unifracweightedcommand.cpp

   1 /*
   2  *  unifracweightedcommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by Sarah Westcott on 2/9/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10 #include "unifracweightedcommand.h"
  11
  12 /***********************************************************/
  13 UnifracWeightedCommand::UnifracWeightedCommand(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","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.weighted 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 += 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                                 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 = isTrue(temp);
  60
  61                         temp = validParameter.validFile(parameters, "random", false);                                   if (temp == "not found") { temp = "true"; }
  62                         random = isTrue(temp);
  63
  64                         if (!random) {  iters = 0;  } //turn off random calcs
  65
  66
  67                         if (abort == false) {
  68                                 T = globaldata->gTree;
  69                                 tmap = globaldata->gTreemap;
  70                                 sumFile = outputDir + getSimpleName(globaldata->getTreeFile()) + ".wsummary";
  71                                 openOutputFile(sumFile, outSum);
  72                                 outputNames.push_back(sumFile);
  73
  74                                 util = new SharedUtil();
  75                                 string s; //to make work with setgroups
  76                                 util->setGroups(globaldata->Groups, tmap->namesOfGroups, s, numGroups, "weighted");     //sets the groups the user wants to analyze
  77                                 util->getCombos(groupComb, globaldata->Groups, numComp);
  78
  79                                 weighted = new Weighted(tmap);
  80
  81                         }
  82                 }
  83
  84
  85         }
  86         catch(exception& e) {
  87                 m->errorOut(e, "UnifracWeightedCommand", "UnifracWeightedCommand");
  88                 exit(1);
  89         }
  90 }
  91 //**********************************************************************************************************************
  92
  93 void UnifracWeightedCommand::help(){
  94         try {
  95                 m->mothurOut("The unifrac.weighted command can only be executed after a successful read.tree command.\n");
  96                 m->mothurOut("The unifrac.weighted command parameters are groups, iters, distance and random.  No parameters are required.\n");
  97                 m->mothurOut("The groups parameter allows you to specify which of the groups in your groupfile you would like analyzed.  You must enter at least 2 valid groups.\n");
  98                 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");
  99                 m->mothurOut("The distance parameter allows you to create a distance file from the results. The default is false.\n");
 100                 m->mothurOut("The random parameter allows you to shut off the comparison to random trees. The default is true, meaning compare your trees with randomly generated trees.\n");
 101                 m->mothurOut("The unifrac.weighted command should be in the following format: unifrac.weighted(groups=yourGroups, iters=yourIters).\n");
 102                 m->mothurOut("Example unifrac.weighted(groups=A-B-C, iters=500).\n");
 103                 m->mothurOut("The default value for groups is all the groups in your groupfile, and iters is 1000.\n");
 104                 m->mothurOut("The unifrac.weighted command output two files: .weighted and .wsummary their descriptions are in the manual.\n");
 105                 m->mothurOut("Note: No spaces between parameter labels (i.e. groups), '=' and parameters (i.e.yourGroups).\n\n");
 106         }
 107         catch(exception& e) {
 108                 m->errorOut(e, "UnifracWeightedCommand", "help");
 109                 exit(1);
 110         }
 111 }
 112
 113 /***********************************************************/
 114 int UnifracWeightedCommand::execute() {
 115         try {
 116
 117                 if (abort == true) { return 0; }
 118
 119                 Progress* reading;
 120                 if (random) {   reading = new Progress("Comparing to random:", iters);  }
 121
 122                 //get weighted for users tree
 123                 userData.resize(numComp,0);  //data[0] = weightedscore AB, data[1] = weightedscore AC...
 124                 randomData.resize(numComp,0); //data[0] = weightedscore AB, data[1] = weightedscore AC...
 125
 126                 //create new tree with same num nodes and leaves as users
 127                 randT = new Tree();
 128
 129                 //get weighted scores for users trees
 130                 for (int i = 0; i < T.size(); i++) {
 131                         counter = 0;
 132                         rScores.resize(numComp);  //data[0] = weightedscore AB, data[1] = weightedscore AC...
 133                         uScores.resize(numComp);  //data[0] = weightedscore AB, data[1] = weightedscore AC...
 134
 135                         if (random) {
 136                                 output = new ColumnFile(outputDir + getSimpleName(globaldata->getTreeFile())  + toString(i+1) + ".weighted", itersString);
 137                                 outputNames.push_back(outputDir + getSimpleName(globaldata->getTreeFile())  + toString(i+1) + ".weighted");
 138                         }
 139
 140                         userData = weighted->getValues(T[i]);  //userData[0] = weightedscore
 141
 142                         //save users score
 143                         for (int s=0; s<numComp; s++) {
 144                                 //add users score to vector of user scores
 145                                 uScores[s].push_back(userData[s]);
 146
 147                                 //save users tree score for summary file
 148                                 utreeScores.push_back(userData[s]);
 149                         }
 150
 151                         //get scores for random trees
 152                         for (int j = 0; j < iters; j++) {
 153                                 int count = 0;
 154                                 for (int r=0; r<numGroups; r++) {
 155                                         for (int l = r+1; l < numGroups; l++) {
 156                                                 //copy T[i]'s info.
 157                                                 randT->getCopy(T[i]);
 158
 159                                                 //create a random tree with same topology as T[i], but different labels
 160                                                 randT->assembleRandomUnifracTree(globaldata->Groups[r], globaldata->Groups[l]);
 161                                                 //get wscore of random tree
 162                                                 randomData = weighted->getValues(randT, globaldata->Groups[r], globaldata->Groups[l]);
 163
 164                                                 //save scores
 165                                                 rScores[count].push_back(randomData[0]);
 166                                                 count++;
 167                                         }
 168                                 }
 169
 170                                 //update progress bar
 171                                 reading->update(j);
 172
 173                         }
 174
 175                         //removeValidScoresDuplicates();
 176                         //find the signifigance of the score for summary file
 177                         if (random) {
 178                                 for (int f = 0; f < numComp; f++) {
 179                                         //sort random scores
 180                                         sort(rScores[f].begin(), rScores[f].end());
 181
 182                                         //the index of the score higher than yours is returned
 183                                         //so if you have 1000 random trees the index returned is 100
 184                                         //then there are 900 trees with a score greater then you.
 185                                         //giving you a signifigance of 0.900
 186                                         int index = findIndex(userData[f], f);    if (index == -1) { m->mothurOut("error in UnifracWeightedCommand"); m->mothurOutEndLine(); exit(1); } //error code
 187
 188                                         //the signifigance is the number of trees with the users score or higher
 189                                         WScoreSig.push_back((iters-index)/(float)iters);
 190                                 }
 191
 192                                 //out << "Tree# " << i << endl;
 193                                 calculateFreqsCumuls();
 194                                 printWeightedFile();
 195
 196                                 delete output;
 197                         }
 198
 199                         //clear data
 200                         rScores.clear();
 201                         uScores.clear();
 202                         validScores.clear();
 203                 }
 204
 205                 //finish progress bar
 206                 if (random) {   reading->finish();      delete reading;         }
 207
 208                 printWSummaryFile();
 209
 210                 if (phylip) {   createPhylipFile();             }
 211
 212                 //clear out users groups
 213                 globaldata->Groups.clear();
 214
 215                 delete randT;
 216
 217                 m->mothurOutEndLine();
 218                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 219                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 220                 m->mothurOutEndLine();
 221
 222                 return 0;
 223
 224         }
 225         catch(exception& e) {
 226                 m->errorOut(e, "UnifracWeightedCommand", "execute");
 227                 exit(1);
 228         }
 229 }
 230 /***********************************************************/
 231 void UnifracWeightedCommand::printWeightedFile() {
 232         try {
 233                 vector<double> data;
 234                 vector<string> tags;
 235                 tags.push_back("Score"); tags.push_back("RandFreq"); tags.push_back("RandCumul");
 236
 237                 for(int a = 0; a < numComp; a++) {
 238                         output->initFile(groupComb[a], tags);
 239                         //print each line
 240                         for (map<float,float>::iterator it = validScores.begin(); it != validScores.end(); it++) {
 241                                 data.push_back(it->first);  data.push_back(rScoreFreq[a][it->first]); data.push_back(rCumul[a][it->first]);
 242                                 output->output(data);
 243                                 data.clear();
 244                         }
 245                         output->resetFile();
 246                 }
 247         }
 248         catch(exception& e) {
 249                 m->errorOut(e, "UnifracWeightedCommand", "printWeightedFile");
 250                 exit(1);
 251         }
 252 }
 253
 254
 255 /***********************************************************/
 256 void UnifracWeightedCommand::printWSummaryFile() {
 257         try {
 258                 //column headers
 259                 outSum << "Tree#" << '\t' << "Groups" << '\t' << "WScore" << '\t' << "WSig" <<  endl;
 260                 m->mothurOut("Tree#\tGroups\tWScore\tWSig"); m->mothurOutEndLine();
 261
 262                 //format output
 263                 outSum.setf(ios::fixed, ios::floatfield); outSum.setf(ios::showpoint);
 264
 265                 //print each line
 266                 int count = 0;
 267                 for (int i = 0; i < T.size(); i++) {
 268                         for (int j = 0; j < numComp; j++) {
 269                                 if (random) {
 270                                         if (WScoreSig[count] > (1/(float)iters)) {
 271                                                 outSum << setprecision(6) << i+1 << '\t' << groupComb[j] << '\t' << utreeScores[count] << '\t' << setprecision(itersString.length()) << WScoreSig[count] << endl;
 272                                                 cout << setprecision(6) << i+1 << '\t' << groupComb[j] << '\t' << utreeScores[count] << '\t' << setprecision(itersString.length()) << WScoreSig[count] << endl;
 273                                                 m->mothurOutJustToLog(toString(i+1) +"\t" + groupComb[j] +"\t" + toString(utreeScores[count]) +"\t" +  toString(WScoreSig[count])); m->mothurOutEndLine();
 274                                         }else{
 275                                                 outSum << setprecision(6) << i+1 << '\t' << groupComb[j] << '\t' << utreeScores[count] << '\t' << setprecision(itersString.length()) << "<" << (1/float(iters)) << endl;
 276                                                 cout << setprecision(6) << i+1 << '\t' << groupComb[j] << '\t' << utreeScores[count] << '\t' << setprecision(itersString.length()) << "<" << (1/float(iters)) << endl;
 277                                                 m->mothurOutJustToLog(toString(i+1) +"\t" + groupComb[j] +"\t" + toString(utreeScores[count]) +"\t<" +  toString((1/float(iters)))); m->mothurOutEndLine();
 278                                         }
 279                                 }else{
 280                                         outSum << setprecision(6) << i+1 << '\t' << groupComb[j] << '\t' << utreeScores[count] << '\t' << "0.00" << endl;
 281                                         cout << setprecision(6) << i+1 << '\t' << groupComb[j] << '\t' << utreeScores[count] << '\t' << "0.00" << endl;
 282                                         m->mothurOutJustToLog(toString(i+1) +"\t" + groupComb[j] +"\t" + toString(utreeScores[count]) +"\t0.00"); m->mothurOutEndLine();
 283                                 }
 284                                 count++;
 285                         }
 286                 }
 287                 outSum.close();
 288         }
 289         catch(exception& e) {
 290                 m->errorOut(e, "UnifracWeightedCommand", "printWSummaryFile");
 291                 exit(1);
 292         }
 293 }
 294 /***********************************************************/
 295 void UnifracWeightedCommand::createPhylipFile() {
 296         try {
 297                 int count = 0;
 298                 //for each tree
 299                 for (int i = 0; i < T.size(); i++) {
 300
 301                         string phylipFileName = outputDir + getSimpleName(globaldata->getTreeFile())  + toString(i+1) + ".weighted.dist";
 302                         outputNames.push_back(phylipFileName);
 303                         ofstream out;
 304                         openOutputFile(phylipFileName, out);
 305
 306                         //output numSeqs
 307                         out << globaldata->Groups.size() << endl;
 308
 309                         //make matrix with scores in it
 310                         vector< vector<float> > dists;  dists.resize(globaldata->Groups.size());
 311                         for (int i = 0; i < globaldata->Groups.size(); i++) {
 312                                 dists[i].resize(globaldata->Groups.size(), 0.0);
 313                         }
 314
 315                         //flip it so you can print it
 316                         for (int r=0; r<globaldata->Groups.size(); r++) {
 317                                 for (int l = r+1; l < globaldata->Groups.size(); l++) {
 318                                         dists[r][l] = (1.0 - utreeScores[count]);
 319                                         dists[l][r] = (1.0 - utreeScores[count]);
 320                                         count++;
 321                                 }
 322                         }
 323
 324                         //output to file
 325                         for (int r=0; r<globaldata->Groups.size(); r++) {
 326                                 //output name
 327                                 string name = globaldata->Groups[r];
 328                                 if (name.length() < 10) { //pad with spaces to make compatible
 329                                         while (name.length() < 10) {  name += " ";  }
 330                                 }
 331                                 out << name << '\t';
 332
 333                                 //output distances
 334                                 for (int l = 0; l < r; l++) {   out  << dists[r][l] << '\t';  }
 335                                 out << endl;
 336                         }
 337                         out.close();
 338                 }
 339         }
 340         catch(exception& e) {
 341                 m->errorOut(e, "UnifracWeightedCommand", "createPhylipFile");
 342                 exit(1);
 343         }
 344 }
 345 /***********************************************************/
 346 int UnifracWeightedCommand::findIndex(float score, int index) {
 347         try{
 348                 for (int i = 0; i < rScores[index].size(); i++) {
 349                         if (rScores[index][i] >= score) {       return i;       }
 350                 }
 351                 return rScores[index].size();
 352         }
 353         catch(exception& e) {
 354                 m->errorOut(e, "UnifracWeightedCommand", "findIndex");
 355                 exit(1);
 356         }
 357 }
 358
 359 /***********************************************************/
 360
 361 void UnifracWeightedCommand::calculateFreqsCumuls() {
 362         try {
 363                 //clear out old tree values
 364                 rScoreFreq.clear();
 365                 rScoreFreq.resize(numComp);
 366                 rCumul.clear();
 367                 rCumul.resize(numComp);
 368                 validScores.clear();
 369
 370                 //calculate frequency
 371                 for (int f = 0; f < numComp; f++) {
 372                         for (int i = 0; i < rScores[f].size(); i++) { //looks like 0,0,1,1,1,2,4,7...  you want to make a map that say rScoreFreq[0] = 2, rScoreFreq[1] = 3...
 373                                 validScores[rScores[f][i]] = rScores[f][i];
 374                                 map<float,float>::iterator it = rScoreFreq[f].find(rScores[f][i]);
 375                                 if (it != rScoreFreq[f].end()) {
 376                                         rScoreFreq[f][rScores[f][i]]++;
 377                                 }else{
 378                                         rScoreFreq[f][rScores[f][i]] = 1;
 379                                 }
 380                         }
 381                 }
 382
 383                 //calculate rcumul
 384                 for(int a = 0; a < numComp; a++) {
 385                         float rcumul = 1.0000;
 386                         //this loop fills the cumulative maps and put 0.0000 in the score freq map to make it easier to print.
 387                         for (map<float,float>::iterator it = validScores.begin(); it != validScores.end(); it++) {
 388                                 //make rscoreFreq map and rCumul
 389                                 map<float,float>::iterator it2 = rScoreFreq[a].find(it->first);
 390                                 rCumul[a][it->first] = rcumul;
 391                                 //get percentage of random trees with that info
 392                                 if (it2 != rScoreFreq[a].end()) {  rScoreFreq[a][it->first] /= iters; rcumul-= it2->second;  }
 393                                 else { rScoreFreq[a][it->first] = 0.0000; } //no random trees with that score
 394                         }
 395                 }
 396
 397         }
 398         catch(exception& e) {
 399                 m->errorOut(e, "UnifracWeightedCommand", "calculateFreqsCums");
 400                 exit(1);
 401         }
 402 }
 403
 404 /***********************************************************/
 405
 406
 407
 408
 409