2 * unifracweightedcommand.cpp
5 * Created by Sarah Westcott on 2/9/09.
6 * Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
10 #include "unifracweightedcommand.h"
12 /***********************************************************/
13 UnifracWeightedCommand::UnifracWeightedCommand(string option) {
15 globaldata = GlobalData::getInstance();
19 //allow user to run help
20 if(option == "help") { help(); abort = true; }
23 //valid paramters for this command
24 string Array[] = {"groups","iters","distance","random","processors","outputdir","inputdir"};
25 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
27 OptionParser parser(option);
28 map<string,string> parameters=parser.getParameters();
30 ValidParameters validParameter;
32 //check to make sure all parameters are valid for command
33 for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
34 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) { abort = true; }
37 if (globaldata->gTree.size() == 0) {//no trees were read
38 m->mothurOut("You must execute the read.tree command, before you may execute the unifrac.weighted command."); m->mothurOutEndLine(); abort = true; }
40 //if the user changes the output directory command factory will send this info to us in the output parameter
41 outputDir = validParameter.validFile(parameters, "outputdir", false); if (outputDir == "not found"){
43 outputDir += m->hasPath(globaldata->inputFileName); //if user entered a file with a path then preserve it
46 //check for optional parameter and set defaults
47 // ...at some point should added some additional type checking...
48 groups = validParameter.validFile(parameters, "groups", false);
49 if (groups == "not found") { groups = ""; }
51 m->splitAtDash(groups, Groups);
52 globaldata->Groups = Groups;
55 itersString = validParameter.validFile(parameters, "iters", false); if (itersString == "not found") { itersString = "1000"; }
56 convert(itersString, iters);
58 string temp = validParameter.validFile(parameters, "distance", false); if (temp == "not found") { temp = "false"; }
59 phylip = m->isTrue(temp);
61 temp = validParameter.validFile(parameters, "random", false); if (temp == "not found") { temp = "F"; }
62 random = m->isTrue(temp);
64 temp = validParameter.validFile(parameters, "processors", false); if (temp == "not found"){ temp = "1"; }
65 convert(temp, processors);
67 if (!random) { iters = 0; } //turn off random calcs
71 T = globaldata->gTree;
72 tmap = globaldata->gTreemap;
73 sumFile = outputDir + m->getSimpleName(globaldata->getTreeFile()) + ".wsummary";
74 m->openOutputFile(sumFile, outSum);
75 outputNames.push_back(sumFile);
77 util = new SharedUtil();
78 string s; //to make work with setgroups
79 util->setGroups(globaldata->Groups, tmap->namesOfGroups, s, numGroups, "weighted"); //sets the groups the user wants to analyze
80 util->getCombos(groupComb, globaldata->Groups, numComp);
82 weighted = new Weighted(tmap);
90 m->errorOut(e, "UnifracWeightedCommand", "UnifracWeightedCommand");
94 //**********************************************************************************************************************
96 void UnifracWeightedCommand::help(){
98 m->mothurOut("The unifrac.weighted command can only be executed after a successful read.tree command.\n");
99 m->mothurOut("The unifrac.weighted command parameters are groups, iters, distance and random. No parameters are required.\n");
100 m->mothurOut("The groups parameter allows you to specify which of the groups in your groupfile you would like analyzed. You must enter at least 2 valid groups.\n");
101 m->mothurOut("The group names are separated by dashes. The iters parameter allows you to specify how many random trees you would like compared to your tree.\n");
102 m->mothurOut("The distance parameter allows you to create a distance file from the results. The default is false.\n");
103 m->mothurOut("The random parameter allows you to shut off the comparison to random trees. The default is false, meaning don't compare your trees with randomly generated trees.\n");
104 m->mothurOut("The unifrac.weighted command should be in the following format: unifrac.weighted(groups=yourGroups, iters=yourIters).\n");
105 m->mothurOut("Example unifrac.weighted(groups=A-B-C, iters=500).\n");
106 m->mothurOut("The default value for groups is all the groups in your groupfile, and iters is 1000.\n");
107 m->mothurOut("The unifrac.weighted command output two files: .weighted and .wsummary their descriptions are in the manual.\n");
108 m->mothurOut("Note: No spaces between parameter labels (i.e. groups), '=' and parameters (i.e.yourGroups).\n\n");
110 catch(exception& e) {
111 m->errorOut(e, "UnifracWeightedCommand", "help");
116 /***********************************************************/
117 int UnifracWeightedCommand::execute() {
120 if (abort == true) { return 0; }
122 int start = time(NULL);
124 //get weighted for users tree
125 userData.resize(numComp,0); //data[0] = weightedscore AB, data[1] = weightedscore AC...
126 randomData.resize(numComp,0); //data[0] = weightedscore AB, data[1] = weightedscore AC...
128 //get weighted scores for users trees
129 for (int i = 0; i < T.size(); i++) {
131 if (m->control_pressed) { outSum.close(); for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } return 0; }
134 rScores.resize(numComp); //data[0] = weightedscore AB, data[1] = weightedscore AC...
135 uScores.resize(numComp); //data[0] = weightedscore AB, data[1] = weightedscore AC...
138 output = new ColumnFile(outputDir + m->getSimpleName(globaldata->getTreeFile()) + toString(i+1) + ".weighted", itersString);
139 outputNames.push_back(outputDir + m->getSimpleName(globaldata->getTreeFile()) + toString(i+1) + ".weighted");
142 userData = weighted->getValues(T[i], processors, outputDir); //userData[0] = weightedscore
144 if (m->control_pressed) { if (random) { delete output; } outSum.close(); for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } return 0; }
147 for (int s=0; s<numComp; s++) {
148 //add users score to vector of user scores
149 uScores[s].push_back(userData[s]);
151 //save users tree score for summary file
152 utreeScores.push_back(userData[s]);
156 vector<double> sums = weighted->getBranchLengthSums(T[i]);
158 //calculate number of comparisons i.e. with groups A,B,C = AB, AC, BC = 3;
159 vector< vector<string> > namesOfGroupCombos;
160 for (int a=0; a<numGroups; a++) {
161 for (int l = 0; l < a; l++) {
162 vector<string> groups; groups.push_back(globaldata->Groups[a]); groups.push_back(globaldata->Groups[l]);
163 namesOfGroupCombos.push_back(groups);
169 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
171 int numPairs = namesOfGroupCombos.size();
172 int numPairsPerProcessor = numPairs / processors;
174 for (int i = 0; i < processors; i++) {
175 int startPos = i * numPairsPerProcessor;
176 if(i == processors - 1){
177 numPairsPerProcessor = numPairs - i * numPairsPerProcessor;
179 lines.push_back(linePair(startPos, numPairsPerProcessor));
185 //get scores for random trees
186 for (int j = 0; j < iters; j++) {
188 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
190 driver(T[i], namesOfGroupCombos, 0, namesOfGroupCombos.size(), sums, rScores);
192 createProcesses(T[i], namesOfGroupCombos, sums, rScores);
195 driver(T[i], namesOfGroupCombos, 0, namesOfGroupCombos.size(), sums, rScores);
198 if (m->control_pressed) { delete output; outSum.close(); for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } return 0; }
201 m->mothurOut("Iter: " + toString(j+1)); m->mothurOutEndLine();
205 //find the signifigance of the score for summary file
206 for (int f = 0; f < numComp; f++) {
208 sort(rScores[f].begin(), rScores[f].end());
210 //the index of the score higher than yours is returned
211 //so if you have 1000 random trees the index returned is 100
212 //then there are 900 trees with a score greater then you.
213 //giving you a signifigance of 0.900
214 int index = findIndex(userData[f], f); if (index == -1) { m->mothurOut("error in UnifracWeightedCommand"); m->mothurOutEndLine(); exit(1); } //error code
216 //the signifigance is the number of trees with the users score or higher
217 WScoreSig.push_back((iters-index)/(float)iters);
220 //out << "Tree# " << i << endl;
221 calculateFreqsCumuls();
235 if (m->control_pressed) { outSum.close(); for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } return 0; }
239 if (phylip) { createPhylipFile(); }
241 //clear out users groups
242 globaldata->Groups.clear();
245 if (m->control_pressed) {
246 for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); }
250 m->mothurOut("It took " + toString(time(NULL) - start) + " secs to run unifrac.weighted."); m->mothurOutEndLine();
252 m->mothurOutEndLine();
253 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
254 for (int i = 0; i < outputNames.size(); i++) { m->mothurOut(outputNames[i]); m->mothurOutEndLine(); }
255 m->mothurOutEndLine();
260 catch(exception& e) {
261 m->errorOut(e, "UnifracWeightedCommand", "execute");
265 /**************************************************************************************************/
267 int UnifracWeightedCommand::createProcesses(Tree* t, vector< vector<string> > namesOfGroupCombos, vector<double>& sums, vector< vector<double> >& scores) {
269 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
272 vector<int> processIDS;
276 //loop through and create all the processes you want
277 while (process != processors) {
281 processIDS.push_back(pid); //create map from line number to pid so you can append files in correct order later
284 driver(t, namesOfGroupCombos, lines[process].start, lines[process].num, sums, scores);
286 //pass numSeqs to parent
288 string tempFile = outputDir + toString(getpid()) + ".weightedcommand.results.temp";
289 m->openOutputFile(tempFile, out);
290 for (int i = lines[process].start; i < (lines[process].start + lines[process].num); i++) { out << scores[i][(scores[i].size()-1)] << '\t'; } out << endl;
294 }else { m->mothurOut("unable to spawn the necessary processes."); m->mothurOutEndLine(); exit(0); }
297 driver(t, namesOfGroupCombos, lines[0].start, lines[0].num, sums, scores);
299 //force parent to wait until all the processes are done
300 for (int i=0;i<(processors-1);i++) {
301 int temp = processIDS[i];
305 //get data created by processes
306 for (int i=0;i<(processors-1);i++) {
309 string s = outputDir + toString(processIDS[i]) + ".weightedcommand.results.temp";
310 m->openInputFile(s, in);
313 for (int j = lines[(i+1)].start; j < (lines[(i+1)].start + lines[(i+1)].num); j++) { in >> tempScore; scores[j].push_back(tempScore); }
321 catch(exception& e) {
322 m->errorOut(e, "UnifracWeightedCommand", "createProcesses");
327 /**************************************************************************************************/
328 int UnifracWeightedCommand::driver(Tree* t, vector< vector<string> > namesOfGroupCombos, int start, int num, vector<double>& sums, vector< vector<double> >& scores) {
330 Tree* randT = new Tree();
332 for (int h = start; h < (start+num); h++) {
334 if (m->control_pressed) { return 0; }
336 //initialize weighted score
337 string groupA = namesOfGroupCombos[h][0];
338 string groupB = namesOfGroupCombos[h][1];
343 //create a random tree with same topology as T[i], but different labels
344 randT->assembleRandomUnifracTree(groupA, groupB);
346 if (m->control_pressed) { delete randT; return 0; }
348 //get wscore of random tree
349 EstOutput randomData = weighted->getValues(randT, groupA, groupB, sums);
351 if (m->control_pressed) { delete randT; return 0; }
354 scores[h].push_back(randomData[0]);
362 catch(exception& e) {
363 m->errorOut(e, "UnifracWeightedCommand", "driver");
367 /***********************************************************/
368 void UnifracWeightedCommand::printWeightedFile() {
372 tags.push_back("Score"); tags.push_back("RandFreq"); tags.push_back("RandCumul");
374 for(int a = 0; a < numComp; a++) {
375 output->initFile(groupComb[a], tags);
377 for (map<float,float>::iterator it = validScores.begin(); it != validScores.end(); it++) {
378 data.push_back(it->first); data.push_back(rScoreFreq[a][it->first]); data.push_back(rCumul[a][it->first]);
379 output->output(data);
385 catch(exception& e) {
386 m->errorOut(e, "UnifracWeightedCommand", "printWeightedFile");
392 /***********************************************************/
393 void UnifracWeightedCommand::printWSummaryFile() {
396 outSum << "Tree#" << '\t' << "Groups" << '\t' << "WScore" << '\t' << "WSig" << endl;
397 m->mothurOut("Tree#\tGroups\tWScore\tWSig"); m->mothurOutEndLine();
400 outSum.setf(ios::fixed, ios::floatfield); outSum.setf(ios::showpoint);
404 for (int i = 0; i < T.size(); i++) {
405 for (int j = 0; j < numComp; j++) {
407 if (WScoreSig[count] > (1/(float)iters)) {
408 outSum << setprecision(6) << i+1 << '\t' << groupComb[j] << '\t' << utreeScores[count] << '\t' << setprecision(itersString.length()) << WScoreSig[count] << endl;
409 cout << setprecision(6) << i+1 << '\t' << groupComb[j] << '\t' << utreeScores[count] << '\t' << setprecision(itersString.length()) << WScoreSig[count] << endl;
410 m->mothurOutJustToLog(toString(i+1) +"\t" + groupComb[j] +"\t" + toString(utreeScores[count]) +"\t" + toString(WScoreSig[count]) + "\n");
412 outSum << setprecision(6) << i+1 << '\t' << groupComb[j] << '\t' << utreeScores[count] << '\t' << setprecision(itersString.length()) << "<" << (1/float(iters)) << endl;
413 cout << setprecision(6) << i+1 << '\t' << groupComb[j] << '\t' << utreeScores[count] << '\t' << setprecision(itersString.length()) << "<" << (1/float(iters)) << endl;
414 m->mothurOutJustToLog(toString(i+1) +"\t" + groupComb[j] +"\t" + toString(utreeScores[count]) +"\t<" + toString((1/float(iters))) + "\n");
417 outSum << setprecision(6) << i+1 << '\t' << groupComb[j] << '\t' << utreeScores[count] << '\t' << "0.00" << endl;
418 cout << setprecision(6) << i+1 << '\t' << groupComb[j] << '\t' << utreeScores[count] << '\t' << "0.00" << endl;
419 m->mothurOutJustToLog(toString(i+1) +"\t" + groupComb[j] +"\t" + toString(utreeScores[count]) +"\t0.00\n");
426 catch(exception& e) {
427 m->errorOut(e, "UnifracWeightedCommand", "printWSummaryFile");
431 /***********************************************************/
432 void UnifracWeightedCommand::createPhylipFile() {
436 for (int i = 0; i < T.size(); i++) {
438 string phylipFileName = outputDir + m->getSimpleName(globaldata->getTreeFile()) + toString(i+1) + ".weighted.dist";
439 outputNames.push_back(phylipFileName);
441 m->openOutputFile(phylipFileName, out);
444 out << globaldata->Groups.size() << endl;
446 //make matrix with scores in it
447 vector< vector<float> > dists; dists.resize(globaldata->Groups.size());
448 for (int i = 0; i < globaldata->Groups.size(); i++) {
449 dists[i].resize(globaldata->Groups.size(), 0.0);
452 //flip it so you can print it
453 for (int r=0; r<globaldata->Groups.size(); r++) {
454 for (int l = r+1; l < globaldata->Groups.size(); l++) {
455 dists[r][l] = utreeScores[count];
456 dists[l][r] = utreeScores[count];
462 for (int r=0; r<globaldata->Groups.size(); r++) {
464 string name = globaldata->Groups[r];
465 if (name.length() < 10) { //pad with spaces to make compatible
466 while (name.length() < 10) { name += " "; }
471 for (int l = 0; l < r; l++) { out << dists[r][l] << '\t'; }
477 catch(exception& e) {
478 m->errorOut(e, "UnifracWeightedCommand", "createPhylipFile");
482 /***********************************************************/
483 int UnifracWeightedCommand::findIndex(float score, int index) {
485 for (int i = 0; i < rScores[index].size(); i++) {
486 if (rScores[index][i] >= score) { return i; }
488 return rScores[index].size();
490 catch(exception& e) {
491 m->errorOut(e, "UnifracWeightedCommand", "findIndex");
496 /***********************************************************/
498 void UnifracWeightedCommand::calculateFreqsCumuls() {
500 //clear out old tree values
502 rScoreFreq.resize(numComp);
504 rCumul.resize(numComp);
507 //calculate frequency
508 for (int f = 0; f < numComp; f++) {
509 for (int i = 0; i < rScores[f].size(); i++) { //looks like 0,0,1,1,1,2,4,7... you want to make a map that say rScoreFreq[0] = 2, rScoreFreq[1] = 3...
510 validScores[rScores[f][i]] = rScores[f][i];
511 map<float,float>::iterator it = rScoreFreq[f].find(rScores[f][i]);
512 if (it != rScoreFreq[f].end()) {
513 rScoreFreq[f][rScores[f][i]]++;
515 rScoreFreq[f][rScores[f][i]] = 1;
521 for(int a = 0; a < numComp; a++) {
522 float rcumul = 1.0000;
523 //this loop fills the cumulative maps and put 0.0000 in the score freq map to make it easier to print.
524 for (map<float,float>::iterator it = validScores.begin(); it != validScores.end(); it++) {
525 //make rscoreFreq map and rCumul
526 map<float,float>::iterator it2 = rScoreFreq[a].find(it->first);
527 rCumul[a][it->first] = rcumul;
528 //get percentage of random trees with that info
529 if (it2 != rScoreFreq[a].end()) { rScoreFreq[a][it->first] /= iters; rcumul-= it2->second; }
530 else { rScoreFreq[a][it->first] = 0.0000; } //no random trees with that score
535 catch(exception& e) {
536 m->errorOut(e, "UnifracWeightedCommand", "calculateFreqsCums");
541 /***********************************************************/