor = numPairs / processors;
312 for (int i = 0; i < processors; i++) {
313 int startPos = i * numPairsPerProcessor;
314 if(i == processors - 1){
315 numPairsPerProcessor = numPairs - i * numPairsPerProcessor;
317 lines.push_back(linePair(startPos, numPairsPerProcessor));
320 data = createProcesses(t, namesOfGroupCombos, true);
325 data = driver(t, namesOfGroupCombos, 0, namesOfGroupCombos.size(), true);
330 catch(exception& e) {
331 m->errorOut(e, "Unweighted", "getValues");
335 /**************************************************************************************************/
337 EstOutput Unweighted::createProcesses(Tree* t, vector< vector<string> > namesOfGroupCombos, bool usingGroups) {
339 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
341 vector<int> processIDS;
345 //loop through and create all the processes you want
346 while (process != processors) {
350 processIDS.push_back(pid); //create map from line number to pid so you can append files in correct order later
354 myresults = driver(t, namesOfGroupCombos, lines[process].start, lines[process].num, usingGroups);
356 if (m->control_pressed) { exit(0); }
358 //pass numSeqs to parent
360 string tempFile = outputDir + toString(getpid()) + ".unweighted.results.temp";
361 m->openOutputFile(tempFile, out);
362 out << myresults.size() << endl;
363 for (int i = 0; i < myresults.size(); i++) { out << myresults[i] << '\t'; } out << endl;
368 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
369 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
374 results = driver(t, namesOfGroupCombos, lines[0].start, lines[0].num, usingGroups);
376 //force parent to wait until all the processes are done
377 for (int i=0;i<(processors-1);i++) {
378 int temp = processIDS[i];
382 if (m->control_pressed) { return results; }
384 //get data created by processes
385 for (int i=0;i<(processors-1);i++) {
387 string s = outputDir + toString(processIDS[i]) + ".unweighted.results.temp";
388 m->openInputFile(s, in);
393 in >> num; m->gobble(in);
395 if (m->control_pressed) { break; }
398 for (int j = 0; j < num; j++) {
401 results.push_back(w);
412 catch(exception& e) {
413 m->errorOut(e, "Unweighted", "createProcesses");
417 /**************************************************************************************************/
418 EstOutput Unweighted::driver(Tree* t, vector< vector<string> > namesOfGroupCombos, int start, int num, bool usingGroups) {
421 EstOutput results; results.resize(num);
424 int numLeaves = t->getNumLeaves();
426 Tree* copyTree = new Tree;
428 for (int h = start; h < (start+num); h++) {
430 if (m->control_pressed) { return results; }
432 //copy random tree passed in
433 copyTree->getCopy(t);
435 //swap labels in the groups you want to compare
436 copyTree->assembleRandomUnifracTree(namesOfGroupCombos[h]);
438 double UniqueBL=0.0000; //a branch length is unique if it's chidren are from the same group
439 double totalBL = 0.00; //all branch lengths
440 double UW = 0.00; //Unweighted Value = UniqueBL / totalBL;
441 map<int, double> tempTotals; //maps node to total Branch Length
442 map<int, int> nodePcountSize; //maps node to pcountSize
444 for(int i=0;i<copyTree->getNumNodes();i++){
446 if (m->control_pressed) { return data; }
448 //pcountSize = 0, they are from a branch that is entirely from a group the user doesn't want
449 //pcountSize = 2, not unique to one group
450 //pcountSize = 1, unique to one group
453 for (int j = 0; j < namesOfGroupCombos[h].size(); j++) {
454 map<string, int>::iterator itGroup = copyTree->tree[i].pcount.find(namesOfGroupCombos[h][j]);
455 if (itGroup != copyTree->tree[i].pcount.end()) { pcountSize++; if (pcountSize > 1) { break; } }
458 nodePcountSize[i] = pcountSize;
460 if (pcountSize == 0) { }
461 else if ((copyTree->tree[i].getBranchLength() != -1) && (pcountSize == 1)) { UniqueBL += abs(copyTree->tree[i].getBranchLength()); }
463 //if you are a leaf from a users group add to total
465 if ((copyTree->tree[i].getBranchLength() != -1) && pcountSize != 0) {
466 totalBL += abs(copyTree->tree[i].getBranchLength());
468 tempTotals[i] = 0.0; //we don't care about you, or we have already added you
469 }else{ //if you are not a leaf
470 //do both your chidren have have descendants from the users groups?
471 int lc = copyTree->tree[i].getLChild();
472 int rc = copyTree->tree[i].getRChild();
474 //if yes, add your childrens tempTotals
475 if ((nodePcountSize[lc] != 0) && (nodePcountSize[rc] != 0)) {
476 totalBL += tempTotals[lc] + tempTotals[rc];
478 if (copyTree->tree[i].getBranchLength() != -1) {
479 tempTotals[i] = abs(copyTree->tree[i].getBranchLength());
483 }else if ((nodePcountSize[lc] == 0) && (nodePcountSize[rc] == 0)) { tempTotals[i] = 0.0; //we don't care about you
484 }else { //if no, your tempTotal is your childrens temp totals + your branch length
485 tempTotals[i] = tempTotals[lc] + tempTotals[rc] + abs(copyTree->tree[i].getBranchLength());
492 UW = (UniqueBL / totalBL);
494 if (isnan(UW) || isinf(UW)) { UW = 0; }
505 catch(exception& e) {
506 m->errorOut(e, "Unweighted", "driver");
510 /**************************************************************************************************/