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                 if (numComp < processors) { processors = numComp;  }
 202
 203                 outSum << "Tree#" << '\t' << "Groups" << '\t'  <<  "UWScore" <<'\t' << "UWSig" <<  endl;
 204                 m->mothurOut("Tree#\tGroups\tUWScore\tUWSig"); m->mothurOutEndLine();
 205
 206                 //get pscores for users trees
 207                 for (int i = 0; i < T.size(); i++) {
 208                         if (m->control_pressed) {
 209                                 outSum.close();
 210                                 for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());  }
 211                                 return 0;
 212                         }
 213
 214                         counter = 0;
 215
 216                         if (random)  {
 217                                 output = new ColumnFile(outputDir + m->getSimpleName(globaldata->getTreeFile())  + toString(i+1) + ".unweighted", itersString);
 218                                 outputNames.push_back(outputDir + m->getSimpleName(globaldata->getTreeFile())  + toString(i+1) + ".unweighted");
 219                                 outputTypes["unweighted"].push_back(outputDir + m->getSimpleName(globaldata->getTreeFile())  + toString(i+1) + ".unweighted");
 220                         }
 221
 222
 223                         //get unweighted for users tree
 224                         rscoreFreq.resize(numComp);
 225                         rCumul.resize(numComp);
 226                         utreeScores.resize(numComp);
 227                         UWScoreSig.resize(numComp);
 228
 229                         userData = unweighted->getValues(T[i], processors, outputDir);  //userData[0] = unweightedscore
 230
 231                         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; }
 232
 233                         //output scores for each combination
 234                         for(int k = 0; k < numComp; k++) {
 235                                 //saves users score
 236                                 utreeScores[k].push_back(userData[k]);
 237
 238                                 //add users score to validscores
 239                                 validScores[userData[k]] = userData[k];
 240                         }
 241
 242                         //get unweighted scores for random trees - if random is false iters = 0
 243                         for (int j = 0; j < iters; j++) {
 244
 245                                 //we need a different getValues because when we swap the labels we only want to swap those in each pairwise comparison
 246                                 randomData = unweighted->getValues(T[i], "", "", processors, outputDir);
 247
 248                                 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; }
 249
 250                                 for(int k = 0; k < numComp; k++) {
 251                                         //add trees unweighted score to map of scores
 252                                         map<float,float>::iterator it = rscoreFreq[k].find(randomData[k]);
 253                                         if (it != rscoreFreq[k].end()) {//already have that score
 254                                                 rscoreFreq[k][randomData[k]]++;
 255                                         }else{//first time we have seen this score
 256                                                 rscoreFreq[k][randomData[k]] = 1;
 257                                         }
 258
 259                                         //add randoms score to validscores
 260                                         validScores[randomData[k]] = randomData[k];
 261                                 }
 262
 263                                 //report progress
 264 //                              m->mothurOut("Iter: " + toString(j+1)); m->mothurOutEndLine();
 265                         }
 266
 267                         for(int a = 0; a < numComp; a++) {
 268                                 float rcumul = 1.0000;
 269
 270                                 if (random) {
 271                                         //this loop fills the cumulative maps and put 0.0000 in the score freq map to make it easier to print.
 272                                         for (map<float,float>::iterator it = validScores.begin(); it != validScores.end(); it++) {
 273                                                 //make rscoreFreq map and rCumul
 274                                                 map<float,float>::iterator it2 = rscoreFreq[a].find(it->first);
 275                                                 rCumul[a][it->first] = rcumul;
 276                                                 //get percentage of random trees with that info
 277                                                 if (it2 != rscoreFreq[a].end()) {  rscoreFreq[a][it->first] /= iters; rcumul-= it2->second;  }
 278                                                 else { rscoreFreq[a][it->first] = 0.0000; } //no random trees with that score
 279                                         }
 280                                         UWScoreSig[a].push_back(rCumul[a][userData[a]]);
 281                                 }else           {       UWScoreSig[a].push_back(0.0);                                           }
 282
 283                         }
 284
 285                         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;  }
 286
 287                         //print output files
 288                         printUWSummaryFile(i);
 289                         if (random)  {  printUnweightedFile();  delete output;  }
 290                         if (phylip) {   createPhylipFile(i);            }
 291
 292                         rscoreFreq.clear();
 293                         rCumul.clear();
 294                         validScores.clear();
 295                         utreeScores.clear();
 296                         UWScoreSig.clear();
 297                 }
 298
 299
 300                 outSum.close();
 301
 302                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str());  }      return 0; }
 303
 304                 m->mothurOut("It took " + toString(time(NULL) - start) + " secs to run unifrac.unweighted."); m->mothurOutEndLine();
 305
 306                 m->mothurOutEndLine();
 307                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 308                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 309                 m->mothurOutEndLine();
 310
 311                 return 0;
 312
 313         }
 314         catch(exception& e) {
 315                 m->errorOut(e, "UnifracUnweightedCommand", "execute");
 316                 exit(1);
 317         }
 318 }
 319 /***********************************************************/
 320 void UnifracUnweightedCommand::printUnweightedFile() {
 321         try {
 322                 vector<double> data;
 323                 vector<string> tags;
 324
 325                 tags.push_back("Score");
 326                 tags.push_back("RandFreq"); tags.push_back("RandCumul");
 327
 328                 for(int a = 0; a < numComp; a++) {
 329                         output->initFile(groupComb[a], tags);
 330                         //print each line
 331                         for (map<float,float>::iterator it = validScores.begin(); it != validScores.end(); it++) {
 332                                 data.push_back(it->first);  data.push_back(rscoreFreq[a][it->first]); data.push_back(rCumul[a][it->first]);
 333                                 output->output(data);
 334                                 data.clear();
 335                         }
 336                         output->resetFile();
 337                 }
 338         }
 339         catch(exception& e) {
 340                 m->errorOut(e, "UnifracUnweightedCommand", "printUnweightedFile");
 341                 exit(1);
 342         }
 343 }
 344
 345 /***********************************************************/
 346 void UnifracUnweightedCommand::printUWSummaryFile(int i) {
 347         try {
 348
 349                 //format output
 350                 outSum.setf(ios::fixed, ios::floatfield); outSum.setf(ios::showpoint);
 351
 352                 //print each line
 353
 354                 for(int a = 0; a < numComp; a++) {
 355                         outSum << i+1 << '\t';
 356                         m->mothurOut(toString(i+1) + "\t");
 357
 358                         if (random) {
 359                                 if (UWScoreSig[a][0] > (1/(float)iters)) {
 360                                         outSum << setprecision(6) << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << setprecision(itersString.length()) << UWScoreSig[a][0] << endl;
 361                                         cout << setprecision(6)  << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << setprecision(itersString.length()) << UWScoreSig[a][0] << endl;
 362                                         m->mothurOutJustToLog(groupComb[a]  + "\t" + toString(utreeScores[a][0])  + "\t" + toString(UWScoreSig[a][0])+ "\n");
 363                                 }else {
 364                                         outSum << setprecision(6) << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << setprecision(itersString.length()) << "<" << (1/float(iters)) << endl;
 365                                         cout << setprecision(6)  << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << setprecision(itersString.length()) << "<" << (1/float(iters)) << endl;
 366                                         m->mothurOutJustToLog(groupComb[a]  + "\t" + toString(utreeScores[a][0])  + "\t<" + toString((1/float(iters))) + "\n");
 367                                 }
 368                         }else{
 369                                 outSum << setprecision(6) << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << "0.00" << endl;
 370                                 cout << setprecision(6)  << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << "0.00" << endl;
 371                                 m->mothurOutJustToLog(groupComb[a]  + "\t" + toString(utreeScores[a][0])  + "\t0.00\n");
 372                         }
 373                 }
 374
 375         }
 376         catch(exception& e) {
 377                 m->errorOut(e, "UnifracUnweightedCommand", "printUWSummaryFile");
 378                 exit(1);
 379         }
 380 }
 381 /***********************************************************/
 382 void UnifracUnweightedCommand::createPhylipFile(int i) {
 383         try {
 384                 string phylipFileName;
 385                 if ((outputForm == "lt") || (outputForm == "square")) {
 386                         phylipFileName = outputDir + m->getSimpleName(globaldata->getTreeFile())  + toString(i+1) + ".unweighted.phylip.dist";
 387                         outputNames.push_back(phylipFileName); outputTypes["phylip"].push_back(phylipFileName);
 388                 }else { //column
 389                         phylipFileName = outputDir + m->getSimpleName(globaldata->getTreeFile())  + toString(i+1) + ".unweighted.column.dist";
 390                         outputNames.push_back(phylipFileName); outputTypes["column"].push_back(phylipFileName);
 391                 }
 392
 393                 ofstream out;
 394                 m->openOutputFile(phylipFileName, out);
 395
 396                 if ((outputForm == "lt") || (outputForm == "square")) {
 397                         //output numSeqs
 398                         out << globaldata->Groups.size() << endl;
 399                 }
 400
 401                 //make matrix with scores in it
 402                 vector< vector<float> > dists;  dists.resize(globaldata->Groups.size());
 403                 for (int i = 0; i < globaldata->Groups.size(); i++) {
 404                         dists[i].resize(globaldata->Groups.size(), 0.0);
 405                 }
 406
 407                 //flip it so you can print it
 408                 int count = 0;
 409                 for (int r=0; r<globaldata->Groups.size(); r++) {
 410                         for (int l = 0; l < r; l++) {
 411                                 dists[r][l] = utreeScores[count][0];
 412                                 dists[l][r] = utreeScores[count][0];
 413                                 count++;
 414                         }
 415                 }
 416
 417                 //output to file
 418                 for (int r=0; r<globaldata->Groups.size(); r++) {
 419                         //output name
 420                         string name = globaldata->Groups[r];
 421                         if (name.length() < 10) { //pad with spaces to make compatible
 422                                 while (name.length() < 10) {  name += " ";  }
 423                         }
 424
 425                         if (outputForm == "lt") {
 426                                 out << name << '\t';
 427
 428                                 //output distances
 429                                 for (int l = 0; l < r; l++) {   out  << dists[r][l] << '\t';  }
 430                                 out << endl;
 431                         }else if (outputForm == "square") {
 432                                 out << name << '\t';
 433
 434                                 //output distances
 435                                 for (int l = 0; l < globaldata->Groups.size(); l++) {   out << dists[r][l] << '\t';  }
 436                                 out << endl;
 437                         }else{
 438                                 //output distances
 439                                 for (int l = 0; l < r; l++) {
 440                                         string otherName = globaldata->Groups[l];
 441                                         if (otherName.length() < 10) { //pad with spaces to make compatible
 442                                                 while (otherName.length() < 10) {  otherName += " ";  }
 443                                         }
 444
 445                                         out  << name << '\t' << otherName << dists[r][l] << endl;
 446                                 }
 447                         }
 448                 }
 449                 out.close();
 450         }
 451         catch(exception& e) {
 452                 m->errorOut(e, "UnifracUnweightedCommand", "createPhylipFile");
 453                 exit(1);
 454         }
 455 }
 456 /***********************************************************/
 457
 458
 459