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", "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                                 mothurOut("You must execute the read.tree command, before you may execute the unifrac.unweighted command."); 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                         //if user selects distance = true and no groups it won't calc the pairwise
  67                         if ((phylip) && (Groups.size() == 0)) {
  68                                 groups = "all";
  69                                 splitAtDash(groups, Groups);
  70                                 globaldata->Groups = Groups;
  71                         }
  72
  73                         if (abort == false) {
  74                                 T = globaldata->gTree;
  75                                 tmap = globaldata->gTreemap;
  76                                 sumFile = outputDir + getSimpleName(globaldata->getTreeFile()) + ".uwsummary";
  77                                 openOutputFile(sumFile, outSum);
  78
  79                                 util = new SharedUtil();
  80                                 util->setGroups(globaldata->Groups, tmap->namesOfGroups, allGroups, numGroups, "unweighted");   //sets the groups the user wants to analyze
  81                                 util->getCombos(groupComb, globaldata->Groups, numComp);
  82
  83                                 if (numGroups == 1) { numComp++; groupComb.push_back(allGroups); }
  84
  85                                 unweighted = new Unweighted(tmap);
  86
  87                         }
  88
  89                 }
  90
  91         }
  92         catch(exception& e) {
  93                 errorOut(e, "UnifracUnweightedCommand", "UnifracUnweightedCommand");
  94                 exit(1);
  95         }
  96 }
  97
  98 //**********************************************************************************************************************
  99
 100 void UnifracUnweightedCommand::help(){
 101         try {
 102                 mothurOut("The unifrac.unweighted command can only be executed after a successful read.tree command.\n");
 103                 mothurOut("The unifrac.unweighted command parameters are groups, iters, distance and random.  No parameters are required.\n");
 104                 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");
 105                 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");
 106                 mothurOut("The distance parameter allows you to create a distance file from the results. The default is false.\n");
 107                 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");
 108                 mothurOut("The unifrac.unweighted command should be in the following format: unifrac.unweighted(groups=yourGroups, iters=yourIters).\n");
 109                 mothurOut("Example unifrac.unweighted(groups=A-B-C, iters=500).\n");
 110                 mothurOut("The default value for groups is all the groups in your groupfile, and iters is 1000.\n");
 111                 mothurOut("The unifrac.unweighted command output two files: .unweighted and .uwsummary their descriptions are in the manual.\n");
 112                 mothurOut("Note: No spaces between parameter labels (i.e. groups), '=' and parameters (i.e.yourGroups).\n\n");
 113         }
 114         catch(exception& e) {
 115                 errorOut(e, "UnifracUnweightedCommand", "help");
 116                 exit(1);
 117         }
 118 }
 119
 120
 121 /***********************************************************/
 122 int UnifracUnweightedCommand::execute() {
 123         try {
 124
 125                 if (abort == true) { return 0; }
 126
 127                 userData.resize(numComp,0);  //data[0] = unweightedscore
 128                 randomData.resize(numComp,0); //data[0] = unweightedscore
 129                 //create new tree with same num nodes and leaves as users
 130
 131                 outSum << "Tree#" << '\t' << "Groups" << '\t'  <<  "UWScore" <<'\t' << "UWSig" <<  endl;
 132                 mothurOut("Tree#\tGroups\tUWScore\tUWSig"); mothurOutEndLine();
 133
 134                 //get pscores for users trees
 135                 for (int i = 0; i < T.size(); i++) {
 136                         counter = 0;
 137
 138                         if (random)  {  output = new ColumnFile(outputDir + getSimpleName(globaldata->getTreeFile())  + toString(i+1) + ".unweighted", itersString);  }
 139
 140                         //get unweighted for users tree
 141                         rscoreFreq.resize(numComp);
 142                         rCumul.resize(numComp);
 143                         utreeScores.resize(numComp);
 144                         UWScoreSig.resize(numComp);
 145
 146                         userData = unweighted->getValues(T[i]);  //userData[0] = unweightedscore
 147
 148                         //output scores for each combination
 149                         for(int k = 0; k < numComp; k++) {
 150                                 //saves users score
 151                                 utreeScores[k].push_back(userData[k]);
 152
 153                                 //add users score to validscores
 154                                 validScores[userData[k]] = userData[k];
 155                         }
 156
 157                         //get unweighted scores for random trees
 158                         for (int j = 0; j < iters; j++) {
 159                                 //we need a different getValues because when we swap the labels we only want to swap those in each pairwise comparison
 160                                 randomData = unweighted->getValues(T[i], "", "");
 161
 162                                 for(int k = 0; k < numComp; k++) {
 163                                         //add trees unweighted score to map of scores
 164                                         map<float,float>::iterator it = rscoreFreq[k].find(randomData[k]);
 165                                         if (it != rscoreFreq[k].end()) {//already have that score
 166                                                 rscoreFreq[k][randomData[k]]++;
 167                                         }else{//first time we have seen this score
 168                                                 rscoreFreq[k][randomData[k]] = 1;
 169                                         }
 170
 171                                         //add randoms score to validscores
 172                                         validScores[randomData[k]] = randomData[k];
 173                                 }
 174                         }
 175
 176                         for(int a = 0; a < numComp; a++) {
 177                                 float rcumul = 1.0000;
 178                                 //this loop fills the cumulative maps and put 0.0000 in the score freq map to make it easier to print.
 179                                 for (map<float,float>::iterator it = validScores.begin(); it != validScores.end(); it++) {
 180                                         //make rscoreFreq map and rCumul
 181                                         map<float,float>::iterator it2 = rscoreFreq[a].find(it->first);
 182                                         rCumul[a][it->first] = rcumul;
 183                                         //get percentage of random trees with that info
 184                                         if (it2 != rscoreFreq[a].end()) {  rscoreFreq[a][it->first] /= iters; rcumul-= it2->second;  }
 185                                         else { rscoreFreq[a][it->first] = 0.0000; } //no random trees with that score
 186                                 }
 187
 188                                 if (random) {   UWScoreSig[a].push_back(rCumul[a][userData[a]]);        }
 189                                 else            {       UWScoreSig[a].push_back(0.0);                                           }
 190                         }
 191
 192                         //print output files
 193                         printUWSummaryFile(i);
 194                         if (random)  {  printUnweightedFile();  delete output;  }
 195                         if (phylip) {   createPhylipFile(i);            }
 196
 197                         rscoreFreq.clear();
 198                         rCumul.clear();
 199                         validScores.clear();
 200                         utreeScores.clear();
 201                         UWScoreSig.clear();
 202                 }
 203
 204                 //reset groups parameter
 205                 globaldata->Groups.clear();
 206                 outSum.close();
 207
 208                 return 0;
 209
 210         }
 211         catch(exception& e) {
 212                 errorOut(e, "UnifracUnweightedCommand", "execute");
 213                 exit(1);
 214         }
 215 }
 216 /***********************************************************/
 217 void UnifracUnweightedCommand::printUnweightedFile() {
 218         try {
 219                 vector<double> data;
 220                 vector<string> tags;
 221
 222                 tags.push_back("Score");
 223                 tags.push_back("RandFreq"); tags.push_back("RandCumul");
 224
 225                 for(int a = 0; a < numComp; a++) {
 226                         output->initFile(groupComb[a], tags);
 227                         //print each line
 228                         for (map<float,float>::iterator it = validScores.begin(); it != validScores.end(); it++) {
 229                                 data.push_back(it->first);  data.push_back(rscoreFreq[a][it->first]); data.push_back(rCumul[a][it->first]);
 230                                 output->output(data);
 231                                 data.clear();
 232                         }
 233                         output->resetFile();
 234                 }
 235         }
 236         catch(exception& e) {
 237                 errorOut(e, "UnifracUnweightedCommand", "printUnweightedFile");
 238                 exit(1);
 239         }
 240 }
 241
 242 /***********************************************************/
 243 void UnifracUnweightedCommand::printUWSummaryFile(int i) {
 244         try {
 245
 246                 //format output
 247                 outSum.setf(ios::fixed, ios::floatfield); outSum.setf(ios::showpoint);
 248
 249                 //print each line
 250
 251                 for(int a = 0; a < numComp; a++) {
 252                         outSum << i+1 << '\t';
 253                         mothurOut(toString(i+1) + "\t");
 254
 255                         if (random) {
 256                                 if (UWScoreSig[a][0] > (1/(float)iters)) {
 257                                         outSum << setprecision(6) << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << setprecision(itersString.length()) << UWScoreSig[a][0] << endl;
 258                                         cout << setprecision(6)  << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << setprecision(itersString.length()) << UWScoreSig[a][0] << endl;
 259                                         mothurOutJustToLog(groupComb[a]  + "\t" + toString(utreeScores[a][0])  + "\t" + toString(UWScoreSig[a][0])); mothurOutEndLine();
 260                                 }else {
 261                                         outSum << setprecision(6) << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << setprecision(itersString.length()) << "<" << (1/float(iters)) << endl;
 262                                         cout << setprecision(6)  << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << setprecision(itersString.length()) << "<" << (1/float(iters)) << endl;
 263                                         mothurOutJustToLog(groupComb[a]  + "\t" + toString(utreeScores[a][0])  + "\t<" + toString((1/float(iters)))); mothurOutEndLine();
 264                                 }
 265                         }else{
 266                                 outSum << setprecision(6) << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << "0.00" << endl;
 267                                 cout << setprecision(6)  << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << "0.00" << endl;
 268                                 mothurOutJustToLog(groupComb[a]  + "\t" + toString(utreeScores[a][0])  + "\t0.00"); mothurOutEndLine();
 269                         }
 270                 }
 271
 272         }
 273         catch(exception& e) {
 274                 errorOut(e, "UnifracUnweightedCommand", "printUWSummaryFile");
 275                 exit(1);
 276         }
 277 }
 278 /***********************************************************/
 279 void UnifracUnweightedCommand::createPhylipFile(int i) {
 280         try {
 281                 string phylipFileName = outputDir + getSimpleName(globaldata->getTreeFile())  + toString(i+1) + ".unweighted.dist";
 282                 ofstream out;
 283                 openOutputFile(phylipFileName, out);
 284
 285                 //output numSeqs
 286                 out << globaldata->Groups.size() << endl;
 287
 288                 //make matrix with scores in it
 289                 vector< vector<float> > dists;  dists.resize(globaldata->Groups.size());
 290                 for (int i = 0; i < globaldata->Groups.size(); i++) {
 291                         dists[i].resize(globaldata->Groups.size(), 0.0);
 292                 }
 293
 294                 //flip it so you can print it
 295                 int count = 0;
 296                 for (int r=0; r<globaldata->Groups.size(); r++) {
 297                         for (int l = r+1; l < globaldata->Groups.size(); l++) {
 298                                 dists[r][l] = (1.0 - utreeScores[count][0]);
 299                                 dists[l][r] = (1.0 - utreeScores[count][0]);
 300                                 count++;
 301                         }
 302                 }
 303
 304                 //output to file
 305                 for (int r=0; r<globaldata->Groups.size(); r++) {
 306                         //output name
 307                         string name = globaldata->Groups[r];
 308                         if (name.length() < 10) { //pad with spaces to make compatible
 309                                 while (name.length() < 10) {  name += " ";  }
 310                         }
 311                         out << name << '\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         catch(exception& e) {
 320                 errorOut(e, "UnifracUnweightedCommand", "createPhylipFile");
 321                 exit(1);
 322         }
 323 }
 324 /***********************************************************/
 325
 326
 327