unifracweightedcommand.cpp

   1 /*
   2  *  unifracweightedcommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by Sarah Westcott on 2/9/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10 #include "unifracweightedcommand.h"
  11
  12 //**********************************************************************************************************************
  13 vector<string> UnifracWeightedCommand::getValidParameters(){
  14         try {
  15                 string Array[] =  {"groups","iters","distance","random","processors","outputdir","inputdir"};
  16                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  17                 return myArray;
  18         }
  19         catch(exception& e) {
  20                 m->errorOut(e, "UnifracWeightedCommand", "getValidParameters");
  21                 exit(1);
  22         }
  23 }
  24 //**********************************************************************************************************************
  25 UnifracWeightedCommand::UnifracWeightedCommand(){
  26         try {
  27                 abort = true;
  28                 //initialize outputTypes
  29                 vector<string> tempOutNames;
  30                 outputTypes["weighted"] = tempOutNames;
  31                 outputTypes["wsummary"] = tempOutNames;
  32                 outputTypes["phylip"] = tempOutNames;
  33                 outputTypes["column"] = tempOutNames;
  34         }
  35         catch(exception& e) {
  36                 m->errorOut(e, "UnifracWeightedCommand", "UnifracWeightedCommand");
  37                 exit(1);
  38         }
  39 }
  40 //**********************************************************************************************************************
  41 vector<string> UnifracWeightedCommand::getRequiredParameters(){
  42         try {
  43                 vector<string> myArray;
  44                 return myArray;
  45         }
  46         catch(exception& e) {
  47                 m->errorOut(e, "UnifracWeightedCommand", "getRequiredParameters");
  48                 exit(1);
  49         }
  50 }
  51 //**********************************************************************************************************************
  52 vector<string> UnifracWeightedCommand::getRequiredFiles(){
  53         try {
  54                 string Array[] =  {"tree","group"};
  55                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  56
  57                 return myArray;
  58         }
  59         catch(exception& e) {
  60                 m->errorOut(e, "UnifracWeightedCommand", "getRequiredFiles");
  61                 exit(1);
  62         }
  63 }
  64 /***********************************************************/
  65 UnifracWeightedCommand::UnifracWeightedCommand(string option) {
  66         try {
  67                 globaldata = GlobalData::getInstance();
  68                 abort = false;
  69                 Groups.clear();
  70
  71                 //allow user to run help
  72                 if(option == "help") { help(); abort = true; }
  73
  74                 else {
  75                         //valid paramters for this command
  76                         string Array[] =  {"groups","iters","distance","random","processors","outputdir","inputdir"};
  77                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  78
  79                         OptionParser parser(option);
  80                         map<string,string> parameters=parser.getParameters();
  81
  82                         ValidParameters validParameter;
  83
  84                         //check to make sure all parameters are valid for command
  85                         for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
  86                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  87                         }
  88
  89                         //initialize outputTypes
  90                         vector<string> tempOutNames;
  91                         outputTypes["weighted"] = tempOutNames;
  92                         outputTypes["wsummary"] = tempOutNames;
  93                         outputTypes["phylip"] = tempOutNames;
  94                         outputTypes["column"] = tempOutNames;
  95
  96                         if (globaldata->gTree.size() == 0) {//no trees were read
  97                                 m->mothurOut("You must execute the read.tree command, before you may execute the unifrac.weighted command."); m->mothurOutEndLine(); abort = true;  }
  98
  99                         //if the user changes the output directory command factory will send this info to us in the output parameter
 100                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 101                                 outputDir = "";
 102                                 outputDir += m->hasPath(globaldata->inputFileName); //if user entered a file with a path then preserve it
 103                         }
 104
 105                         //check for optional parameter and set defaults
 106                         // ...at some point should added some additional type checking...
 107                         groups = validParameter.validFile(parameters, "groups", false);
 108                         if (groups == "not found") { groups = ""; }
 109                         else {
 110                                 m->splitAtDash(groups, Groups);
 111                                 globaldata->Groups = Groups;
 112                         }
 113
 114                         itersString = validParameter.validFile(parameters, "iters", false);                     if (itersString == "not found") { itersString = "1000"; }
 115                         convert(itersString, iters);
 116
 117                         string temp = validParameter.validFile(parameters, "distance", false);
 118                         if (temp == "not found") { phylip = false; outputForm = ""; }
 119                         else{
 120                                 if ((temp == "lt") || (temp == "column") || (temp == "square")) {  phylip = true;  outputForm = temp; }
 121                                 else { m->mothurOut("Options for distance are: lt, square, or column. Using lt."); m->mothurOutEndLine(); phylip = true; outputForm = "lt"; }
 122                         }
 123
 124                         temp = validParameter.validFile(parameters, "random", false);                                   if (temp == "not found") { temp = "F"; }
 125                         random = m->isTrue(temp);
 126
 127                         temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found"){       temp = "1";                             }
 128                         convert(temp, processors);
 129
 130                         if (!random) {  iters = 0;  } //turn off random calcs
 131
 132
 133                         if (abort == false) {
 134                                 T = globaldata->gTree;
 135                                 tmap = globaldata->gTreemap;
 136                                 sumFile = outputDir + m->getSimpleName(globaldata->getTreeFile()) + ".wsummary";
 137                                 m->openOutputFile(sumFile, outSum);
 138                                 outputNames.push_back(sumFile);  outputTypes["wsummary"].push_back(sumFile);
 139
 140                                 util = new SharedUtil();
 141                                 string s; //to make work with setgroups
 142                                 util->setGroups(globaldata->Groups, tmap->namesOfGroups, s, numGroups, "weighted");     //sets the groups the user wants to analyze
 143                                 util->getCombos(groupComb, globaldata->Groups, numComp);
 144
 145                                 weighted = new Weighted(tmap);
 146
 147                         }
 148                 }
 149
 150
 151         }
 152         catch(exception& e) {
 153                 m->errorOut(e, "UnifracWeightedCommand", "UnifracWeightedCommand");
 154                 exit(1);
 155         }
 156 }
 157 //**********************************************************************************************************************
 158
 159 void UnifracWeightedCommand::help(){
 160         try {
 161                 m->mothurOut("The unifrac.weighted command can only be executed after a successful read.tree command.\n");
 162                 m->mothurOut("The unifrac.weighted command parameters are groups, iters, distance, processors and random.  No parameters are required.\n");
 163                 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");
 164                 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");
 165                 m->mothurOut("The distance parameter allows you to create a distance file from the results. The default is false.\n");
 166                 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");
 167                 m->mothurOut("The processors parameter allows you to specify the number of processors to use. The default is 1.\n");
 168                 m->mothurOut("The unifrac.weighted command should be in the following format: unifrac.weighted(groups=yourGroups, iters=yourIters).\n");
 169                 m->mothurOut("Example unifrac.weighted(groups=A-B-C, iters=500).\n");
 170                 m->mothurOut("The default value for groups is all the groups in your groupfile, and iters is 1000.\n");
 171                 m->mothurOut("The unifrac.weighted command output two files: .weighted and .wsummary their descriptions are in the manual.\n");
 172                 m->mothurOut("Note: No spaces between parameter labels (i.e. groups), '=' and parameters (i.e.yourGroups).\n\n");
 173         }
 174         catch(exception& e) {
 175                 m->errorOut(e, "UnifracWeightedCommand", "help");
 176                 exit(1);
 177         }
 178 }
 179
 180 /***********************************************************/
 181 int UnifracWeightedCommand::execute() {
 182         try {
 183
 184                 if (abort == true) { return 0; }
 185
 186                 int start = time(NULL);
 187
 188                 //get weighted for users tree
 189                 userData.resize(numComp,0);  //data[0] = weightedscore AB, data[1] = weightedscore AC...
 190                 randomData.resize(numComp,0); //data[0] = weightedscore AB, data[1] = weightedscore AC...
 191
 192                 //get weighted scores for users trees
 193                 for (int i = 0; i < T.size(); i++) {
 194
 195                         if (m->control_pressed) { outSum.close(); for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str());  } return 0; }
 196
 197                         counter = 0;
 198                         rScores.resize(numComp);  //data[0] = weightedscore AB, data[1] = weightedscore AC...
 199                         uScores.resize(numComp);  //data[0] = weightedscore AB, data[1] = weightedscore AC...
 200
 201                         if (random) {
 202                                 output = new ColumnFile(outputDir + m->getSimpleName(globaldata->getTreeFile())  + toString(i+1) + ".weighted", itersString);
 203                                 outputNames.push_back(outputDir + m->getSimpleName(globaldata->getTreeFile())  + toString(i+1) + ".weighted");
 204                                 outputTypes["weighted"].push_back(outputDir + m->getSimpleName(globaldata->getTreeFile())  + toString(i+1) + ".weighted");
 205                         }
 206
 207                         userData = weighted->getValues(T[i], processors, outputDir);  //userData[0] = weightedscore
 208
 209                         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; }
 210
 211                         //save users score
 212                         for (int s=0; s<numComp; s++) {
 213                                 //add users score to vector of user scores
 214                                 uScores[s].push_back(userData[s]);
 215
 216                                 //save users tree score for summary file
 217                                 utreeScores.push_back(userData[s]);
 218                         }
 219
 220                         if (random) {
 221
 222                                 //calculate number of comparisons i.e. with groups A,B,C = AB, AC, BC = 3;
 223                                 vector< vector<string> > namesOfGroupCombos;
 224                                 for (int a=0; a<numGroups; a++) {
 225                                         for (int l = 0; l < a; l++) {
 226                                                 vector<string> groups; groups.push_back(globaldata->Groups[a]); groups.push_back(globaldata->Groups[l]);
 227                                                 namesOfGroupCombos.push_back(groups);
 228                                         }
 229                                 }
 230
 231                                 lines.clear();
 232
 233                                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 234                                         if(processors != 1){
 235                                                 int numPairs = namesOfGroupCombos.size();
 236                                                 int numPairsPerProcessor = numPairs / processors;
 237
 238                                                 for (int i = 0; i < processors; i++) {
 239                                                         int startPos = i * numPairsPerProcessor;
 240                                                         if(i == processors - 1){
 241                                                                 numPairsPerProcessor = numPairs - i * numPairsPerProcessor;
 242                                                         }
 243                                                         lines.push_back(linePair(startPos, numPairsPerProcessor));
 244                                                 }
 245                                         }
 246                                 #endif
 247
 248
 249                                 //get scores for random trees
 250                                 for (int j = 0; j < iters; j++) {
 251
 252                                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 253                                                 if(processors == 1){
 254                                                         driver(T[i],  namesOfGroupCombos, 0, namesOfGroupCombos.size(),  rScores);
 255                                                 }else{
 256                                                         createProcesses(T[i],  namesOfGroupCombos, rScores);
 257                                                 }
 258                                         #else
 259                                                 driver(T[i], namesOfGroupCombos, 0, namesOfGroupCombos.size(), rScores);
 260                                         #endif
 261
 262                                         if (m->control_pressed) { delete output; outSum.close(); for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str());  } return 0; }
 263
 264                                         //report progress
 265                                         m->mothurOut("Iter: " + toString(j+1)); m->mothurOutEndLine();
 266                                 }
 267                                 lines.clear();
 268
 269                                 //find the signifigance of the score for summary file
 270                                 for (int f = 0; f < numComp; f++) {
 271                                         //sort random scores
 272                                         sort(rScores[f].begin(), rScores[f].end());
 273
 274                                         //the index of the score higher than yours is returned
 275                                         //so if you have 1000 random trees the index returned is 100
 276                                         //then there are 900 trees with a score greater then you.
 277                                         //giving you a signifigance of 0.900
 278                                         int index = findIndex(userData[f], f);    if (index == -1) { m->mothurOut("error in UnifracWeightedCommand"); m->mothurOutEndLine(); exit(1); } //error code
 279
 280                                         //the signifigance is the number of trees with the users score or higher
 281                                         WScoreSig.push_back((iters-index)/(float)iters);
 282                                 }
 283
 284                                 //out << "Tree# " << i << endl;
 285                                 calculateFreqsCumuls();
 286                                 printWeightedFile();
 287
 288                                 delete output;
 289
 290                         }
 291
 292                         //clear data
 293                         rScores.clear();
 294                         uScores.clear();
 295                         validScores.clear();
 296                 }
 297
 298
 299                 if (m->control_pressed) { outSum.close(); for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str());  } return 0;  }
 300
 301                 printWSummaryFile();
 302
 303                 if (phylip) {   createPhylipFile();             }
 304
 305                 //clear out users groups
 306                 globaldata->Groups.clear();
 307
 308
 309                 if (m->control_pressed) {
 310                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());  }
 311                         return 0;
 312                 }
 313
 314                 m->mothurOut("It took " + toString(time(NULL) - start) + " secs to run unifrac.weighted."); m->mothurOutEndLine();
 315
 316                 m->mothurOutEndLine();
 317                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 318                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 319                 m->mothurOutEndLine();
 320
 321                 return 0;
 322
 323         }
 324         catch(exception& e) {
 325                 m->errorOut(e, "UnifracWeightedCommand", "execute");
 326                 exit(1);
 327         }
 328 }
 329 /**************************************************************************************************/
 330
 331 int UnifracWeightedCommand::createProcesses(Tree* t, vector< vector<string> > namesOfGroupCombos, vector< vector<double> >& scores) {
 332         try {
 333 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 334                 int process = 1;
 335                 vector<int> processIDS;
 336
 337                 EstOutput results;
 338
 339                 //loop through and create all the processes you want
 340                 while (process != processors) {
 341                         int pid = fork();
 342
 343                         if (pid > 0) {
 344                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 345                                 process++;
 346                         }else if (pid == 0){
 347                                 driver(t, namesOfGroupCombos, lines[process].start, lines[process].num, scores);
 348
 349                                 //pass numSeqs to parent
 350                                 ofstream out;
 351                                 string tempFile = outputDir + toString(getpid()) + ".weightedcommand.results.temp";
 352                                 m->openOutputFile(tempFile, out);
 353                                 for (int i = lines[process].start; i < (lines[process].start + lines[process].num); i++) { out << scores[i][(scores[i].size()-1)] << '\t';  } out << endl;
 354                                 out.close();
 355
 356                                 exit(0);
 357                         }else {
 358                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 359                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 360                                 exit(0);
 361                         }
 362                 }
 363
 364                 driver(t, namesOfGroupCombos, lines[0].start, lines[0].num, scores);
 365
 366                 //force parent to wait until all the processes are done
 367                 for (int i=0;i<(processors-1);i++) {
 368                         int temp = processIDS[i];
 369                         wait(&temp);
 370                 }
 371
 372                 //get data created by processes
 373                 for (int i=0;i<(processors-1);i++) {
 374
 375                         ifstream in;
 376                         string s = outputDir + toString(processIDS[i]) + ".weightedcommand.results.temp";
 377                         m->openInputFile(s, in);
 378
 379                         double tempScore;
 380                         for (int j = lines[(i+1)].start; j < (lines[(i+1)].start + lines[(i+1)].num); j++) { in >> tempScore; scores[j].push_back(tempScore); }
 381                         in.close();
 382                         remove(s.c_str());
 383                 }
 384
 385                 return 0;
 386 #endif
 387         }
 388         catch(exception& e) {
 389                 m->errorOut(e, "UnifracWeightedCommand", "createProcesses");
 390                 exit(1);
 391         }
 392 }
 393
 394 /**************************************************************************************************/
 395 int UnifracWeightedCommand::driver(Tree* t, vector< vector<string> > namesOfGroupCombos, int start, int num, vector< vector<double> >& scores) {
 396  try {
 397                 Tree* randT = new Tree();
 398
 399                 for (int h = start; h < (start+num); h++) {
 400
 401                         if (m->control_pressed) { return 0; }
 402
 403                         //initialize weighted score
 404                         string groupA = namesOfGroupCombos[h][0];
 405                         string groupB = namesOfGroupCombos[h][1];
 406
 407                         //copy T[i]'s info.
 408                         randT->getCopy(t);
 409
 410                         //create a random tree with same topology as T[i], but different labels
 411                         randT->assembleRandomUnifracTree(groupA, groupB);
 412
 413                         if (m->control_pressed) { delete randT;  return 0;  }
 414
 415                         //get wscore of random tree
 416                         EstOutput randomData = weighted->getValues(randT, groupA, groupB);
 417
 418                         if (m->control_pressed) { delete randT;  return 0;  }
 419
 420                         //save scores
 421                         scores[h].push_back(randomData[0]);
 422                 }
 423
 424                 delete randT;
 425
 426                 return 0;
 427
 428         }
 429         catch(exception& e) {
 430                 m->errorOut(e, "UnifracWeightedCommand", "driver");
 431                 exit(1);
 432         }
 433 }
 434 /***********************************************************/
 435 void UnifracWeightedCommand::printWeightedFile() {
 436         try {
 437                 vector<double> data;
 438                 vector<string> tags;
 439                 tags.push_back("Score"); tags.push_back("RandFreq"); tags.push_back("RandCumul");
 440
 441                 for(int a = 0; a < numComp; a++) {
 442                         output->initFile(groupComb[a], tags);
 443                         //print each line
 444                         for (map<float,float>::iterator it = validScores.begin(); it != validScores.end(); it++) {
 445                                 data.push_back(it->first);  data.push_back(rScoreFreq[a][it->first]); data.push_back(rCumul[a][it->first]);
 446                                 output->output(data);
 447                                 data.clear();
 448                         }
 449                         output->resetFile();
 450                 }
 451         }
 452         catch(exception& e) {
 453                 m->errorOut(e, "UnifracWeightedCommand", "printWeightedFile");
 454                 exit(1);
 455         }
 456 }
 457
 458
 459 /***********************************************************/
 460 void UnifracWeightedCommand::printWSummaryFile() {
 461         try {
 462                 //column headers
 463                 outSum << "Tree#" << '\t' << "Groups" << '\t' << "WScore" << '\t' << "WSig" <<  endl;
 464                 m->mothurOut("Tree#\tGroups\tWScore\tWSig"); m->mothurOutEndLine();
 465
 466                 //format output
 467                 outSum.setf(ios::fixed, ios::floatfield); outSum.setf(ios::showpoint);
 468
 469                 //print each line
 470                 int count = 0;
 471                 for (int i = 0; i < T.size(); i++) {
 472                         for (int j = 0; j < numComp; j++) {
 473                                 if (random) {
 474                                         if (WScoreSig[count] > (1/(float)iters)) {
 475                                                 outSum << setprecision(6) << i+1 << '\t' << groupComb[j] << '\t' << utreeScores[count] << '\t' << setprecision(itersString.length()) << WScoreSig[count] << endl;
 476                                                 cout << setprecision(6) << i+1 << '\t' << groupComb[j] << '\t' << utreeScores[count] << '\t' << setprecision(itersString.length()) << WScoreSig[count] << endl;
 477                                                 m->mothurOutJustToLog(toString(i+1) +"\t" + groupComb[j] +"\t" + toString(utreeScores[count]) +"\t" +  toString(WScoreSig[count]) + "\n");
 478                                         }else{
 479                                                 outSum << setprecision(6) << i+1 << '\t' << groupComb[j] << '\t' << utreeScores[count] << '\t' << setprecision(itersString.length()) << "<" << (1/float(iters)) << endl;
 480                                                 cout << setprecision(6) << i+1 << '\t' << groupComb[j] << '\t' << utreeScores[count] << '\t' << setprecision(itersString.length()) << "<" << (1/float(iters)) << endl;
 481                                                 m->mothurOutJustToLog(toString(i+1) +"\t" + groupComb[j] +"\t" + toString(utreeScores[count]) +"\t<" +  toString((1/float(iters))) + "\n");
 482                                         }
 483                                 }else{
 484                                         outSum << setprecision(6) << i+1 << '\t' << groupComb[j] << '\t' << utreeScores[count] << '\t' << "0.00" << endl;
 485                                         cout << setprecision(6) << i+1 << '\t' << groupComb[j] << '\t' << utreeScores[count] << '\t' << "0.00" << endl;
 486                                         m->mothurOutJustToLog(toString(i+1) +"\t" + groupComb[j] +"\t" + toString(utreeScores[count]) +"\t0.00\n");
 487                                 }
 488                                 count++;
 489                         }
 490                 }
 491                 outSum.close();
 492         }
 493         catch(exception& e) {
 494                 m->errorOut(e, "UnifracWeightedCommand", "printWSummaryFile");
 495                 exit(1);
 496         }
 497 }
 498 /***********************************************************/
 499 void UnifracWeightedCommand::createPhylipFile() {
 500         try {
 501                 int count = 0;
 502                 //for each tree
 503                 for (int i = 0; i < T.size(); i++) {
 504
 505                         string phylipFileName;
 506                         if ((outputForm == "lt") || (outputForm == "square")) {
 507                                 phylipFileName = outputDir + m->getSimpleName(globaldata->getTreeFile())  + toString(i+1) + ".weighted.phylip.dist";
 508                                 outputNames.push_back(phylipFileName); outputTypes["phylip"].push_back(phylipFileName);
 509                         }else { //column
 510                                 phylipFileName = outputDir + m->getSimpleName(globaldata->getTreeFile())  + toString(i+1) + ".weighted.column.dist";
 511                                 outputNames.push_back(phylipFileName); outputTypes["column"].push_back(phylipFileName);
 512                         }
 513
 514                         ofstream out;
 515                         m->openOutputFile(phylipFileName, out);
 516
 517                         if ((outputForm == "lt") || (outputForm == "square")) {
 518                                 //output numSeqs
 519                                 out << globaldata->Groups.size() << endl;
 520                         }
 521
 522                         //make matrix with scores in it
 523                         vector< vector<float> > dists;  dists.resize(globaldata->Groups.size());
 524                         for (int i = 0; i < globaldata->Groups.size(); i++) {
 525                                 dists[i].resize(globaldata->Groups.size(), 0.0);
 526                         }
 527
 528                         //flip it so you can print it
 529                         for (int r=0; r<globaldata->Groups.size(); r++) {
 530                                 for (int l = r+1; l < globaldata->Groups.size(); l++) {
 531                                         dists[r][l] = utreeScores[count];
 532                                         dists[l][r] = utreeScores[count];
 533                                         count++;
 534                                 }
 535                         }
 536
 537                         //output to file
 538                         for (int r=0; r<globaldata->Groups.size(); r++) {
 539                                 //output name
 540                                 string name = globaldata->Groups[r];
 541                                 if (name.length() < 10) { //pad with spaces to make compatible
 542                                         while (name.length() < 10) {  name += " ";  }
 543                                 }
 544
 545                                 if (outputForm == "lt") {
 546                                         out << name << '\t';
 547
 548                                         //output distances
 549                                         for (int l = 0; l < r; l++) {   out  << dists[r][l] << '\t';  }
 550                                         out << endl;
 551                                 }else if (outputForm == "square") {
 552                                         out << name << '\t';
 553
 554                                         //output distances
 555                                         for (int l = 0; l < globaldata->Groups.size(); l++) {   out  << dists[r][l] << '\t';  }
 556                                         out << endl;
 557                                 }else{
 558                                         //output distances
 559                                         for (int l = 0; l < r; l++) {
 560                                                 string otherName = globaldata->Groups[l];
 561                                                 if (otherName.length() < 10) { //pad with spaces to make compatible
 562                                                         while (otherName.length() < 10) {  otherName += " ";  }
 563                                                 }
 564
 565                                                 out  << name << '\t' << otherName << dists[r][l] << endl;
 566                                         }
 567                                 }
 568                         }
 569                         out.close();
 570                 }
 571         }
 572         catch(exception& e) {
 573                 m->errorOut(e, "UnifracWeightedCommand", "createPhylipFile");
 574                 exit(1);
 575         }
 576 }
 577 /***********************************************************/
 578 int UnifracWeightedCommand::findIndex(float score, int index) {
 579         try{
 580                 for (int i = 0; i < rScores[index].size(); i++) {
 581                         if (rScores[index][i] >= score) {       return i;       }
 582                 }
 583                 return rScores[index].size();
 584         }
 585         catch(exception& e) {
 586                 m->errorOut(e, "UnifracWeightedCommand", "findIndex");
 587                 exit(1);
 588         }
 589 }
 590
 591 /***********************************************************/
 592
 593 void UnifracWeightedCommand::calculateFreqsCumuls() {
 594         try {
 595                 //clear out old tree values
 596                 rScoreFreq.clear();
 597                 rScoreFreq.resize(numComp);
 598                 rCumul.clear();
 599                 rCumul.resize(numComp);
 600                 validScores.clear();
 601
 602                 //calculate frequency
 603                 for (int f = 0; f < numComp; f++) {
 604                         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...
 605                                 validScores[rScores[f][i]] = rScores[f][i];
 606                                 map<float,float>::iterator it = rScoreFreq[f].find(rScores[f][i]);
 607                                 if (it != rScoreFreq[f].end()) {
 608                                         rScoreFreq[f][rScores[f][i]]++;
 609                                 }else{
 610                                         rScoreFreq[f][rScores[f][i]] = 1;
 611                                 }
 612                         }
 613                 }
 614
 615                 //calculate rcumul
 616                 for(int a = 0; a < numComp; a++) {
 617                         float rcumul = 1.0000;
 618                         //this loop fills the cumulative maps and put 0.0000 in the score freq map to make it easier to print.
 619                         for (map<float,float>::iterator it = validScores.begin(); it != validScores.end(); it++) {
 620                                 //make rscoreFreq map and rCumul
 621                                 map<float,float>::iterator it2 = rScoreFreq[a].find(it->first);
 622                                 rCumul[a][it->first] = rcumul;
 623                                 //get percentage of random trees with that info
 624                                 if (it2 != rScoreFreq[a].end()) {  rScoreFreq[a][it->first] /= iters; rcumul-= it2->second;  }
 625                                 else { rScoreFreq[a][it->first] = 0.0000; } //no random trees with that score
 626                         }
 627                 }
 628
 629         }
 630         catch(exception& e) {
 631                 m->errorOut(e, "UnifracWeightedCommand", "calculateFreqsCums");
 632                 exit(1);
 633         }
 634 }
 635
 636 /***********************************************************/
 637
 638
 639
 640
 641