unifracunweightedcommand.cpp

   1 /*
   2  *  unifracunweightedcommand.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 "unifracunweightedcommand.h"
  11
  12 //**********************************************************************************************************************
  13 vector<string> UnifracUnweightedCommand::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, "UnifracUnweightedCommand", "getValidParameters");
  21                 exit(1);
  22         }
  23 }
  24 //**********************************************************************************************************************
  25 UnifracUnweightedCommand::UnifracUnweightedCommand(){
  26         try {
  27                 globaldata = GlobalData::getInstance();
  28                 abort = true; calledHelp = true;
  29                 vector<string> tempOutNames;
  30                 outputTypes["unweighted"] = tempOutNames;
  31                 outputTypes["uwsummary"] = tempOutNames;
  32                 outputTypes["phylip"] = tempOutNames;
  33                 outputTypes["column"] = tempOutNames;
  34         }
  35         catch(exception& e) {
  36                 m->errorOut(e, "UnifracUnweightedCommand", "UnifracUnweightedCommand");
  37                 exit(1);
  38         }
  39 }
  40 //**********************************************************************************************************************
  41 vector<string> UnifracUnweightedCommand::getRequiredParameters(){
  42         try {
  43                 vector<string> myArray;
  44                 return myArray;
  45         }
  46         catch(exception& e) {
  47                 m->errorOut(e, "UnifracUnweightedCommand", "getRequiredParameters");
  48                 exit(1);
  49         }
  50 }
  51 //**********************************************************************************************************************
  52 vector<string> UnifracUnweightedCommand::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, "UnifracUnweightedCommand", "getRequiredFiles");
  61                 exit(1);
  62         }
  63 }
  64 /***********************************************************/
  65 UnifracUnweightedCommand::UnifracUnweightedCommand(string option)  {
  66         try {
  67                 globaldata = GlobalData::getInstance();
  68                 abort = false; calledHelp = false;
  69                 Groups.clear();
  70
  71                 //allow user to run help
  72                 if(option == "help") { help(); abort = true; calledHelp = 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["unweighted"] = tempOutNames;
  92                         outputTypes["uwsummary"] = 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.unweighted 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                         //if user selects distance = true and no groups it won't calc the pairwise
 133                         if ((phylip) && (Groups.size() == 0)) {
 134                                 groups = "all";
 135                                 m->splitAtDash(groups, Groups);
 136                                 globaldata->Groups = Groups;
 137                         }
 138
 139                         if (abort == false) {
 140                                 T = globaldata->gTree;
 141                                 tmap = globaldata->gTreemap;
 142                                 sumFile = outputDir + m->getSimpleName(globaldata->getTreeFile()) + ".uwsummary";
 143                                 outputNames.push_back(sumFile); outputTypes["uwsummary"].push_back(sumFile);
 144                                 m->openOutputFile(sumFile, outSum);
 145
 146                                 util = new SharedUtil();
 147                                 util->setGroups(globaldata->Groups, tmap->namesOfGroups, allGroups, numGroups, "unweighted");   //sets the groups the user wants to analyze
 148                                 util->getCombos(groupComb, globaldata->Groups, numComp);
 149
 150                                 if (numGroups == 1) { numComp++; groupComb.push_back(allGroups); }
 151
 152                                 unweighted = new Unweighted(tmap);
 153
 154                         }
 155
 156                 }
 157
 158         }
 159         catch(exception& e) {
 160                 m->errorOut(e, "UnifracUnweightedCommand", "UnifracUnweightedCommand");
 161                 exit(1);
 162         }
 163 }
 164
 165 //**********************************************************************************************************************
 166
 167 void UnifracUnweightedCommand::help(){
 168         try {
 169                 m->mothurOut("The unifrac.unweighted command can only be executed after a successful read.tree command.\n");
 170                 m->mothurOut("The unifrac.unweighted command parameters are groups, iters, distance, processors and random.  No parameters are required.\n");
 171                 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 1 valid group.\n");
 172                 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");
 173                 m->mothurOut("The distance parameter allows you to create a distance file from the results. The default is false. You may set distance to lt, square or column.\n");
 174                 m->mothurOut("The random parameter allows you to shut off the comparison to random trees. The default is false, meaning compare don't your trees with randomly generated trees.\n");
 175                 m->mothurOut("The processors parameter allows you to specify the number of processors to use. The default is 1.\n");
 176                 m->mothurOut("The unifrac.unweighted command should be in the following format: unifrac.unweighted(groups=yourGroups, iters=yourIters).\n");
 177                 m->mothurOut("Example unifrac.unweighted(groups=A-B-C, iters=500).\n");
 178                 m->mothurOut("The default value for groups is all the groups in your groupfile, and iters is 1000.\n");
 179                 m->mothurOut("The unifrac.unweighted command output two files: .unweighted and .uwsummary their descriptions are in the manual.\n");
 180                 m->mothurOut("Note: No spaces between parameter labels (i.e. groups), '=' and parameters (i.e.yourGroups).\n\n");
 181         }
 182         catch(exception& e) {
 183                 m->errorOut(e, "UnifracUnweightedCommand", "help");
 184                 exit(1);
 185         }
 186 }
 187
 188
 189 /***********************************************************/
 190 int UnifracUnweightedCommand::execute() {
 191         try {
 192
 193                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 194
 195                 int start = time(NULL);
 196
 197                 userData.resize(numComp,0);  //data[0] = unweightedscore
 198                 randomData.resize(numComp,0); //data[0] = unweightedscore
 199                 //create new tree with same num nodes and leaves as users
 200
 201                 outSum << "Tree#" << '\t' << "Groups" << '\t'  <<  "UWScore" <<'\t' << "UWSig" <<  endl;
 202                 m->mothurOut("Tree#\tGroups\tUWScore\tUWSig"); m->mothurOutEndLine();
 203
 204                 //get pscores for users trees
 205                 for (int i = 0; i < T.size(); i++) {
 206                         if (m->control_pressed) {
 207                                 outSum.close();
 208                                 for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());  }
 209                                 return 0;
 210                         }
 211
 212                         counter = 0;
 213
 214                         if (random)  {
 215                                 output = new ColumnFile(outputDir + m->getSimpleName(globaldata->getTreeFile())  + toString(i+1) + ".unweighted", itersString);
 216                                 outputNames.push_back(outputDir + m->getSimpleName(globaldata->getTreeFile())  + toString(i+1) + ".unweighted");
 217                                 outputTypes["unweighted"].push_back(outputDir + m->getSimpleName(globaldata->getTreeFile())  + toString(i+1) + ".unweighted");
 218                         }
 219
 220
 221                         //get unweighted for users tree
 222                         rscoreFreq.resize(numComp);
 223                         rCumul.resize(numComp);
 224                         utreeScores.resize(numComp);
 225                         UWScoreSig.resize(numComp);
 226
 227                         userData = unweighted->getValues(T[i], processors, outputDir);  //userData[0] = unweightedscore
 228
 229                         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; }
 230
 231                         //output scores for each combination
 232                         for(int k = 0; k < numComp; k++) {
 233                                 //saves users score
 234                                 utreeScores[k].push_back(userData[k]);
 235
 236                                 //add users score to validscores
 237                                 validScores[userData[k]] = userData[k];
 238                         }
 239
 240                         //get unweighted scores for random trees - if random is false iters = 0
 241                         for (int j = 0; j < iters; j++) {
 242
 243                                 //we need a different getValues because when we swap the labels we only want to swap those in each pairwise comparison
 244                                 randomData = unweighted->getValues(T[i], "", "", processors, outputDir);
 245
 246                                 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; }
 247
 248                                 for(int k = 0; k < numComp; k++) {
 249                                         //add trees unweighted score to map of scores
 250                                         map<float,float>::iterator it = rscoreFreq[k].find(randomData[k]);
 251                                         if (it != rscoreFreq[k].end()) {//already have that score
 252                                                 rscoreFreq[k][randomData[k]]++;
 253                                         }else{//first time we have seen this score
 254                                                 rscoreFreq[k][randomData[k]] = 1;
 255                                         }
 256
 257                                         //add randoms score to validscores
 258                                         validScores[randomData[k]] = randomData[k];
 259                                 }
 260
 261                                 //report progress
 262                                 m->mothurOut("Iter: " + toString(j+1)); m->mothurOutEndLine();
 263                         }
 264
 265                         for(int a = 0; a < numComp; a++) {
 266                                 float rcumul = 1.0000;
 267
 268                                 if (random) {
 269                                         //this loop fills the cumulative maps and put 0.0000 in the score freq map to make it easier to print.
 270                                         for (map<float,float>::iterator it = validScores.begin(); it != validScores.end(); it++) {
 271                                                 //make rscoreFreq map and rCumul
 272                                                 map<float,float>::iterator it2 = rscoreFreq[a].find(it->first);
 273                                                 rCumul[a][it->first] = rcumul;
 274                                                 //get percentage of random trees with that info
 275                                                 if (it2 != rscoreFreq[a].end()) {  rscoreFreq[a][it->first] /= iters; rcumul-= it2->second;  }
 276                                                 else { rscoreFreq[a][it->first] = 0.0000; } //no random trees with that score
 277                                         }
 278                                         UWScoreSig[a].push_back(rCumul[a][userData[a]]);
 279                                 }else           {       UWScoreSig[a].push_back(0.0);                                           }
 280
 281                         }
 282
 283                         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;  }
 284
 285                         //print output files
 286                         printUWSummaryFile(i);
 287                         if (random)  {  printUnweightedFile();  delete output;  }
 288                         if (phylip) {   createPhylipFile(i);            }
 289
 290                         rscoreFreq.clear();
 291                         rCumul.clear();
 292                         validScores.clear();
 293                         utreeScores.clear();
 294                         UWScoreSig.clear();
 295                 }
 296
 297
 298                 outSum.close();
 299
 300                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str());  }      return 0; }
 301
 302                 m->mothurOut("It took " + toString(time(NULL) - start) + " secs to run unifrac.unweighted."); m->mothurOutEndLine();
 303
 304                 m->mothurOutEndLine();
 305                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 306                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 307                 m->mothurOutEndLine();
 308
 309                 return 0;
 310
 311         }
 312         catch(exception& e) {
 313                 m->errorOut(e, "UnifracUnweightedCommand", "execute");
 314                 exit(1);
 315         }
 316 }
 317 /***********************************************************/
 318 void UnifracUnweightedCommand::printUnweightedFile() {
 319         try {
 320                 vector<double> data;
 321                 vector<string> tags;
 322
 323                 tags.push_back("Score");
 324                 tags.push_back("RandFreq"); tags.push_back("RandCumul");
 325
 326                 for(int a = 0; a < numComp; a++) {
 327                         output->initFile(groupComb[a], tags);
 328                         //print each line
 329                         for (map<float,float>::iterator it = validScores.begin(); it != validScores.end(); it++) {
 330                                 data.push_back(it->first);  data.push_back(rscoreFreq[a][it->first]); data.push_back(rCumul[a][it->first]);
 331                                 output->output(data);
 332                                 data.clear();
 333                         }
 334                         output->resetFile();
 335                 }
 336         }
 337         catch(exception& e) {
 338                 m->errorOut(e, "UnifracUnweightedCommand", "printUnweightedFile");
 339                 exit(1);
 340         }
 341 }
 342
 343 /***********************************************************/
 344 void UnifracUnweightedCommand::printUWSummaryFile(int i) {
 345         try {
 346
 347                 //format output
 348                 outSum.setf(ios::fixed, ios::floatfield); outSum.setf(ios::showpoint);
 349
 350                 //print each line
 351
 352                 for(int a = 0; a < numComp; a++) {
 353                         outSum << i+1 << '\t';
 354                         m->mothurOut(toString(i+1) + "\t");
 355
 356                         if (random) {
 357                                 if (UWScoreSig[a][0] > (1/(float)iters)) {
 358                                         outSum << setprecision(6) << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << setprecision(itersString.length()) << UWScoreSig[a][0] << endl;
 359                                         cout << setprecision(6)  << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << setprecision(itersString.length()) << UWScoreSig[a][0] << endl;
 360                                         m->mothurOutJustToLog(groupComb[a]  + "\t" + toString(utreeScores[a][0])  + "\t" + toString(UWScoreSig[a][0])+ "\n");
 361                                 }else {
 362                                         outSum << setprecision(6) << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << setprecision(itersString.length()) << "<" << (1/float(iters)) << endl;
 363                                         cout << setprecision(6)  << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << setprecision(itersString.length()) << "<" << (1/float(iters)) << endl;
 364                                         m->mothurOutJustToLog(groupComb[a]  + "\t" + toString(utreeScores[a][0])  + "\t<" + toString((1/float(iters))) + "\n");
 365                                 }
 366                         }else{
 367                                 outSum << setprecision(6) << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << "0.00" << endl;
 368                                 cout << setprecision(6)  << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << "0.00" << endl;
 369                                 m->mothurOutJustToLog(groupComb[a]  + "\t" + toString(utreeScores[a][0])  + "\t0.00\n");
 370                         }
 371                 }
 372
 373         }
 374         catch(exception& e) {
 375                 m->errorOut(e, "UnifracUnweightedCommand", "printUWSummaryFile");
 376                 exit(1);
 377         }
 378 }
 379 /***********************************************************/
 380 void UnifracUnweightedCommand::createPhylipFile(int i) {
 381         try {
 382                 string phylipFileName;
 383                 if ((outputForm == "lt") || (outputForm == "square")) {
 384                         phylipFileName = outputDir + m->getSimpleName(globaldata->getTreeFile())  + toString(i+1) + ".unweighted.phylip.dist";
 385                         outputNames.push_back(phylipFileName); outputTypes["phylip"].push_back(phylipFileName);
 386                 }else { //column
 387                         phylipFileName = outputDir + m->getSimpleName(globaldata->getTreeFile())  + toString(i+1) + ".unweighted.column.dist";
 388                         outputNames.push_back(phylipFileName); outputTypes["column"].push_back(phylipFileName);
 389                 }
 390
 391                 ofstream out;
 392                 m->openOutputFile(phylipFileName, out);
 393
 394                 if ((outputForm == "lt") || (outputForm == "square")) {
 395                         //output numSeqs
 396                         out << globaldata->Groups.size() << endl;
 397                 }
 398
 399                 //make matrix with scores in it
 400                 vector< vector<float> > dists;  dists.resize(globaldata->Groups.size());
 401                 for (int i = 0; i < globaldata->Groups.size(); i++) {
 402                         dists[i].resize(globaldata->Groups.size(), 0.0);
 403                 }
 404
 405                 //flip it so you can print it
 406                 int count = 0;
 407                 for (int r=0; r<globaldata->Groups.size(); r++) {
 408                         for (int l = 0; l < r; l++) {
 409                                 dists[r][l] = utreeScores[count][0];
 410                                 dists[l][r] = utreeScores[count][0];
 411                                 count++;
 412                         }
 413                 }
 414
 415                 //output to file
 416                 for (int r=0; r<globaldata->Groups.size(); r++) {
 417                         //output name
 418                         string name = globaldata->Groups[r];
 419                         if (name.length() < 10) { //pad with spaces to make compatible
 420                                 while (name.length() < 10) {  name += " ";  }
 421                         }
 422
 423                         if (outputForm == "lt") {
 424                                 out << name << '\t';
 425
 426                                 //output distances
 427                                 for (int l = 0; l < r; l++) {   out  << dists[r][l] << '\t';  }
 428                                 out << endl;
 429                         }else if (outputForm == "square") {
 430                                 out << name << '\t';
 431
 432                                 //output distances
 433                                 for (int l = 0; l < globaldata->Groups.size(); l++) {   out  << dists[r][l] << '\t';  }
 434                                 out << endl;
 435                         }else{
 436                                 //output distances
 437                                 for (int l = 0; l < r; l++) {
 438                                         string otherName = globaldata->Groups[l];
 439                                         if (otherName.length() < 10) { //pad with spaces to make compatible
 440                                                 while (otherName.length() < 10) {  otherName += " ";  }
 441                                         }
 442
 443                                         out  << name << '\t' << otherName << dists[r][l] << endl;
 444                                 }
 445                         }
 446                 }
 447                 out.close();
 448         }
 449         catch(exception& e) {
 450                 m->errorOut(e, "UnifracUnweightedCommand", "createPhylipFile");
 451                 exit(1);
 452         }
 453 }
 454 /***********************************************************/
 455
 456
 457