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"};
  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                                 mothurOut("You must execute the read.tree command, before you may execute the unifrac.weighted command."); mothurOutEndLine(); abort = true;  }
  39
  40                         //check for optional parameter and set defaults
  41                         // ...at some point should added some additional type checking...
  42                         groups = validParameter.validFile(parameters, "groups", false);
  43                         if (groups == "not found") { groups = ""; }
  44                         else {
  45                                 splitAtDash(groups, Groups);
  46                                 globaldata->Groups = Groups;
  47                         }
  48
  49                         itersString = validParameter.validFile(parameters, "iters", false);                     if (itersString == "not found") { itersString = "1000"; }
  50                         convert(itersString, iters);
  51
  52                         string temp = validParameter.validFile(parameters, "distance", false);                  if (temp == "not found") { temp = "false"; }
  53                         phylip = isTrue(temp);
  54
  55                         temp = validParameter.validFile(parameters, "random", false);                                   if (temp == "not found") { temp = "true"; }
  56                         random = isTrue(temp);
  57
  58                         if (!random) {  iters = 0;  } //turn off random calcs
  59
  60
  61                         if (abort == false) {
  62                                 T = globaldata->gTree;
  63                                 tmap = globaldata->gTreemap;
  64                                 sumFile = globaldata->getTreeFile() + ".wsummary";
  65                                 openOutputFile(sumFile, outSum);
  66
  67                                 util = new SharedUtil();
  68                                 string s; //to make work with setgroups
  69                                 util->setGroups(globaldata->Groups, tmap->namesOfGroups, s, numGroups, "weighted");     //sets the groups the user wants to analyze
  70                                 util->getCombos(groupComb, globaldata->Groups, numComp);
  71
  72                                 weighted = new Weighted(tmap);
  73
  74                         }
  75                 }
  76
  77
  78         }
  79         catch(exception& e) {
  80                 errorOut(e, "UnifracWeightedCommand", "UnifracWeightedCommand");
  81                 exit(1);
  82         }
  83 }
  84 //**********************************************************************************************************************
  85
  86 void UnifracWeightedCommand::help(){
  87         try {
  88                 mothurOut("The unifrac.weighted command can only be executed after a successful read.tree command.\n");
  89                 mothurOut("The unifrac.weighted command parameters are groups, iters, distance and random.  No parameters are required.\n");
  90                 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");
  91                 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");
  92                 mothurOut("The distance parameter allows you to create a distance file from the results. The default is false.\n");
  93                 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");
  94                 mothurOut("The unifrac.weighted command should be in the following format: unifrac.weighted(groups=yourGroups, iters=yourIters).\n");
  95                 mothurOut("Example unifrac.weighted(groups=A-B-C, iters=500).\n");
  96                 mothurOut("The default value for groups is all the groups in your groupfile, and iters is 1000.\n");
  97                 mothurOut("The unifrac.weighted command output two files: .weighted and .wsummary their descriptions are in the manual.\n");
  98                 mothurOut("Note: No spaces between parameter labels (i.e. groups), '=' and parameters (i.e.yourGroups).\n\n");
  99         }
 100         catch(exception& e) {
 101                 errorOut(e, "UnifracWeightedCommand", "help");
 102                 exit(1);
 103         }
 104 }
 105
 106 /***********************************************************/
 107 int UnifracWeightedCommand::execute() {
 108         try {
 109
 110                 if (abort == true) { return 0; }
 111
 112                 Progress* reading;
 113                 if (random) {   reading = new Progress("Comparing to random:", iters);  }
 114
 115                 //get weighted for users tree
 116                 userData.resize(numComp,0);  //data[0] = weightedscore AB, data[1] = weightedscore AC...
 117                 randomData.resize(numComp,0); //data[0] = weightedscore AB, data[1] = weightedscore AC...
 118
 119                 //create new tree with same num nodes and leaves as users
 120                 randT = new Tree();
 121
 122                 //get weighted scores for users trees
 123                 for (int i = 0; i < T.size(); i++) {
 124                         counter = 0;
 125                         rScores.resize(numComp);  //data[0] = weightedscore AB, data[1] = weightedscore AC...
 126                         uScores.resize(numComp);  //data[0] = weightedscore AB, data[1] = weightedscore AC...
 127
 128                         if (random) {  output = new ColumnFile(globaldata->getTreeFile()  + toString(i+1) + ".weighted", itersString);  }
 129
 130                         userData = weighted->getValues(T[i]);  //userData[0] = weightedscore
 131
 132                         //save users score
 133                         for (int s=0; s<numComp; s++) {
 134                                 //add users score to vector of user scores
 135                                 uScores[s].push_back(userData[s]);
 136
 137                                 //save users tree score for summary file
 138                                 utreeScores.push_back(userData[s]);
 139                         }
 140
 141                         //get scores for random trees
 142                         for (int j = 0; j < iters; j++) {
 143                                 int count = 0;
 144                                 for (int r=0; r<numGroups; r++) {
 145                                         for (int l = r+1; l < numGroups; l++) {
 146                                                 //copy T[i]'s info.
 147                                                 randT->getCopy(T[i]);
 148
 149                                                 //create a random tree with same topology as T[i], but different labels
 150                                                 randT->assembleRandomUnifracTree(globaldata->Groups[r], globaldata->Groups[l]);
 151                                                 //get wscore of random tree
 152                                                 randomData = weighted->getValues(randT, globaldata->Groups[r], globaldata->Groups[l]);
 153
 154                                                 //save scores
 155                                                 rScores[count].push_back(randomData[0]);
 156                                                 count++;
 157                                         }
 158                                 }
 159
 160                                 //update progress bar
 161                                 reading->update(j);
 162
 163                         }
 164
 165                         //removeValidScoresDuplicates();
 166                         //find the signifigance of the score for summary file
 167                         if (random) {
 168                                 for (int f = 0; f < numComp; f++) {
 169                                         //sort random scores
 170                                         sort(rScores[f].begin(), rScores[f].end());
 171
 172                                         //the index of the score higher than yours is returned
 173                                         //so if you have 1000 random trees the index returned is 100
 174                                         //then there are 900 trees with a score greater then you.
 175                                         //giving you a signifigance of 0.900
 176                                         int index = findIndex(userData[f], f);    if (index == -1) { mothurOut("error in UnifracWeightedCommand"); mothurOutEndLine(); exit(1); } //error code
 177
 178                                         //the signifigance is the number of trees with the users score or higher
 179                                         WScoreSig.push_back((iters-index)/(float)iters);
 180                                 }
 181
 182                                 //out << "Tree# " << i << endl;
 183                                 calculateFreqsCumuls();
 184                                 printWeightedFile();
 185
 186                                 delete output;
 187                         }
 188
 189                         //clear data
 190                         rScores.clear();
 191                         uScores.clear();
 192                         validScores.clear();
 193                 }
 194
 195                 //finish progress bar
 196                 if (random) {   reading->finish();      delete reading;         }
 197
 198                 printWSummaryFile();
 199
 200                 if (phylip) {   createPhylipFile();             }
 201
 202                 //clear out users groups
 203                 globaldata->Groups.clear();
 204
 205                 delete randT;
 206
 207                 return 0;
 208
 209         }
 210         catch(exception& e) {
 211                 errorOut(e, "UnifracWeightedCommand", "execute");
 212                 exit(1);
 213         }
 214 }
 215 /***********************************************************/
 216 void UnifracWeightedCommand::printWeightedFile() {
 217         try {
 218                 vector<double> data;
 219                 vector<string> tags;
 220                 tags.push_back("Score"); tags.push_back("RandFreq"); tags.push_back("RandCumul");
 221
 222                 for(int a = 0; a < numComp; a++) {
 223                         output->initFile(groupComb[a], tags);
 224                         //print each line
 225                         for (map<float,float>::iterator it = validScores.begin(); it != validScores.end(); it++) {
 226                                 data.push_back(it->first);  data.push_back(rScoreFreq[a][it->first]); data.push_back(rCumul[a][it->first]);
 227                                 output->output(data);
 228                                 data.clear();
 229                         }
 230                         output->resetFile();
 231                 }
 232         }
 233         catch(exception& e) {
 234                 errorOut(e, "UnifracWeightedCommand", "printWeightedFile");
 235                 exit(1);
 236         }
 237 }
 238
 239
 240 /***********************************************************/
 241 void UnifracWeightedCommand::printWSummaryFile() {
 242         try {
 243                 //column headers
 244                 outSum << "Tree#" << '\t' << "Groups" << '\t' << "WScore" << '\t' << "WSig" <<  endl;
 245                 mothurOut("Tree#\tGroups\tWScore\tWSig"); mothurOutEndLine();
 246
 247                 //format output
 248                 outSum.setf(ios::fixed, ios::floatfield); outSum.setf(ios::showpoint);
 249
 250                 //print each line
 251                 int count = 0;
 252                 for (int i = 0; i < T.size(); i++) {
 253                         for (int j = 0; j < numComp; j++) {
 254                                 if (random) {
 255                                         if (WScoreSig[count] > (1/(float)iters)) {
 256                                                 outSum << setprecision(6) << i+1 << '\t' << groupComb[j] << '\t' << utreeScores[count] << '\t' << setprecision(itersString.length()) << WScoreSig[count] << endl;
 257                                                 cout << setprecision(6) << i+1 << '\t' << groupComb[j] << '\t' << utreeScores[count] << '\t' << setprecision(itersString.length()) << WScoreSig[count] << endl;
 258                                                 mothurOutJustToLog(toString(i+1) +"\t" + groupComb[j] +"\t" + toString(utreeScores[count]) +"\t" +  toString(WScoreSig[count])); mothurOutEndLine();
 259                                         }else{
 260                                                 outSum << setprecision(6) << i+1 << '\t' << groupComb[j] << '\t' << utreeScores[count] << '\t' << setprecision(itersString.length()) << "<" << (1/float(iters)) << endl;
 261                                                 cout << setprecision(6) << i+1 << '\t' << groupComb[j] << '\t' << utreeScores[count] << '\t' << setprecision(itersString.length()) << "<" << (1/float(iters)) << endl;
 262                                                 mothurOutJustToLog(toString(i+1) +"\t" + groupComb[j] +"\t" + toString(utreeScores[count]) +"\t<" +  toString((1/float(iters)))); mothurOutEndLine();
 263                                         }
 264                                 }else{
 265                                         outSum << setprecision(6) << i+1 << '\t' << groupComb[j] << '\t' << utreeScores[count] << '\t' << "0.00" << endl;
 266                                         cout << setprecision(6) << i+1 << '\t' << groupComb[j] << '\t' << utreeScores[count] << '\t' << "0.00" << endl;
 267                                         mothurOutJustToLog(toString(i+1) +"\t" + groupComb[j] +"\t" + toString(utreeScores[count]) +"\t0.00"); mothurOutEndLine();
 268                                 }
 269                                 count++;
 270                         }
 271                 }
 272                 outSum.close();
 273         }
 274         catch(exception& e) {
 275                 errorOut(e, "UnifracWeightedCommand", "printWSummaryFile");
 276                 exit(1);
 277         }
 278 }
 279 /***********************************************************/
 280 void UnifracWeightedCommand::createPhylipFile() {
 281         try {
 282                 int count = 0;
 283                 //for each tree
 284                 for (int i = 0; i < T.size(); i++) {
 285
 286                         string phylipFileName = globaldata->getTreeFile()  + toString(i+1) + ".weighted.dist";
 287                         ofstream out;
 288                         openOutputFile(phylipFileName, out);
 289
 290                         //output numSeqs
 291                         out << globaldata->Groups.size() << endl;
 292
 293                         //make matrix with scores in it
 294                         vector< vector<float> > dists;  dists.resize(globaldata->Groups.size());
 295                         for (int i = 0; i < globaldata->Groups.size(); i++) {
 296                                 dists[i].resize(globaldata->Groups.size(), 0.0);
 297                         }
 298
 299                         //flip it so you can print it
 300                         for (int r=0; r<globaldata->Groups.size(); r++) {
 301                                 for (int l = r+1; l < globaldata->Groups.size(); l++) {
 302                                         dists[r][l] = (1.0 - utreeScores[count]);
 303                                         dists[l][r] = (1.0 - utreeScores[count]);
 304                                         count++;
 305                                 }
 306                         }
 307
 308                         //output to file
 309                         for (int r=0; r<globaldata->Groups.size(); r++) {
 310                                 //output name
 311                                 out << globaldata->Groups[r] << '\t';
 312
 313                                 //output distances
 314                                 for (int l = 0; l < r; l++) {   out  << dists[r][l] << '\t';  }
 315                                 out << endl;
 316                         }
 317                         out.close();
 318                 }
 319         }
 320         catch(exception& e) {
 321                 errorOut(e, "UnifracWeightedCommand", "createPhylipFile");
 322                 exit(1);
 323         }
 324 }
 325 /***********************************************************/
 326 int UnifracWeightedCommand::findIndex(float score, int index) {
 327         try{
 328                 for (int i = 0; i < rScores[index].size(); i++) {
 329                         if (rScores[index][i] >= score) {       return i;       }
 330                 }
 331                 return rScores[index].size();
 332         }
 333         catch(exception& e) {
 334                 errorOut(e, "UnifracWeightedCommand", "findIndex");
 335                 exit(1);
 336         }
 337 }
 338
 339 /***********************************************************/
 340
 341 void UnifracWeightedCommand::calculateFreqsCumuls() {
 342         try {
 343                 //clear out old tree values
 344                 rScoreFreq.clear();
 345                 rScoreFreq.resize(numComp);
 346                 rCumul.clear();
 347                 rCumul.resize(numComp);
 348                 validScores.clear();
 349
 350                 //calculate frequency
 351                 for (int f = 0; f < numComp; f++) {
 352                         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...
 353                                 validScores[rScores[f][i]] = rScores[f][i];
 354                                 map<float,float>::iterator it = rScoreFreq[f].find(rScores[f][i]);
 355                                 if (it != rScoreFreq[f].end()) {
 356                                         rScoreFreq[f][rScores[f][i]]++;
 357                                 }else{
 358                                         rScoreFreq[f][rScores[f][i]] = 1;
 359                                 }
 360                         }
 361                 }
 362
 363                 //calculate rcumul
 364                 for(int a = 0; a < numComp; a++) {
 365                         float rcumul = 1.0000;
 366                         //this loop fills the cumulative maps and put 0.0000 in the score freq map to make it easier to print.
 367                         for (map<float,float>::iterator it = validScores.begin(); it != validScores.end(); it++) {
 368                                 //make rscoreFreq map and rCumul
 369                                 map<float,float>::iterator it2 = rScoreFreq[a].find(it->first);
 370                                 rCumul[a][it->first] = rcumul;
 371                                 //get percentage of random trees with that info
 372                                 if (it2 != rScoreFreq[a].end()) {  rScoreFreq[a][it->first] /= iters; rcumul-= it2->second;  }
 373                                 else { rScoreFreq[a][it->first] = 0.0000; } //no random trees with that score
 374                         }
 375                 }
 376
 377         }
 378         catch(exception& e) {
 379                 errorOut(e, "UnifracWeightedCommand", "calculateFreqsCums");
 380                 exit(1);
 381         }
 382 }
 383
 384 /***********************************************************/
 385
 386
 387
 388
 389