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"};
  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                         //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
  53                         if (abort == false) {
  54                                 T = globaldata->gTree;
  55                                 tmap = globaldata->gTreemap;
  56                                 sumFile = globaldata->getTreeFile() + ".uwsummary";
  57                                 openOutputFile(sumFile, outSum);
  58
  59                                 util = new SharedUtil();
  60                                 util->setGroups(globaldata->Groups, tmap->namesOfGroups, allGroups, numGroups, "unweighted");   //sets the groups the user wants to analyze
  61                                 util->getCombos(groupComb, globaldata->Groups, numComp);
  62
  63                                 if (numGroups == 1) { numComp++; groupComb.push_back(allGroups); }
  64
  65                                 unweighted = new Unweighted(tmap);
  66
  67                         }
  68
  69                 }
  70
  71         }
  72         catch(exception& e) {
  73                 errorOut(e, "UnifracUnweightedCommand", "UnifracUnweightedCommand");
  74                 exit(1);
  75         }
  76 }
  77
  78 //**********************************************************************************************************************
  79
  80 void UnifracUnweightedCommand::help(){
  81         try {
  82                 mothurOut("The unifrac.unweighted command can only be executed after a successful read.tree command.\n");
  83                 mothurOut("The unifrac.unweighted command parameters are groups and iters.  No parameters are required.\n");
  84                 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");
  85                 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");
  86                 mothurOut("The unifrac.unweighted command should be in the following format: unifrac.unweighted(groups=yourGroups, iters=yourIters).\n");
  87                 mothurOut("Example unifrac.unweighted(groups=A-B-C, iters=500).\n");
  88                 mothurOut("The default value for groups is all the groups in your groupfile, and iters is 1000.\n");
  89                 mothurOut("The unifrac.unweighted command output two files: .unweighted and .uwsummary their descriptions are in the manual.\n");
  90                 mothurOut("Note: No spaces between parameter labels (i.e. groups), '=' and parameters (i.e.yourGroups).\n\n");
  91         }
  92         catch(exception& e) {
  93                 errorOut(e, "UnifracUnweightedCommand", "help");
  94                 exit(1);
  95         }
  96 }
  97
  98
  99 /***********************************************************/
 100 int UnifracUnweightedCommand::execute() {
 101         try {
 102
 103                 if (abort == true) { return 0; }
 104
 105                 userData.resize(numComp,0);  //data[0] = unweightedscore
 106                 randomData.resize(numComp,0); //data[0] = unweightedscore
 107                 //create new tree with same num nodes and leaves as users
 108
 109                 outSum << "Tree#" << '\t' << "Groups" << '\t'  <<  "UWScore" <<'\t' << "UWSig" <<  endl;
 110                 mothurOut("Tree#\tGroups\tUWScore\tUWSig"); mothurOutEndLine();
 111
 112                 //get pscores for users trees
 113                 for (int i = 0; i < T.size(); i++) {
 114                         counter = 0;
 115
 116                         output = new ColumnFile(globaldata->getTreeFile()  + toString(i+1) + ".unweighted", itersString);
 117
 118                         //get unweighted for users tree
 119                         rscoreFreq.resize(numComp);
 120                         rCumul.resize(numComp);
 121                         utreeScores.resize(numComp);
 122                         UWScoreSig.resize(numComp);
 123
 124                         userData = unweighted->getValues(T[i]);  //userData[0] = unweightedscore
 125
 126                         //output scores for each combination
 127                         for(int k = 0; k < numComp; k++) {
 128                                 //saves users score
 129                                 utreeScores[k].push_back(userData[k]);
 130
 131                         }
 132
 133                         //get unweighted scores for random trees
 134                         for (int j = 0; j < iters; j++) {
 135                                 //we need a different getValues because when we swap the labels we only want to swap those in each parwise comparison
 136                                 randomData = unweighted->getValues(T[i], "", "");
 137
 138                                 for(int k = 0; k < numComp; k++) {
 139                                         //add trees unweighted score to map of scores
 140                                         map<float,float>::iterator it = rscoreFreq[k].find(randomData[k]);
 141                                         if (it != rscoreFreq[k].end()) {//already have that score
 142                                                 rscoreFreq[k][randomData[k]]++;
 143                                         }else{//first time we have seen this score
 144                                                 rscoreFreq[k][randomData[k]] = 1;
 145                                         }
 146
 147                                         //add randoms score to validscores
 148                                         validScores[randomData[k]] = randomData[k];
 149                                 }
 150
 151                         }
 152
 153                         for(int a = 0; a < numComp; a++) {
 154                                 float rcumul = 1.0000;
 155                                 //this loop fills the cumulative maps and put 0.0000 in the score freq map to make it easier to print.
 156                                 for (map<float,float>::iterator it = validScores.begin(); it != validScores.end(); it++) {
 157                                         //make rscoreFreq map and rCumul
 158                                         map<float,float>::iterator it2 = rscoreFreq[a].find(it->first);
 159                                         rCumul[a][it->first] = rcumul;
 160                                         //get percentage of random trees with that info
 161                                         if (it2 != rscoreFreq[a].end()) {  rscoreFreq[a][it->first] /= iters; rcumul-= it2->second;  }
 162                                         else { rscoreFreq[a][it->first] = 0.0000; } //no random trees with that score
 163                                 }
 164                                 UWScoreSig[a].push_back(rCumul[a][userData[a]]);
 165                         }
 166
 167
 168
 169                         printUnweightedFile();
 170                         printUWSummaryFile(i);
 171
 172                         delete output;
 173                         rscoreFreq.clear();
 174                         rCumul.clear();
 175                         validScores.clear();
 176                         utreeScores.clear();
 177                         UWScoreSig.clear();
 178                 }
 179
 180                 //reset groups parameter
 181                 globaldata->Groups.clear();
 182                 outSum.close();
 183
 184                 return 0;
 185
 186         }
 187         catch(exception& e) {
 188                 errorOut(e, "UnifracUnweightedCommand", "execute");
 189                 exit(1);
 190         }
 191 }
 192 /***********************************************************/
 193 void UnifracUnweightedCommand::printUnweightedFile() {
 194         try {
 195                 vector<double> data;
 196                 vector<string> tags;
 197                 tags.push_back("Score"); tags.push_back("RandFreq"); tags.push_back("RandCumul");
 198
 199                 for(int a = 0; a < numComp; a++) {
 200                         output->initFile(groupComb[a], tags);
 201                         //print each line
 202                         for (map<float,float>::iterator it = validScores.begin(); it != validScores.end(); it++) {
 203                                 data.push_back(it->first);  data.push_back(rscoreFreq[a][it->first]); data.push_back(rCumul[a][it->first]);
 204                                 output->output(data);
 205                                 data.clear();
 206                         }
 207                         output->resetFile();
 208                 }
 209         }
 210         catch(exception& e) {
 211                 errorOut(e, "UnifracUnweightedCommand", "printUnweightedFile");
 212                 exit(1);
 213         }
 214 }
 215
 216 /***********************************************************/
 217 void UnifracUnweightedCommand::printUWSummaryFile(int i) {
 218         try {
 219
 220                 //format output
 221                 outSum.setf(ios::fixed, ios::floatfield); outSum.setf(ios::showpoint);
 222
 223                 //print each line
 224
 225                 for(int a = 0; a < numComp; a++) {
 226                         outSum << i+1 << '\t';
 227                         mothurOut(toString(i+1) + "\t");
 228
 229                         if (UWScoreSig[a][0] > (1/(float)iters)) {
 230                                 outSum << setprecision(6) << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << setprecision(itersString.length()) << UWScoreSig[a][0] << endl;
 231                                 cout << setprecision(6)  << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << setprecision(itersString.length()) << UWScoreSig[a][0] << endl;
 232                                 mothurOutJustToLog(groupComb[a]  + "\t" + toString(utreeScores[a][0])  + "\t" + toString(UWScoreSig[a][0])); mothurOutEndLine();
 233                         }else {
 234                                 outSum << setprecision(6) << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << setprecision(itersString.length()) << "<" << (1/float(iters)) << endl;
 235                                 cout << setprecision(6)  << groupComb[a]  << '\t' << utreeScores[a][0] << '\t' << setprecision(itersString.length()) << "<" << (1/float(iters)) << endl;
 236                                 mothurOutJustToLog(groupComb[a]  + "\t" + toString(utreeScores[a][0])  + "\t<" + toString((1/float(iters)))); mothurOutEndLine();
 237                         }
 238                 }
 239
 240         }
 241         catch(exception& e) {
 242                 errorOut(e, "UnifracUnweightedCommand", "printUWSummaryFile");
 243                 exit(1);
 244         }
 245 }
 246
 247 /***********************************************************/
 248
 249