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;
  29                 //initialize outputTypes
  30                 vector<string> tempOutNames;
  31                 outputTypes["unweighted"] = tempOutNames;
  32                 outputTypes["uwsummary"] = tempOutNames;
  33                 outputTypes["phylip"] = tempOutNames;
  34                 outputTypes["column"] = tempOutNames;
  35         }
  36         catch(exception& e) {
  37                 m->errorOut(e, "UnifracUnweightedCommand", "UnifracUnweightedCommand");
  38                 exit(1);
  39         }
  40 }
  41 //**********************************************************************************************************************
  42 vector<string> UnifracUnweightedCommand::getRequiredParameters(){
  43         try {
  44                 vector<string> myArray;
  45                 return myArray;
  46         }
  47         catch(exception& e) {
  48                 m->errorOut(e, "UnifracUnweightedCommand", "getRequiredParameters");
  49                 exit(1);
  50         }
  51 }
  52 //**********************************************************************************************************************
  53 vector<string> UnifracUnweightedCommand::getRequiredFiles(){
  54         try {
  55                 string Array[] =  {"tree","group"};
  56                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  57
  58                 return myArray;
  59         }
  60         catch(exception& e) {
  61                 m->errorOut(e, "UnifracUnweightedCommand", "getRequiredFiles");
  62                 exit(1);
  63         }
  64 }
  65 /***********************************************************/
  66 UnifracUnweightedCommand::UnifracUnweightedCommand(string option)  {
  67         try {
  68                 globaldata = GlobalData::getInstance();
  69                 abort = false;
  70                 Groups.clear();
  71
  72                 //allow user to run help
  73                 if(option == "help") { help(); abort = true; }
  74
  75                 else {
  76                         //valid paramters for this command
  77                         string Array[] =  {"groups","iters","distance","random", "processors","outputdir","inputdir"};
  78                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  79
  80                         OptionParser parser(option);
  81                         map<string,string> parameters = parser.getParameters();
  82
  83                         ValidParameters validParameter;
  84
  85                         //check to make sure all parameters are valid for command
  86                         for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
  87                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  88                         }
  89
  90                         //initialize outputTypes
  91                         vector<string> tempOutNames;
  92                         outputTypes["unweighted"] = tempOutNames;
  93                         outputTypes["uwsummary"] = tempOutNames;
  94                         outputTypes["phylip"] = tempOutNames;
  95                         outputTypes["column"] = tempOutNames;
  96
  97                         if (globaldata->gTree.size() == 0) {//no trees were read
  98                                 m->mothurOut("You must execute the read.tree command, before you may execute the unifrac.unweighted command."); m->mothurOutEndLine(); abort = true;  }
  99
 100                         //if the user changes the output directory command factory will send this info to us in the output parameter
 101                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 102                                 outputDir = "";
 103                                 outputDir += m->hasPath(globaldata->inputFileName); //if user entered a file with a path then preserve it
 104                         }
 105
 106                         //check for optional parameter and set defaults
 107                         // ...at some point should added some additional type checking...
 108                         groups = validParameter.validFile(parameters, "groups", false);
 109                         if (groups == "not found") { groups = ""; }
 110                         else {
 111                                 m->splitAtDash(groups, Groups);
 112                                 globaldata->Groups = Groups;
 113                         }
 114
 115                         itersString = validParameter.validFile(parameters, "iters", false);                             if (itersString == "not found") { itersString = "1000"; }
 116                         convert(itersString, iters);
 117
 118                         string temp = validParameter.validFile(parameters, "distance", false);
 119                         if (temp == "not found") { phylip = false; outputForm = ""; }
 120                         else{
 121                                 if ((temp == "lt") || (temp == "column") || (temp == "square")) {  phylip = true;  outputForm = temp; }
 122                                 else { m->mothurOut("Options for distance are: lt, square, or column. Using lt."); m->mothurOutEndLine(); phylip = true; outputForm = "lt"; }
 123                         }
 124
 125                         temp = validParameter.validFile(parameters, "random", false);                                   if (temp == "not found") { temp = "f"; }
 126                         random = m->isTrue(temp);
 127
 128                         temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found"){       temp = "1";                             }
 129                         convert(temp, processors);
 130
 131                         if (!random) {  iters = 0;  } //turn off random calcs
 132
 133                         //if user selects distance = true and no groups it won't calc the pairwise
 134                         if ((phylip) && (Groups.size() == 0)) {
 135                                 groups = "all";
 136                                 m->splitAtDash(groups, Groups);
 137                                 globaldata->Groups = Groups;
 138                         }
 139
 140                         if (abort == false) {
 141                                 T = globaldata->gTree;
 142                                 tmap = globaldata->gTreemap;
 143                                 sumFile = outputDir + m->getSimpleName(globaldata->getTreeFile()) + ".uwsummary";
 144                                 outputNames.push_back(sumFile); outputTypes["uwsummary"].push_back(sumFile);
 145                                 m->openOutputFile(sumFile, outSum);
 146
 147                                 util = new SharedUtil();
 148                                 util->setGroups(globaldata->Groups, tmap->namesOfGroups, allGroups, numGroups, "unweighted");   //sets the groups the user wants to analyze
 149                                 util->getCombos(groupComb, globaldata->Groups, numComp);
 150
 151                                 if (numGroups == 1) { numComp++; groupComb.push_back(allGroups); }
 152
 153                                 unweighted = new Unweighted(tmap);
 154
 155                         }
 156
 157                 }
 158
 159         }
 160         catch(exception& e) {
 161                 m->errorOut(e, "UnifracUnweightedCommand", "UnifracUnweightedCommand");
 162                 exit(1);
 163         }
 164 }
 165
 166 //**********************************************************************************************************************
 167
 168 void UnifracUnweightedCommand::help(){
 169         try {
 170                 m->mothurOut("The unifrac.unweighted command can only be executed after a successful read.tree command.\n");
 171                 m->mothurOut("The unifrac.unweighted command parameters are groups, iters, distance, processors and random.  No parameters are required.\n");
 172                 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");
 173                 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");
 174                 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");
 175                 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");
 176                 m->mothurOut("The processors parameter allows you to specify the number of processors to use. The default is 1.\n");
 177                 m->mothurOut("The unifrac.unweighted command should be in the following format: unifrac.unweighted(groups=yourGroups, iters=yourIters).\n");
 178                 m->mothurOut("Example unifrac.unweighted(groups=A-B-C, iters=500).\n");
 179                 m->mothurOut("The default value for groups is all the groups in your groupfile, and iters is 1000.\n");
 180                 m->mothurOut("The unifrac.unweighted command output two files: .unweighted and .uwsummary their descriptions are in the manual.\n");
 181                 m->mothurOut("Note: No spaces between parameter labels (i.e. groups), '=' and parameters (i.e.yourGroups).\n\n");
 182         }
 183         catch(exception& e) {
 184                 m->errorOut(e, "UnifracUnweightedCommand", "help");
 185                 exit(1);
 186         }
 187 }
 188
 189
 190 /***********************************************************/
 191 int UnifracUnweightedCommand::execute() {
 192         try {
 193
 194                 if (abort == true) { return 0; }
 195
 196                 int start = time(NULL);
 197
 198                 userData.resize(numComp,0);  //data[0] = unweightedscore
 199                 randomData.resize(numComp,0); //data[0] = unweightedscore
 200                 //create new tree with same num nodes and leaves as users
 201
 202                 outSum << "Tree#" << '\t' << "Groups" << '\t'  <<  "UWScore" <<'\t' << "UWSig" <<  endl;
 203                 m->mothurOut("Tree#\tGroups\tUWScore\tUWSig"); m->mothurOutEndLine();
 204
 205                 //get pscores for users trees
 206                 for (int i = 0; i < T.size(); i++) {
 207                         if (m->control_pressed) {
 208                                 outSum.close();
 209                                 for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());  }
 210                                 return 0;
 211                         }
 212
 213                         counter = 0;
 214
 215                         if (random)  {
 216                                 output = new ColumnFile(outputDir + m->getSimpleName(globaldata->getTreeFile())  + toString(i+1) + ".unweighted", itersString);
 217                                 outputNames.push_back(outputDir + m->getSimpleName(globaldata->getTreeFile())  + toString(i+1) + ".unweighted");
 218                                 outputTypes["unweighted"].push_back(outputDir + m->getSimpleName(globaldata->getTreeFile())  + toString(i+1) + ".unweighted");
 219                         }
 220
 221
 222                         //get unweighted for users tree
 223                         rscoreFreq.resize(numComp);
 224                         rCumul.resize(numComp);
 225                         utreeScores.resize(numComp);
 226                         UWScoreSig.resize(numComp);
 227
 228                         userData = unweighted->getValues(T[i], processors, outputDir);  //userData[0] = unweightedscore
 229
 230                         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; }
 231
 232                         //output scores for each combination
 233                         for(int k = 0; k < numComp; k++) {
 234                                 //saves users score
 235                                 utreeScores[k].push_back(userData[k]);
 236
 237                                 //add users score to validscores
 238                                 validScores[userData[k]] = userData[k];
 239                         }
 240
 241                         //get unweighted scores for random trees - if random is false iters = 0
 242                         for (int j = 0; j < iters; j++) {
 243
 244                                 //we need a different getValues because when we swap the labels we only want to swap those in each pairwise comparison
 245                                 randomData = unweighted->getValues(T[i], "", "", processors, outputDir);
 246
 247                                 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; }
 248
 249                                 for(int k = 0; k < numComp; k++) {
 250                                         //add trees unweighted score to map of scores
 251                                         map<float,float>::iterator it = rscoreFreq[k].find(randomData[k]);
 252                                         if (it != rscoreFreq[k].end()) {//already have that score
 253                                                 rscoreFreq[k][randomData[k]]++;
 254                                         }else{//first time we have seen this score
 255                                                 rscoreFreq[k][randomData[k]] = 1;
 256                                         }
 257
 258                                         //add randoms score to validscores
 259                                         validScores[randomData[k]] = randomData[k];
 260                                 }
 261
 262                                 //report progress
 263                                 m->mothurOut("Iter: " + toString(j+1)); m->mothurOutEndLine();
 264                         }
 265
 266                         for(int a = 0; a < numComp; a++) {
 267                                 float rcumul = 1.0000;
 268
 269                                 if (random) {
 270                                         //this loop fills the cumulative maps and put 0.0000 in the score freq map to make it easier to print.
 271                                         for (map<float,float>::iterator it = validScores.begin(); it != validScores.end(); it++) {
 272                                                 //make rscoreFreq map and rCumul
 273                                                 map<float,float>::iterator it2 = rscoreFreq[a].find(it->first);
 274                                                 rCumul[a][it->first] = rcumul;
 275                                                 //get percentage of random trees with that info
 276                                                 if (it2 != rscoreFreq[a].end()) {  rscoreFreq[a][it->first] /= iters; rcumul-= it2->second;  }
 277                                                 else { rscoreFreq[a][it->first] = 0.0000; } //no random trees with that score
 278                                         }
 279                                         UWScoreSig[a].push_back(rCumul[a][userData[a]]);
 280                                 }else           {       UWScoreSig[a].push_back(0.0);                                           }
 281
 282                         }
 283
 284                         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;  }
 285
 286                         //print output files
 287                         printUWSummaryFile(i);
 288                         if (random)  {  printUnweightedFile();  delete output;  }
 289                         if (phylip) {   createPhylipFile(i);            }
 290
 291                         rscoreFreq.clear();
 292                         rCumul.clear();
 293                         validScores.clear();
 294                         utreeScores.clear();
 295                         UWScoreSig.clear();
 296                 }
 297
 298
 299                 outSum.close();
 300
 301                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str());  }      return 0; }
 302
 303                 m->mothurOut("It took " + toString(time(NULL) - start) + " secs to run unifrac.unweighted."); m->mothurOutEndLine();
 304
 305                 m->mothurOutEndLine();
 306                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 307                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 308                 m->mothurOutEndLine();
 309
 310                 return 0;
 311
 312         }
 313         catch(exception& e) {
 314                 m->errorOut(e, "UnifracUnweightedCommand", "execute");
 315                 exit(1);
 316         }
 317 }
 318 /***********************************************************/
 319 void UnifracUnweightedCommand::printUnweightedFile() {
 320         try {
 321                 vector<double> data;
 322                 vector<string> tags;
 323
 324                 tags.push_back("Score");
 325                 tags.push_back("RandFreq"); tags.push_back("RandCumul");
 326
 327                 for(int a = 0; a < numComp; a++) {
 328                         output->initFile(groupComb[a], tags);
 329                         //print each line
 330                         for (map<float,float>::iterator it = validScores.begin(); it != validScores.end(); it++) {
 331                                 data.push_back(it->first);  data.push_back(rscoreFreq[a][it->first]); data.push_back(rCumul[a][it->first]);
 332                                 output->output(data);
 333                                 data.clear();
 334                         }
 335                         output->resetFile();
 336                 }
 337         }
 338         catch(exception& e) {
 339                 m->errorOut(e, "UnifracUnweightedCommand", "printUnweightedFile");
 340                 exit(1);
 341         }
 342 }
 343
 344 /***********************************************************/
 345 void UnifracUnweightedCommand::printUWSummaryFile(int i) {
 346         try {
 347
 348                 //format output
 349                 outSum.setf(ios::fixed, ios::floatfield); outSum.setf(ios::showpoint);
 350
 351                 //print each line
 352
 353                 for(int a = 0; a < numComp; a++) {
 354                         outSum << i+1 << '\t';
 355                         m->mothurOut(toString(i+1) + "\t");
 356
 357                         if (random) {
 358                                 if (UWScoreSig[a][0] > (1/(float)iters)) {
 359                                         outSum << setprecision(6) << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << setprecision(itersString.length()) << UWScoreSig[a][0] << endl;
 360                                         cout << setprecision(6)  << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << setprecision(itersString.length()) << UWScoreSig[a][0] << endl;
 361                                         m->mothurOutJustToLog(groupComb[a]  + "\t" + toString(utreeScores[a][0])  + "\t" + toString(UWScoreSig[a][0])+ "\n");
 362                                 }else {
 363                                         outSum << setprecision(6) << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << setprecision(itersString.length()) << "<" << (1/float(iters)) << endl;
 364                                         cout << setprecision(6)  << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << setprecision(itersString.length()) << "<" << (1/float(iters)) << endl;
 365                                         m->mothurOutJustToLog(groupComb[a]  + "\t" + toString(utreeScores[a][0])  + "\t<" + toString((1/float(iters))) + "\n");
 366                                 }
 367                         }else{
 368                                 outSum << setprecision(6) << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << "0.00" << endl;
 369                                 cout << setprecision(6)  << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << "0.00" << endl;
 370                                 m->mothurOutJustToLog(groupComb[a]  + "\t" + toString(utreeScores[a][0])  + "\t0.00\n");
 371                         }
 372                 }
 373
 374         }
 375         catch(exception& e) {
 376                 m->errorOut(e, "UnifracUnweightedCommand", "printUWSummaryFile");
 377                 exit(1);
 378         }
 379 }
 380 /***********************************************************/
 381 void UnifracUnweightedCommand::createPhylipFile(int i) {
 382         try {
 383                 string phylipFileName;
 384                 if ((outputForm == "lt") || (outputForm == "square")) {
 385                         phylipFileName = outputDir + m->getSimpleName(globaldata->getTreeFile())  + toString(i+1) + ".unweighted.phylip.dist";
 386                         outputNames.push_back(phylipFileName); outputTypes["phylip"].push_back(phylipFileName);
 387                 }else { //column
 388                         phylipFileName = outputDir + m->getSimpleName(globaldata->getTreeFile())  + toString(i+1) + ".unweighted.column.dist";
 389                         outputNames.push_back(phylipFileName); outputTypes["column"].push_back(phylipFileName);
 390                 }
 391
 392                 ofstream out;
 393                 m->openOutputFile(phylipFileName, out);
 394
 395                 if ((outputForm == "lt") || (outputForm == "square")) {
 396                         //output numSeqs
 397                         out << globaldata->Groups.size() << endl;
 398                 }
 399
 400                 //make matrix with scores in it
 401                 vector< vector<float> > dists;  dists.resize(globaldata->Groups.size());
 402                 for (int i = 0; i < globaldata->Groups.size(); i++) {
 403                         dists[i].resize(globaldata->Groups.size(), 0.0);
 404                 }
 405
 406                 //flip it so you can print it
 407                 int count = 0;
 408                 for (int r=0; r<globaldata->Groups.size(); r++) {
 409                         for (int l = r+1; l < globaldata->Groups.size(); l++) {
 410                                 dists[r][l] = utreeScores[count][0];
 411                                 dists[l][r] = utreeScores[count][0];
 412                                 count++;
 413                         }
 414                 }
 415
 416                 //output to file
 417                 for (int r=0; r<globaldata->Groups.size(); r++) {
 418                         //output name
 419                         string name = globaldata->Groups[r];
 420                         if (name.length() < 10) { //pad with spaces to make compatible
 421                                 while (name.length() < 10) {  name += " ";  }
 422                         }
 423
 424                         if (outputForm == "lt") {
 425                                 out << name << '\t';
 426
 427                                 //output distances
 428                                 for (int l = 0; l < r; l++) {   out  << dists[r][l] << '\t';  }
 429                                 out << endl;
 430                         }else if (outputForm == "square") {
 431                                 out << name << '\t';
 432
 433                                 //output distances
 434                                 for (int l = 0; l < globaldata->Groups.size(); l++) {   out  << dists[r][l] << '\t';  }
 435                                 out << endl;
 436                         }else{
 437                                 //output distances
 438                                 for (int l = 0; l < r; l++) {
 439                                         string otherName = globaldata->Groups[l];
 440                                         if (otherName.length() < 10) { //pad with spaces to make compatible
 441                                                 while (otherName.length() < 10) {  otherName += " ";  }
 442                                         }
 443
 444                                         out  << name << '\t' << otherName << dists[r][l] << endl;
 445                                 }
 446                         }
 447                 }
 448                 out.close();
 449         }
 450         catch(exception& e) {
 451                 m->errorOut(e, "UnifracUnweightedCommand", "createPhylipFile");
 452                 exit(1);
 453         }
 454 }
 455 /***********************************************************/
 456
 457
 458