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 UnifracUnweightedCommand::UnifracUnweightedCommand(string option)  {
  14         try {
  15                 globaldata = GlobalData::getInstance();
  16                 abort = false;
  17                 Groups.clear();
  18
  19                 //allow user to run help
  20                 if(option == "help") { help(); abort = true; }
  21
  22                 else {
  23                         //valid paramters for this command
  24                         string Array[] =  {"groups","iters","distance","random", "processors","outputdir","inputdir"};
  25                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  26
  27                         OptionParser parser(option);
  28                         map<string,string> parameters = parser.getParameters();
  29
  30                         ValidParameters validParameter;
  31
  32                         //check to make sure all parameters are valid for command
  33                         for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
  34                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  35                         }
  36
  37                         if (globaldata->gTree.size() == 0) {//no trees were read
  38                                 m->mothurOut("You must execute the read.tree command, before you may execute the unifrac.unweighted command."); m->mothurOutEndLine(); abort = true;  }
  39
  40                         //if the user changes the output directory command factory will send this info to us in the output parameter
  41                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
  42                                 outputDir = "";
  43                                 outputDir += m->hasPath(globaldata->inputFileName); //if user entered a file with a path then preserve it
  44                         }
  45
  46                         //check for optional parameter and set defaults
  47                         // ...at some point should added some additional type checking...
  48                         groups = validParameter.validFile(parameters, "groups", false);
  49                         if (groups == "not found") { groups = ""; }
  50                         else {
  51                                 m->splitAtDash(groups, Groups);
  52                                 globaldata->Groups = Groups;
  53                         }
  54
  55                         itersString = validParameter.validFile(parameters, "iters", false);                             if (itersString == "not found") { itersString = "1000"; }
  56                         convert(itersString, iters);
  57
  58                         string temp = validParameter.validFile(parameters, "distance", false);                  if (temp == "not found") { temp = "false"; }
  59                         phylip = m->isTrue(temp);
  60
  61                         temp = validParameter.validFile(parameters, "random", false);                                   if (temp == "not found") { temp = "f"; }
  62                         random = m->isTrue(temp);
  63
  64                         temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found"){       temp = "1";                             }
  65                         convert(temp, processors);
  66
  67                         if (!random) {  iters = 0;  } //turn off random calcs
  68
  69                         //if user selects distance = true and no groups it won't calc the pairwise
  70                         if ((phylip) && (Groups.size() == 0)) {
  71                                 groups = "all";
  72                                 m->splitAtDash(groups, Groups);
  73                                 globaldata->Groups = Groups;
  74                         }
  75
  76                         if (abort == false) {
  77                                 T = globaldata->gTree;
  78                                 tmap = globaldata->gTreemap;
  79                                 sumFile = outputDir + m->getSimpleName(globaldata->getTreeFile()) + ".uwsummary";
  80                                 outputNames.push_back(sumFile);
  81                                 m->openOutputFile(sumFile, outSum);
  82
  83                                 util = new SharedUtil();
  84                                 util->setGroups(globaldata->Groups, tmap->namesOfGroups, allGroups, numGroups, "unweighted");   //sets the groups the user wants to analyze
  85                                 util->getCombos(groupComb, globaldata->Groups, numComp);
  86
  87                                 if (numGroups == 1) { numComp++; groupComb.push_back(allGroups); }
  88
  89                                 unweighted = new Unweighted(tmap);
  90
  91                         }
  92
  93                 }
  94
  95         }
  96         catch(exception& e) {
  97                 m->errorOut(e, "UnifracUnweightedCommand", "UnifracUnweightedCommand");
  98                 exit(1);
  99         }
 100 }
 101
 102 //**********************************************************************************************************************
 103
 104 void UnifracUnweightedCommand::help(){
 105         try {
 106                 m->mothurOut("The unifrac.unweighted command can only be executed after a successful read.tree command.\n");
 107                 m->mothurOut("The unifrac.unweighted command parameters are groups, iters, distance, processors and random.  No parameters are required.\n");
 108                 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");
 109                 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");
 110                 m->mothurOut("The distance parameter allows you to create a distance file from the results. The default is false.\n");
 111                 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");
 112                 m->mothurOut("The processors parameter allows you to specify the number of processors to use. The default is 1.\n");
 113                 m->mothurOut("The unifrac.unweighted command should be in the following format: unifrac.unweighted(groups=yourGroups, iters=yourIters).\n");
 114                 m->mothurOut("Example unifrac.unweighted(groups=A-B-C, iters=500).\n");
 115                 m->mothurOut("The default value for groups is all the groups in your groupfile, and iters is 1000.\n");
 116                 m->mothurOut("The unifrac.unweighted command output two files: .unweighted and .uwsummary their descriptions are in the manual.\n");
 117                 m->mothurOut("Note: No spaces between parameter labels (i.e. groups), '=' and parameters (i.e.yourGroups).\n\n");
 118         }
 119         catch(exception& e) {
 120                 m->errorOut(e, "UnifracUnweightedCommand", "help");
 121                 exit(1);
 122         }
 123 }
 124
 125
 126 /***********************************************************/
 127 int UnifracUnweightedCommand::execute() {
 128         try {
 129
 130                 if (abort == true) { return 0; }
 131
 132                 int start = time(NULL);
 133
 134                 userData.resize(numComp,0);  //data[0] = unweightedscore
 135                 randomData.resize(numComp,0); //data[0] = unweightedscore
 136                 //create new tree with same num nodes and leaves as users
 137
 138                 outSum << "Tree#" << '\t' << "Groups" << '\t'  <<  "UWScore" <<'\t' << "UWSig" <<  endl;
 139                 m->mothurOut("Tree#\tGroups\tUWScore\tUWSig"); m->mothurOutEndLine();
 140
 141                 //get pscores for users trees
 142                 for (int i = 0; i < T.size(); i++) {
 143                         if (m->control_pressed) {
 144                                 outSum.close();
 145                                 for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());  }
 146                                 return 0;
 147                         }
 148
 149                         counter = 0;
 150
 151                         if (random)  {
 152                                 output = new ColumnFile(outputDir + m->getSimpleName(globaldata->getTreeFile())  + toString(i+1) + ".unweighted", itersString);
 153                                 outputNames.push_back(outputDir + m->getSimpleName(globaldata->getTreeFile())  + toString(i+1) + ".unweighted");
 154                         }
 155
 156
 157                         //get unweighted for users tree
 158                         rscoreFreq.resize(numComp);
 159                         rCumul.resize(numComp);
 160                         utreeScores.resize(numComp);
 161                         UWScoreSig.resize(numComp);
 162
 163                         userData = unweighted->getValues(T[i], processors, outputDir);  //userData[0] = unweightedscore
 164
 165                         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; }
 166
 167                         //output scores for each combination
 168                         for(int k = 0; k < numComp; k++) {
 169                                 //saves users score
 170                                 utreeScores[k].push_back(userData[k]);
 171
 172                                 //add users score to validscores
 173                                 validScores[userData[k]] = userData[k];
 174                         }
 175
 176                         //get unweighted scores for random trees - if random is false iters = 0
 177                         for (int j = 0; j < iters; j++) {
 178
 179                                 //we need a different getValues because when we swap the labels we only want to swap those in each pairwise comparison
 180                                 randomData = unweighted->getValues(T[i], "", "", processors, outputDir);
 181
 182                                 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; }
 183
 184                                 for(int k = 0; k < numComp; k++) {
 185                                         //add trees unweighted score to map of scores
 186                                         map<float,float>::iterator it = rscoreFreq[k].find(randomData[k]);
 187                                         if (it != rscoreFreq[k].end()) {//already have that score
 188                                                 rscoreFreq[k][randomData[k]]++;
 189                                         }else{//first time we have seen this score
 190                                                 rscoreFreq[k][randomData[k]] = 1;
 191                                         }
 192
 193                                         //add randoms score to validscores
 194                                         validScores[randomData[k]] = randomData[k];
 195                                 }
 196
 197                                 //report progress
 198                                 m->mothurOut("Iter: " + toString(j+1)); m->mothurOutEndLine();
 199                         }
 200
 201                         for(int a = 0; a < numComp; a++) {
 202                                 float rcumul = 1.0000;
 203
 204                                 if (random) {
 205                                         //this loop fills the cumulative maps and put 0.0000 in the score freq map to make it easier to print.
 206                                         for (map<float,float>::iterator it = validScores.begin(); it != validScores.end(); it++) {
 207                                                 //make rscoreFreq map and rCumul
 208                                                 map<float,float>::iterator it2 = rscoreFreq[a].find(it->first);
 209                                                 rCumul[a][it->first] = rcumul;
 210                                                 //get percentage of random trees with that info
 211                                                 if (it2 != rscoreFreq[a].end()) {  rscoreFreq[a][it->first] /= iters; rcumul-= it2->second;  }
 212                                                 else { rscoreFreq[a][it->first] = 0.0000; } //no random trees with that score
 213                                         }
 214                                         UWScoreSig[a].push_back(rCumul[a][userData[a]]);
 215                                 }else           {       UWScoreSig[a].push_back(0.0);                                           }
 216
 217                         }
 218
 219                         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;  }
 220
 221                         //print output files
 222                         printUWSummaryFile(i);
 223                         if (random)  {  printUnweightedFile();  delete output;  }
 224                         if (phylip) {   createPhylipFile(i);            }
 225
 226                         rscoreFreq.clear();
 227                         rCumul.clear();
 228                         validScores.clear();
 229                         utreeScores.clear();
 230                         UWScoreSig.clear();
 231                 }
 232
 233
 234                 outSum.close();
 235
 236                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str());  }      return 0; }
 237
 238                 m->mothurOut("It took " + toString(time(NULL) - start) + " secs to run unifrac.unweighted."); m->mothurOutEndLine();
 239
 240                 m->mothurOutEndLine();
 241                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 242                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 243                 m->mothurOutEndLine();
 244
 245                 return 0;
 246
 247         }
 248         catch(exception& e) {
 249                 m->errorOut(e, "UnifracUnweightedCommand", "execute");
 250                 exit(1);
 251         }
 252 }
 253 /***********************************************************/
 254 void UnifracUnweightedCommand::printUnweightedFile() {
 255         try {
 256                 vector<double> data;
 257                 vector<string> tags;
 258
 259                 tags.push_back("Score");
 260                 tags.push_back("RandFreq"); tags.push_back("RandCumul");
 261
 262                 for(int a = 0; a < numComp; a++) {
 263                         output->initFile(groupComb[a], tags);
 264                         //print each line
 265                         for (map<float,float>::iterator it = validScores.begin(); it != validScores.end(); it++) {
 266                                 data.push_back(it->first);  data.push_back(rscoreFreq[a][it->first]); data.push_back(rCumul[a][it->first]);
 267                                 output->output(data);
 268                                 data.clear();
 269                         }
 270                         output->resetFile();
 271                 }
 272         }
 273         catch(exception& e) {
 274                 m->errorOut(e, "UnifracUnweightedCommand", "printUnweightedFile");
 275                 exit(1);
 276         }
 277 }
 278
 279 /***********************************************************/
 280 void UnifracUnweightedCommand::printUWSummaryFile(int i) {
 281         try {
 282
 283                 //format output
 284                 outSum.setf(ios::fixed, ios::floatfield); outSum.setf(ios::showpoint);
 285
 286                 //print each line
 287
 288                 for(int a = 0; a < numComp; a++) {
 289                         outSum << i+1 << '\t';
 290                         m->mothurOut(toString(i+1) + "\t");
 291
 292                         if (random) {
 293                                 if (UWScoreSig[a][0] > (1/(float)iters)) {
 294                                         outSum << setprecision(6) << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << setprecision(itersString.length()) << UWScoreSig[a][0] << endl;
 295                                         cout << setprecision(6)  << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << setprecision(itersString.length()) << UWScoreSig[a][0] << endl;
 296                                         m->mothurOutJustToLog(groupComb[a]  + "\t" + toString(utreeScores[a][0])  + "\t" + toString(UWScoreSig[a][0])+ "\n");
 297                                 }else {
 298                                         outSum << setprecision(6) << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << setprecision(itersString.length()) << "<" << (1/float(iters)) << endl;
 299                                         cout << setprecision(6)  << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << setprecision(itersString.length()) << "<" << (1/float(iters)) << endl;
 300                                         m->mothurOutJustToLog(groupComb[a]  + "\t" + toString(utreeScores[a][0])  + "\t<" + toString((1/float(iters))) + "\n");
 301                                 }
 302                         }else{
 303                                 outSum << setprecision(6) << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << "0.00" << endl;
 304                                 cout << setprecision(6)  << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << "0.00" << endl;
 305                                 m->mothurOutJustToLog(groupComb[a]  + "\t" + toString(utreeScores[a][0])  + "\t0.00\n");
 306                         }
 307                 }
 308
 309         }
 310         catch(exception& e) {
 311                 m->errorOut(e, "UnifracUnweightedCommand", "printUWSummaryFile");
 312                 exit(1);
 313         }
 314 }
 315 /***********************************************************/
 316 void UnifracUnweightedCommand::createPhylipFile(int i) {
 317         try {
 318                 string phylipFileName = outputDir + m->getSimpleName(globaldata->getTreeFile())  + toString(i+1) + ".unweighted.dist";
 319                 outputNames.push_back(phylipFileName);
 320
 321                 ofstream out;
 322                 m->openOutputFile(phylipFileName, out);
 323
 324                 //output numSeqs
 325                 out << globaldata->Groups.size() << endl;
 326
 327                 //make matrix with scores in it
 328                 vector< vector<float> > dists;  dists.resize(globaldata->Groups.size());
 329                 for (int i = 0; i < globaldata->Groups.size(); i++) {
 330                         dists[i].resize(globaldata->Groups.size(), 0.0);
 331                 }
 332
 333                 //flip it so you can print it
 334                 int count = 0;
 335                 for (int r=0; r<globaldata->Groups.size(); r++) {
 336                         for (int l = r+1; l < globaldata->Groups.size(); l++) {
 337                                 dists[r][l] = utreeScores[count][0];
 338                                 dists[l][r] = utreeScores[count][0];
 339                                 count++;
 340                         }
 341                 }
 342
 343                 //output to file
 344                 for (int r=0; r<globaldata->Groups.size(); r++) {
 345                         //output name
 346                         string name = globaldata->Groups[r];
 347                         if (name.length() < 10) { //pad with spaces to make compatible
 348                                 while (name.length() < 10) {  name += " ";  }
 349                         }
 350                         out << name << '\t';
 351
 352                         //output distances
 353                         for (int l = 0; l < r; l++) {   out  << dists[r][l] << '\t';  }
 354                         out << endl;
 355                 }
 356                 out.close();
 357         }
 358         catch(exception& e) {
 359                 m->errorOut(e, "UnifracUnweightedCommand", "createPhylipFile");
 360                 exit(1);
 361         }
 362 }
 363 /***********************************************************/
 364
 365
 366