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() {
  14         try {
  15                 globaldata = GlobalData::getInstance();
  16
  17                 T = globaldata->gTree;
  18                 tmap = globaldata->gTreemap;
  19                 sumFile = globaldata->getTreeFile() + ".uwsummary";
  20                 openOutputFile(sumFile, outSum);
  21
  22                 setGroups(); //sets users groups to analyze
  23                 convert(globaldata->getIters(), iters);  //how many random trees to generate
  24                 unweighted = new Unweighted(tmap);
  25
  26         }
  27         catch(exception& e) {
  28                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracUnweightedCommand class Function UnifracUnweightedCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  29                 exit(1);
  30         }
  31         catch(...) {
  32                 cout << "An unknown error has occurred in the UnifracUnweightedCommand class function UnifracUnweightedCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  33                 exit(1);
  34         }
  35 }
  36 /***********************************************************/
  37 int UnifracUnweightedCommand::execute() {
  38         try {
  39
  40                 userData.resize(numComp,0);  //data[0] = unweightedscore
  41                 randomData.resize(numComp,0); //data[0] = unweightedscore
  42                 //create new tree with same num nodes and leaves as users
  43
  44                 //get pscores for users trees
  45                 for (int i = 0; i < T.size(); i++) {
  46                         counter = 0;
  47                         unweightedFile = globaldata->getTreeFile()  + toString(i+1) + ".unweighted";
  48                         unweightedFileout = globaldata->getTreeFile() + "temp." + toString(i+1) + ".unweighted";
  49
  50                         //column headers
  51                         outSum << "Tree# " << i+1 << endl;
  52                         outSum << "Comb" << '\t'  <<  "UWScore" << '\t' << '\t' << "UWSig" <<  endl;
  53                         cout << "Tree# " << i+1 << endl;
  54                         cout << "Comb" << '\t'  <<  "UWScore" << '\t' << '\t' << "UWSig" <<  endl;
  55
  56
  57                         //get unweighted for users tree
  58                         rscoreFreq.resize(numComp);
  59                         rCumul.resize(numComp);
  60                         utreeScores.resize(numComp);
  61                         UWScoreSig.resize(numComp);
  62
  63                         userData = unweighted->getValues(T[i]);  //userData[0] = unweightedscore
  64
  65                         //output scores for each combination
  66                         for(int k = 0; k < numComp; k++) {
  67                                 //saves users score
  68                                 utreeScores[k].push_back(userData[k]);
  69                         }
  70
  71                         //get unweighted scores for random trees
  72                         for (int j = 0; j < iters; j++) {
  73                                 //we need a different getValues because when we swap the labels we only want to swap those in each parwise comparison
  74                                 randomData = unweighted->getValues(T[i], "", "");
  75
  76                                 for(int k = 0; k < numComp; k++) {
  77                                         //add trees unweighted score to map of scores
  78                                         it2 = rscoreFreq[k].find(randomData[k]);
  79                                         if (it2 != rscoreFreq[k].end()) {//already have that score
  80                                                 rscoreFreq[k][randomData[k]]++;
  81                                         }else{//first time we have seen this score
  82                                                 rscoreFreq[k][randomData[k]] = 1;
  83                                         }
  84
  85                                         //add randoms score to validscores
  86                                         validScores[randomData[k]] = randomData[k];
  87                                 }
  88                         }
  89
  90                 for(int a = 0; a < numComp; a++) {
  91                         float rcumul = 1.0000;
  92                         //this loop fills the cumulative maps and put 0.0000 in the score freq map to make it easier to print.
  93                         for (it = validScores.begin(); it != validScores.end(); it++) {
  94                                 //make rscoreFreq map and rCumul
  95                                 it2 = rscoreFreq[a].find(it->first);
  96                                 rCumul[a][it->first] = rcumul;
  97                                 //get percentage of random trees with that info
  98                                 if (it2 != rscoreFreq[a].end()) {  rscoreFreq[a][it->first] /= iters; rcumul-= it2->second;  }
  99                                 else { rscoreFreq[a][it->first] = 0.0000; } //no random trees with that score
 100                         }
 101                         UWScoreSig[a].push_back(rCumul[a][userData[a]]);
 102                 }
 103
 104                 printUnweightedFile();
 105                 printUWSummaryFile();
 106
 107                 rscoreFreq.clear();
 108                 rCumul.clear();
 109                 validScores.clear();
 110                 utreeScores.clear();
 111                 UWScoreSig.clear();
 112         }
 113                 //reset groups parameter
 114                 globaldata->Groups.clear(); globaldata->setGroups("");
 115                 outSum.close();
 116
 117                 return 0;
 118
 119         }
 120         catch(exception& e) {
 121                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracUnweightedCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 122                 exit(1);
 123         }
 124         catch(...) {
 125                 cout << "An unknown error has occurred in the UnifracUnweightedCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 126                 exit(1);
 127         }
 128 }
 129 /***********************************************************/
 130 void UnifracUnweightedCommand::printUnweightedFile() {
 131         try {
 132                 vector<double> data;
 133
 134                 for(int a = 0; a < numComp; a++) {
 135                         initFile(groupComb[a]);
 136                         //print each line
 137                         for (it = validScores.begin(); it != validScores.end(); it++) {
 138                                 data.push_back(it->first);  data.push_back(rscoreFreq[a][it->first]); data.push_back(rCumul[a][it->first]);
 139                                 output(data);
 140                                 data.clear();
 141                         }
 142                         resetFile();
 143                 }
 144
 145                 out.close();
 146                 inFile.close();
 147                 remove(unweightedFileout.c_str());
 148
 149         }
 150         catch(exception& e) {
 151                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracUnweightedCommand class Function printUnweightedFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 152                 exit(1);
 153         }
 154         catch(...) {
 155                 cout << "An unknown error has occurred in the UnifracUnweightedCommand class function printUnweightedFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 156                 exit(1);
 157         }
 158 }
 159
 160 /***********************************************************/
 161 void UnifracUnweightedCommand::printUWSummaryFile() {
 162         try {
 163
 164                 //format output
 165                 outSum.setf(ios::fixed, ios::floatfield); outSum.setf(ios::showpoint);
 166
 167                 //print each line
 168                 for(int a = 0; a < numComp; a++) {
 169                         if (UWScoreSig[a][0] > (1/(float)iters)) {
 170                                 outSum << setprecision(globaldata->getIters().length()) << groupComb[a] << '\t' << '\t' << utreeScores[a][0] << '\t' << UWScoreSig[a][0] << endl;
 171                                 cout << setprecision(globaldata->getIters().length())  << groupComb[a] << '\t' << '\t' << utreeScores[a][0] << '\t' << UWScoreSig[a][0] << endl;
 172                         }else {
 173                                 outSum << setprecision(globaldata->getIters().length()) << groupComb[a] << '\t' << '\t' << utreeScores[a][0] << '\t' << "<" << (1/float(iters)) << endl;
 174                                 cout << setprecision(globaldata->getIters().length())  << groupComb[a] << '\t' << '\t' << utreeScores[a][0] << '\t' << "<" << (1/float(iters)) << endl;
 175                         }
 176                 }
 177
 178         }
 179         catch(exception& e) {
 180                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracUnweightedCommand class Function printUWSummaryFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 181                 exit(1);
 182         }
 183         catch(...) {
 184                 cout << "An unknown error has occurred in the UnifracUnweightedCommand class function printUWSummaryFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 185                 exit(1);
 186         }
 187 }
 188 /***********************************************************/
 189
 190 void UnifracUnweightedCommand::setGroups() {
 191         try {
 192                 string allGroups = "";
 193                 numGroups = 0;
 194                 //if the user has not entered specific groups to analyze then do them all
 195                 if (globaldata->Groups.size() != 0) {
 196                         if (globaldata->Groups[0] != "all") {
 197                                 //check that groups are valid
 198                                 for (int i = 0; i < globaldata->Groups.size(); i++) {
 199                                         if (tmap->isValidGroup(globaldata->Groups[i]) != true) {
 200                                                 cout << globaldata->Groups[i] << " is not a valid group, and will be disregarded." << endl;
 201                                                 // erase the invalid group from globaldata->Groups
 202                                                 globaldata->Groups.erase(globaldata->Groups.begin()+i);
 203                                         }
 204                                 }
 205
 206                                 //if the user only entered invalid groups
 207                                 if (globaldata->Groups.size() == 0) {
 208                                         cout << "When using the groups parameter you must have at least 1 valid group. I will run the command using all the groups in your groupfile." << endl;
 209                                         for (int i = 0; i < tmap->namesOfGroups.size(); i++) {
 210                                                 globaldata->Groups.push_back(tmap->namesOfGroups[i]);
 211                                                 numGroups++;
 212                                                 allGroups += tmap->namesOfGroups[i] + "-";
 213                                         }
 214                                         allGroups = allGroups.substr(0, allGroups.length()-1);
 215                                 }else {
 216                                         for (int i = 0; i < globaldata->Groups.size(); i++) {
 217                                                 allGroups += globaldata->Groups[i] + "-";
 218                                                 numGroups++;
 219                                         }
 220                                         allGroups = allGroups.substr(0, allGroups.length()-1);
 221                                 }
 222                         }else{//user has enter "all" and wants the default groups
 223                                 globaldata->Groups.clear();
 224                                 for (int i = 0; i < tmap->namesOfGroups.size(); i++) {
 225                                         globaldata->Groups.push_back(tmap->namesOfGroups[i]);
 226                                         numGroups++;
 227                                         allGroups += tmap->namesOfGroups[i] + "-";
 228                                 }
 229                                 allGroups = allGroups.substr(0, allGroups.length()-1);
 230                                 globaldata->setGroups("");
 231                         }
 232                 }else {
 233                         for (int i = 0; i < tmap->namesOfGroups.size(); i++) {
 234                                 allGroups += tmap->namesOfGroups[i] + "-";
 235                         }
 236                         allGroups = allGroups.substr(0, allGroups.length()-1);
 237                         numGroups = 1;
 238                 }
 239
 240                 //calculate number of comparsions
 241                 numComp = 0;
 242                 for (int r=0; r<numGroups; r++) {
 243                         for (int l = r+1; l < numGroups; l++) {
 244                                 groupComb.push_back(globaldata->Groups[r]+ "-" + globaldata->Groups[l]);
 245                                 numComp++;
 246                         }
 247                 }
 248
 249                 //ABC
 250                 if (numComp != 1) {
 251                         groupComb.push_back(allGroups);
 252                         numComp++;
 253                 }
 254         }
 255         catch(exception& e) {
 256                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracUnweightedCommand class Function setGroups. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 257                 exit(1);
 258         }
 259         catch(...) {
 260                 cout << "An unknown error has occurred in the UnifracUnweightedCommand class function setGroups. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 261                 exit(1);
 262         }
 263
 264 }
 265 /*****************************************************************/
 266
 267 void UnifracUnweightedCommand::initFile(string label){
 268         try {
 269                 if(counter != 0){
 270                         openOutputFile(unweightedFileout, out);
 271                         openInputFile(unweightedFile, inFile);
 272
 273                         string inputBuffer;
 274                         getline(inFile, inputBuffer);
 275
 276                         out     <<  inputBuffer << '\t' << label + "Score" << '\t' << label + "RandFreq" << '\t' << label + "RandCumul" << endl;
 277                 }else{
 278                         openOutputFile(unweightedFileout, out);
 279                         out     << label + "Score" << '\t' << label + "RandFreq" << '\t' << label + "RandCumul" << endl;
 280                 }
 281
 282                 out.setf(ios::fixed, ios::floatfield);
 283                 out.setf(ios::showpoint);
 284         }
 285         catch(exception& e) {
 286                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracUnweightedCommand class Function initFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 287                 exit(1);
 288         }
 289         catch(...) {
 290                 cout << "An unknown error has occurred in the UnifracUnweightedCommand class function initFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 291                 exit(1);
 292         }
 293 }
 294
 295 /***********************************************************************/
 296
 297 void UnifracUnweightedCommand::output(vector<double> data){
 298         try {
 299                 if(counter != 0){
 300                         string inputBuffer;
 301                         getline(inFile, inputBuffer);
 302
 303                         out     <<  inputBuffer << setprecision(globaldata->getIters().length()) << '\t' << data[0] << '\t' << data[1] << '\t' << data[2] << endl;
 304                 }
 305                 else{
 306                         out << setprecision(globaldata->getIters().length()) << data[0] << '\t' << data[1] << '\t' << data[2] << endl;
 307                 }
 308
 309         }
 310         catch(exception& e) {
 311                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracUnweightedCommand class Function output. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 312                 exit(1);
 313         }
 314         catch(...) {
 315                 cout << "An unknown error has occurred in the UnifracUnweightedCommand class function output. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 316                 exit(1);
 317         }
 318 };
 319
 320 /***********************************************************************/
 321
 322 void UnifracUnweightedCommand::resetFile(){
 323         try {
 324                 if(counter != 0){
 325                         out.close();
 326                         inFile.close();
 327                 }
 328                 else{
 329                         out.close();
 330                 }
 331                 counter = 1;
 332
 333                 remove(unweightedFile.c_str());
 334                 rename(unweightedFileout.c_str(), unweightedFile.c_str());
 335         }
 336         catch(exception& e) {
 337                 cout << "Standard Error: " << e.what() << " has occurred in the UnifracUnweightedCommand class Function resetFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 338                 exit(1);
 339         }
 340         catch(...) {
 341                 cout << "An unknown error has occurred in the UnifracUnweightedCommand class function resetFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 342                 exit(1);
 343         }
 344 }
 345