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 vector<string> UnifracWeightedCommand::getValidParameters(){
15 string Array[] = {"groups","iters","distance","random","processors","outputdir","inputdir"};
16 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
20 m->errorOut(e, "UnifracWeightedCommand", "getValidParameters");
24 //**********************************************************************************************************************
25 UnifracWeightedCommand::UnifracWeightedCommand(){
27 abort = true; calledHelp = true;
28 vector<string> tempOutNames;
29 outputTypes["weighted"] = tempOutNames;
30 outputTypes["wsummary"] = tempOutNames;
31 outputTypes["phylip"] = tempOutNames;
32 outputTypes["column"] = tempOutNames;
35 m->errorOut(e, "UnifracWeightedCommand", "UnifracWeightedCommand");
39 //**********************************************************************************************************************
40 vector<string> UnifracWeightedCommand::getRequiredParameters(){
42 vector<string> myArray;
46 m->errorOut(e, "UnifracWeightedCommand", "getRequiredParameters");
50 //**********************************************************************************************************************
51 vector<string> UnifracWeightedCommand::getRequiredFiles(){
53 string Array[] = {"tree","group"};
54 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
59 m->errorOut(e, "UnifracWeightedCommand", "getRequiredFiles");
63 /***********************************************************/
64 UnifracWeightedCommand::UnifracWeightedCommand(string option) {
66 globaldata = GlobalData::getInstance();
67 abort = false; calledHelp = false;
70 //allow user to run help
71 if(option == "help") { help(); abort = true; calledHelp = true; }
74 //valid paramters for this command
75 string Array[] = {"groups","iters","distance","random","processors","outputdir","inputdir"};
76 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
78 OptionParser parser(option);
79 map<string,string> parameters=parser.getParameters();
81 ValidParameters validParameter;
83 //check to make sure all parameters are valid for command
84 for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
85 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) { abort = true; }
88 //initialize outputTypes
89 vector<string> tempOutNames;
90 outputTypes["weighted"] = tempOutNames;
91 outputTypes["wsummary"] = tempOutNames;
92 outputTypes["phylip"] = tempOutNames;
93 outputTypes["column"] = tempOutNames;
95 if (globaldata->gTree.size() == 0) {//no trees were read
96 m->mothurOut("You must execute the read.tree command, before you may execute the unifrac.weighted command."); m->mothurOutEndLine(); abort = true; }
98 //if the user changes the output directory command factory will send this info to us in the output parameter
99 outputDir = validParameter.validFile(parameters, "outputdir", false); if (outputDir == "not found"){
101 outputDir += m->hasPath(globaldata->inputFileName); //if user entered a file with a path then preserve it
104 //check for optional parameter and set defaults
105 // ...at some point should added some additional type checking...
106 groups = validParameter.validFile(parameters, "groups", false);
107 if (groups == "not found") { groups = ""; }
109 m->splitAtDash(groups, Groups);
110 globaldata->Groups = Groups;
113 itersString = validParameter.validFile(parameters, "iters", false); if (itersString == "not found") { itersString = "1000"; }
114 convert(itersString, iters);
116 string temp = validParameter.validFile(parameters, "distance", false);
117 if (temp == "not found") { phylip = false; outputForm = ""; }
119 if ((temp == "lt") || (temp == "column") || (temp == "square")) { phylip = true; outputForm = temp; }
120 else { m->mothurOut("Options for distance are: lt, square, or column. Using lt."); m->mothurOutEndLine(); phylip = true; outputForm = "lt"; }
123 temp = validParameter.validFile(parameters, "random", false); if (temp == "not found") { temp = "F"; }
124 random = m->isTrue(temp);
126 temp = validParameter.validFile(parameters, "processors", false); if (temp == "not found"){ temp = "1"; }
127 convert(temp, processors);
129 if (!random) { iters = 0; } //turn off random calcs
132 if (abort == false) {
133 T = globaldata->gTree;
134 tmap = globaldata->gTreemap;
135 sumFile = outputDir + m->getSimpleName(globaldata->getTreeFile()) + ".wsummary";
136 m->openOutputFile(sumFile, outSum);
137 outputNames.push_back(sumFile); outputTypes["wsummary"].push_back(sumFile);
139 util = new SharedUtil();
140 string s; //to make work with setgroups
141 util->setGroups(globaldata->Groups, tmap->namesOfGroups, s, numGroups, "weighted"); //sets the groups the user wants to analyze
142 util->getCombos(groupComb, globaldata->Groups, numComp);
144 weighted = new Weighted(tmap);
151 catch(exception& e) {
152 m->errorOut(e, "UnifracWeightedCommand", "UnifracWeightedCommand");
156 //**********************************************************************************************************************
158 void UnifracWeightedCommand::help(){
160 m->mothurOut("The unifrac.weighted command can only be executed after a successful read.tree command.\n");
161 m->mothurOut("The unifrac.weighted command parameters are groups, iters, distance, processors and random. No parameters are required.\n");
162 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");
163 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");
164 m->mothurOut("The distance parameter allows you to create a distance file from the results. The default is false.\n");
165 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");
166 m->mothurOut("The processors parameter allows you to specify the number of processors to use. The default is 1.\n");
167 m->mothurOut("The unifrac.weighted command should be in the following format: unifrac.weighted(groups=yourGroups, iters=yourIters).\n");
168 m->mothurOut("Example unifrac.weighted(groups=A-B-C, iters=500).\n");
169 m->mothurOut("The default value for groups is all the groups in your groupfile, and iters is 1000.\n");
170 m->mothurOut("The unifrac.weighted command output two files: .weighted and .wsummary their descriptions are in the manual.\n");
171 m->mothurOut("Note: No spaces between parameter labels (i.e. groups), '=' and parameters (i.e.yourGroups).\n\n");
173 catch(exception& e) {
174 m->errorOut(e, "UnifracWeightedCommand", "help");
179 /***********************************************************/
180 int UnifracWeightedCommand::execute() {
183 if (abort == true) { if (calledHelp) { return 0; } return 2; }
185 int start = time(NULL);
187 //get weighted for users tree
188 userData.resize(numComp,0); //data[0] = weightedscore AB, data[1] = weightedscore AC...
189 randomData.resize(numComp,0); //data[0] = weightedscore AB, data[1] = weightedscore AC...
191 //get weighted scores for users trees
192 for (int i = 0; i < T.size(); i++) {
194 if (m->control_pressed) { outSum.close(); for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } return 0; }
197 rScores.resize(numComp); //data[0] = weightedscore AB, data[1] = weightedscore AC...
198 uScores.resize(numComp); //data[0] = weightedscore AB, data[1] = weightedscore AC...
201 output = new ColumnFile(outputDir + m->getSimpleName(globaldata->getTreeFile()) + toString(i+1) + ".weighted", itersString);
202 outputNames.push_back(outputDir + m->getSimpleName(globaldata->getTreeFile()) + toString(i+1) + ".weighted");
203 outputTypes["weighted"].push_back(outputDir + m->getSimpleName(globaldata->getTreeFile()) + toString(i+1) + ".weighted");
206 userData = weighted->getValues(T[i], processors, outputDir); //userData[0] = weightedscore
208 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; }
211 for (int s=0; s<numComp; s++) {
212 //add users score to vector of user scores
213 uScores[s].push_back(userData[s]);
215 //save users tree score for summary file
216 utreeScores.push_back(userData[s]);
221 //calculate number of comparisons i.e. with groups A,B,C = AB, AC, BC = 3;
222 vector< vector<string> > namesOfGroupCombos;
223 for (int a=0; a<numGroups; a++) {
224 for (int l = 0; l < a; l++) {
225 vector<string> groups; groups.push_back(globaldata->Groups[a]); groups.push_back(globaldata->Groups[l]);
226 namesOfGroupCombos.push_back(groups);
232 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
234 int numPairs = namesOfGroupCombos.size();
235 int numPairsPerProcessor = numPairs / processors;
237 for (int i = 0; i < processors; i++) {
238 int startPos = i * numPairsPerProcessor;
239 if(i == processors - 1){
240 numPairsPerProcessor = numPairs - i * numPairsPerProcessor;
242 lines.push_back(linePair(startPos, numPairsPerProcessor));
248 //get scores for random trees
249 for (int j = 0; j < iters; j++) {
251 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
253 driver(T[i], namesOfGroupCombos, 0, namesOfGroupCombos.size(), rScores);
255 createProcesses(T[i], namesOfGroupCombos, rScores);
258 driver(T[i], namesOfGroupCombos, 0, namesOfGroupCombos.size(), rScores);
261 if (m->control_pressed) { delete output; outSum.close(); for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } return 0; }
264 // m->mothurOut("Iter: " + toString(j+1)); m->mothurOutEndLine();
268 //find the signifigance of the score for summary file
269 for (int f = 0; f < numComp; f++) {
271 sort(rScores[f].begin(), rScores[f].end());
273 //the index of the score higher than yours is returned
274 //so if you have 1000 random trees the index returned is 100
275 //then there are 900 trees with a score greater then you.
276 //giving you a signifigance of 0.900
277 int index = findIndex(userData[f], f); if (index == -1) { m->mothurOut("error in UnifracWeightedCommand"); m->mothurOutEndLine(); exit(1); } //error code
279 //the signifigance is the number of trees with the users score or higher
280 WScoreSig.push_back((iters-index)/(float)iters);
283 //out << "Tree# " << i << endl;
284 calculateFreqsCumuls();
298 if (m->control_pressed) { outSum.close(); for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } return 0; }
302 if (phylip) { createPhylipFile(); }
304 //clear out users groups
305 globaldata->Groups.clear();
308 if (m->control_pressed) {
309 for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); }
313 m->mothurOut("It took " + toString(time(NULL) - start) + " secs to run unifrac.weighted."); m->mothurOutEndLine();
315 m->mothurOutEndLine();
316 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
317 for (int i = 0; i < outputNames.size(); i++) { m->mothurOut(outputNames[i]); m->mothurOutEndLine(); }
318 m->mothurOutEndLine();
323 catch(exception& e) {
324 m->errorOut(e, "UnifracWeightedCommand", "execute");
328 /**************************************************************************************************/
330 int UnifracWeightedCommand::createProcesses(Tree* t, vector< vector<string> > namesOfGroupCombos, vector< vector<double> >& scores) {
332 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
334 vector<int> processIDS;
338 //loop through and create all the processes you want
339 while (process != processors) {
343 processIDS.push_back(pid); //create map from line number to pid so you can append files in correct order later
346 driver(t, namesOfGroupCombos, lines[process].start, lines[process].num, scores);
348 //pass numSeqs to parent
350 string tempFile = outputDir + toString(getpid()) + ".weightedcommand.results.temp";
351 m->openOutputFile(tempFile, out);
352 for (int i = lines[process].start; i < (lines[process].start + lines[process].num); i++) { out << scores[i][(scores[i].size()-1)] << '\t'; } out << endl;
357 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
358 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
363 driver(t, namesOfGroupCombos, lines[0].start, lines[0].num, scores);
365 //force parent to wait until all the processes are done
366 for (int i=0;i<(processors-1);i++) {
367 int temp = processIDS[i];
371 //get data created by processes
372 for (int i=0;i<(processors-1);i++) {
375 string s = outputDir + toString(processIDS[i]) + ".weightedcommand.results.temp";
376 m->openInputFile(s, in);
379 for (int j = lines[(i+1)].start; j < (lines[(i+1)].start + lines[(i+1)].num); j++) { in >> tempScore; scores[j].push_back(tempScore); }
387 catch(exception& e) {
388 m->errorOut(e, "UnifracWeightedCommand", "createProcesses");
393 /**************************************************************************************************/
394 int UnifracWeightedCommand::driver(Tree* t, vector< vector<string> > namesOfGroupCombos, int start, int num, vector< vector<double> >& scores) {
396 Tree* randT = new Tree();
398 for (int h = start; h < (start+num); h++) {
400 if (m->control_pressed) { return 0; }
402 //initialize weighted score
403 string groupA = namesOfGroupCombos[h][0];
404 string groupB = namesOfGroupCombos[h][1];
409 //create a random tree with same topology as T[i], but different labels
410 randT->assembleRandomUnifracTree(groupA, groupB);
412 if (m->control_pressed) { delete randT; return 0; }
414 //get wscore of random tree
415 EstOutput randomData = weighted->getValues(randT, groupA, groupB);
417 if (m->control_pressed) { delete randT; return 0; }
420 scores[h].push_back(randomData[0]);
428 catch(exception& e) {
429 m->errorOut(e, "UnifracWeightedCommand", "driver");
433 /***********************************************************/
434 void UnifracWeightedCommand::printWeightedFile() {
438 tags.push_back("Score"); tags.push_back("RandFreq"); tags.push_back("RandCumul");
440 for(int a = 0; a < numComp; a++) {
441 output->initFile(groupComb[a], tags);
443 for (map<float,float>::iterator it = validScores.begin(); it != validScores.end(); it++) {
444 data.push_back(it->first); data.push_back(rScoreFreq[a][it->first]); data.push_back(rCumul[a][it->first]);
445 output->output(data);
451 catch(exception& e) {
452 m->errorOut(e, "UnifracWeightedCommand", "printWeightedFile");
458 /***********************************************************/
459 void UnifracWeightedCommand::printWSummaryFile() {
462 outSum << "Tree#" << '\t' << "Groups" << '\t' << "WScore" << '\t' << "WSig" << endl;
463 m->mothurOut("Tree#\tGroups\tWScore\tWSig"); m->mothurOutEndLine();
466 outSum.setf(ios::fixed, ios::floatfield); outSum.setf(ios::showpoint);
470 for (int i = 0; i < T.size(); i++) {
471 for (int j = 0; j < numComp; j++) {
473 if (WScoreSig[count] > (1/(float)iters)) {
474 outSum << setprecision(6) << i+1 << '\t' << groupComb[j] << '\t' << utreeScores[count] << '\t' << setprecision(itersString.length()) << WScoreSig[count] << endl;
475 cout << setprecision(6) << i+1 << '\t' << groupComb[j] << '\t' << utreeScores[count] << '\t' << setprecision(itersString.length()) << WScoreSig[count] << endl;
476 m->mothurOutJustToLog(toString(i+1) +"\t" + groupComb[j] +"\t" + toString(utreeScores[count]) +"\t" + toString(WScoreSig[count]) + "\n");
478 outSum << setprecision(6) << i+1 << '\t' << groupComb[j] << '\t' << utreeScores[count] << '\t' << setprecision(itersString.length()) << "<" << (1/float(iters)) << endl;
479 cout << setprecision(6) << i+1 << '\t' << groupComb[j] << '\t' << utreeScores[count] << '\t' << setprecision(itersString.length()) << "<" << (1/float(iters)) << endl;
480 m->mothurOutJustToLog(toString(i+1) +"\t" + groupComb[j] +"\t" + toString(utreeScores[count]) +"\t<" + toString((1/float(iters))) + "\n");
483 outSum << setprecision(6) << i+1 << '\t' << groupComb[j] << '\t' << utreeScores[count] << '\t' << "0.00" << endl;
484 cout << setprecision(6) << i+1 << '\t' << groupComb[j] << '\t' << utreeScores[count] << '\t' << "0.00" << endl;
485 m->mothurOutJustToLog(toString(i+1) +"\t" + groupComb[j] +"\t" + toString(utreeScores[count]) +"\t0.00\n");
492 catch(exception& e) {
493 m->errorOut(e, "UnifracWeightedCommand", "printWSummaryFile");
497 /***********************************************************/
498 void UnifracWeightedCommand::createPhylipFile() {
502 for (int i = 0; i < T.size(); i++) {
504 string phylipFileName;
505 if ((outputForm == "lt") || (outputForm == "square")) {
506 phylipFileName = outputDir + m->getSimpleName(globaldata->getTreeFile()) + toString(i+1) + ".weighted.phylip.dist";
507 outputNames.push_back(phylipFileName); outputTypes["phylip"].push_back(phylipFileName);
509 phylipFileName = outputDir + m->getSimpleName(globaldata->getTreeFile()) + toString(i+1) + ".weighted.column.dist";
510 outputNames.push_back(phylipFileName); outputTypes["column"].push_back(phylipFileName);
514 m->openOutputFile(phylipFileName, out);
516 if ((outputForm == "lt") || (outputForm == "square")) {
518 out << globaldata->Groups.size() << endl;
521 //make matrix with scores in it
522 vector< vector<float> > dists; dists.resize(globaldata->Groups.size());
523 for (int i = 0; i < globaldata->Groups.size(); i++) {
524 dists[i].resize(globaldata->Groups.size(), 0.0);
527 //flip it so you can print it
528 for (int r=0; r<globaldata->Groups.size(); r++) {
529 for (int l = 0; l < r; l++) {
530 dists[r][l] = utreeScores[count];
531 dists[l][r] = utreeScores[count];
537 for (int r=0; r<globaldata->Groups.size(); r++) {
539 string name = globaldata->Groups[r];
540 if (name.length() < 10) { //pad with spaces to make compatible
541 while (name.length() < 10) { name += " "; }
544 if (outputForm == "lt") {
548 for (int l = 0; l < r; l++) { out << dists[r][l] << '\t'; }
550 }else if (outputForm == "square") {
554 for (int l = 0; l < globaldata->Groups.size(); l++) { out << dists[r][l] << '\t'; }
558 for (int l = 0; l < r; l++) {
559 string otherName = globaldata->Groups[l];
560 if (otherName.length() < 10) { //pad with spaces to make compatible
561 while (otherName.length() < 10) { otherName += " "; }
564 out << name << '\t' << otherName << dists[r][l] << endl;
571 catch(exception& e) {
572 m->errorOut(e, "UnifracWeightedCommand", "createPhylipFile");
576 /***********************************************************/
577 int UnifracWeightedCommand::findIndex(float score, int index) {
579 for (int i = 0; i < rScores[index].size(); i++) {
580 if (rScores[index][i] >= score) { return i; }
582 return rScores[index].size();
584 catch(exception& e) {
585 m->errorOut(e, "UnifracWeightedCommand", "findIndex");
590 /***********************************************************/
592 void UnifracWeightedCommand::calculateFreqsCumuls() {
594 //clear out old tree values
596 rScoreFreq.resize(numComp);
598 rCumul.resize(numComp);
601 //calculate frequency
602 for (int f = 0; f < numComp; f++) {
603 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...
604 validScores[rScores[f][i]] = rScores[f][i];
605 map<float,float>::iterator it = rScoreFreq[f].find(rScores[f][i]);
606 if (it != rScoreFreq[f].end()) {
607 rScoreFreq[f][rScores[f][i]]++;
609 rScoreFreq[f][rScores[f][i]] = 1;
615 for(int a = 0; a < numComp; a++) {
616 float rcumul = 1.0000;
617 //this loop fills the cumulative maps and put 0.0000 in the score freq map to make it easier to print.
618 for (map<float,float>::iterator it = validScores.begin(); it != validScores.end(); it++) {
619 //make rscoreFreq map and rCumul
620 map<float,float>::iterator it2 = rScoreFreq[a].find(it->first);
621 rCumul[a][it->first] = rcumul;
622 //get percentage of random trees with that info
623 if (it2 != rScoreFreq[a].end()) { rScoreFreq[a][it->first] /= iters; rcumul-= it2->second; }
624 else { rScoreFreq[a][it->first] = 0.0000; } //no random trees with that score
629 catch(exception& e) {
630 m->errorOut(e, "UnifracWeightedCommand", "calculateFreqsCums");
635 /***********************************************************/