summarycommand.cpp

   1 /*
   2  *  summarycommand.cpp
   3  *  Dotur
   4  *
   5  *  Created by Sarah Westcott on 1/2/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10 #include "summarycommand.h"
  11 #include "ace.h"
  12 #include "sobs.h"
  13 #include "nseqs.h"
  14 #include "chao1.h"
  15 #include "bootstrap.h"
  16 #include "simpson.h"
  17 #include "npshannon.h"
  18 #include "shannon.h"
  19 #include "jackknife.h"
  20 #include "geom.h"
  21 #include "logsd.h"
  22 #include "qstat.h"
  23 #include "bergerparker.h"
  24 #include "bstick.h"
  25 #include "goodscoverage.h"
  26 #include "coverage.h"
  27 #include "efron.h"
  28 #include "boneh.h"
  29 #include "solow.h"
  30 #include "shen.h"
  31
  32 //**********************************************************************************************************************
  33
  34 SummaryCommand::SummaryCommand(string option){
  35         try {
  36                 globaldata = GlobalData::getInstance();
  37                 abort = false;
  38                 allLines = 1;
  39                 labels.clear();
  40                 Estimators.clear();
  41
  42                 //allow user to run help
  43                 if(option == "help") { validCalculator = new ValidCalculators(); help(); delete validCalculator; abort = true; }
  44
  45                 else {
  46                         //valid paramters for this command
  47                         string Array[] =  {"label","calc","abund","size"};
  48                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  49
  50                         OptionParser parser(option);
  51                         map<string,string> parameters = parser.getParameters();
  52
  53                         ValidParameters validParameter;
  54
  55                         //check to make sure all parameters are valid for command
  56                         for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
  57                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  58                         }
  59
  60                         //make sure the user has already run the read.otu command
  61                         if ((globaldata->getSharedFile() == "") && (globaldata->getListFile() == "") && (globaldata->getRabundFile() == "") && (globaldata->getSabundFile() == "")) { mothurOut("You must read a list, sabund, rabund or shared file before you can use the summary.single command."); mothurOutEndLine(); abort = true; }
  62
  63                         //check for optional parameter and set defaults
  64                         // ...at some point should added some additional type checking...
  65                         label = validParameter.validFile(parameters, "label", false);
  66                         if (label == "not found") { label = ""; }
  67                         else {
  68                                 if(label != "all") {  splitAtDash(label, labels);  allLines = 0;  }
  69                                 else { allLines = 1;  }
  70                         }
  71
  72                         //if the user has not specified any labels use the ones from read.otu
  73                         if(label == "") {
  74                                 allLines = globaldata->allLines;
  75                                 labels = globaldata->labels;
  76                         }
  77
  78                         calc = validParameter.validFile(parameters, "calc", false);
  79                         if (calc == "not found") { calc = "sobs-chao-ace-jack-shannon-npshannon-simpson";  }
  80                         else {
  81                                  if (calc == "default")  {  calc = "sobs-chao-ace-jack-shannon-npshannon-simpson";  }
  82                         }
  83                         splitAtDash(calc, Estimators);
  84
  85                         string temp;
  86                         temp = validParameter.validFile(parameters, "abund", false);            if (temp == "not found") { temp = "10"; }
  87                         convert(temp, abund);
  88
  89                         temp = validParameter.validFile(parameters, "size", false);                     if (temp == "not found") { temp = "0"; }
  90                         convert(temp, size);
  91
  92                 }
  93         }
  94         catch(exception& e) {
  95                 errorOut(e, "SummaryCommand", "SummaryCommand");
  96                 exit(1);
  97         }
  98 }
  99 //**********************************************************************************************************************
 100
 101 void SummaryCommand::help(){
 102         try {
 103                 mothurOut("The summary.single command can only be executed after a successful read.otu WTIH ONE EXECEPTION.\n");
 104                 mothurOut("The summary.single command can be executed after a successful cluster command.  It will use the .list file from the output of the cluster.\n");
 105                 mothurOut("The summary.single command parameters are label, calc, abund.  No parameters are required.\n");
 106                 mothurOut("The summary.single command should be in the following format: \n");
 107                 mothurOut("summary.single(label=yourLabel, calc=yourEstimators).\n");
 108                 mothurOut("Example summary.single(label=unique-.01-.03, calc=sobs-chao-ace-jack-bootstrap-shannon-npshannon-simpson).\n");
 109                 validCalculator->printCalc("summary", cout);
 110                 mothurOut("The default value calc is sobs-chao-ace-jack-shannon-npshannon-simpson\n");
 111                 mothurOut("The label parameter is used to analyze specific labels in your input.\n");
 112                 mothurOut("Note: No spaces between parameter labels (i.e. label), '=' and parameters (i.e.yourLabels).\n\n");
 113         }
 114         catch(exception& e) {
 115                 errorOut(e, "SummaryCommand", "help");
 116                 exit(1);
 117         }
 118 }
 119
 120 //**********************************************************************************************************************
 121
 122 SummaryCommand::~SummaryCommand(){}
 123
 124 //**********************************************************************************************************************
 125
 126 int SummaryCommand::execute(){
 127         try {
 128
 129                 if (abort == true) { return 0; }
 130
 131                 if ((globaldata->getFormat() != "sharedfile")) { inputFileNames.push_back(globaldata->inputFileName);  }
 132                 else {  inputFileNames = parseSharedFile(globaldata->getSharedFile());  globaldata->setFormat("rabund");  }
 133
 134                 for (int p = 0; p < inputFileNames.size(); p++) {
 135
 136                         string fileNameRoot = getRootName(inputFileNames[p]) + "summary";
 137                         globaldata->inputFileName = inputFileNames[p];
 138
 139                         if (inputFileNames.size() > 1) {
 140                                 mothurOutEndLine(); mothurOut("Processing group " + groups[p]); mothurOutEndLine(); mothurOutEndLine();
 141                         }
 142
 143
 144                         validCalculator = new ValidCalculators();
 145
 146                         for (int i=0; i<Estimators.size(); i++) {
 147                                 if (validCalculator->isValidCalculator("summary", Estimators[i]) == true) {
 148                                         if(Estimators[i] == "sobs"){
 149                                                 sumCalculators.push_back(new Sobs());
 150                                         }else if(Estimators[i] == "chao"){
 151                                                 sumCalculators.push_back(new Chao1());
 152                                         }else if(Estimators[i] == "coverage"){
 153                                                 sumCalculators.push_back(new Coverage());
 154                                         }else if(Estimators[i] == "geometric"){
 155                                                 sumCalculators.push_back(new Geom());
 156                                         }else if(Estimators[i] == "logseries"){
 157                                                 sumCalculators.push_back(new LogSD());
 158                                         }else if(Estimators[i] == "qstat"){
 159                                                 sumCalculators.push_back(new QStat());
 160                                         }else if(Estimators[i] == "bergerparker"){
 161                                                 sumCalculators.push_back(new BergerParker());
 162                                         }else if(Estimators[i] == "bstick"){
 163                                                 sumCalculators.push_back(new BStick());
 164                                         }else if(Estimators[i] == "ace"){
 165                                                 if(abund < 5)
 166                                                         abund = 10;
 167                                                 sumCalculators.push_back(new Ace(abund));
 168                                         }else if(Estimators[i] == "jack"){
 169                                                 sumCalculators.push_back(new Jackknife());
 170                                         }else if(Estimators[i] == "shannon"){
 171                                                 sumCalculators.push_back(new Shannon());
 172                                         }else if(Estimators[i] == "npshannon"){
 173                                                 sumCalculators.push_back(new NPShannon());
 174                                         }else if(Estimators[i] == "simpson"){
 175                                                 sumCalculators.push_back(new Simpson());
 176                                         }else if(Estimators[i] == "bootstrap"){
 177                                                 sumCalculators.push_back(new Bootstrap());
 178                                         }else if (Estimators[i] == "nseqs") {
 179                                                 sumCalculators.push_back(new NSeqs());
 180                                         }else if (Estimators[i] == "goodscoverage") {
 181                                                 sumCalculators.push_back(new GoodsCoverage());
 182                                         }else if (Estimators[i] == "efron") {
 183                                                 sumCalculators.push_back(new Efron(size));
 184                                         }else if (Estimators[i] == "boneh") {
 185                                                 sumCalculators.push_back(new Boneh(size));
 186                                         }else if (Estimators[i] == "solow") {
 187                                                 sumCalculators.push_back(new Solow(size));
 188                                         }else if (Estimators[i] == "shen") {
 189                                                 sumCalculators.push_back(new Shen(size, abund));
 190                                         }
 191                                 }
 192                         }
 193
 194                         //if the users entered no valid calculators don't execute command
 195                         if (sumCalculators.size() == 0) { return 0; }
 196
 197                         ofstream outputFileHandle;
 198                         openOutputFile(fileNameRoot, outputFileHandle);
 199                         outputFileHandle << "label";
 200
 201                         read = new ReadOTUFile(globaldata->inputFileName);
 202                         read->read(&*globaldata);
 203
 204                         sabund = globaldata->sabund;
 205                         string lastLabel = sabund->getLabel();
 206                         input = globaldata->ginput;
 207
 208                         for(int i=0;i<sumCalculators.size();i++){
 209                                 if(sumCalculators[i]->getCols() == 1){
 210                                         outputFileHandle << '\t' << sumCalculators[i]->getName();
 211                                 }
 212                                 else{
 213                                         outputFileHandle << '\t' << sumCalculators[i]->getName() << "\t" << sumCalculators[i]->getName() << "_lci\t" << sumCalculators[i]->getName() << "_hci";
 214                                 }
 215                         }
 216                         outputFileHandle << endl;
 217
 218                         //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 219                         set<string> processedLabels;
 220                         set<string> userLabels = labels;
 221
 222                         while((sabund != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 223
 224                                 if(allLines == 1 || labels.count(sabund->getLabel()) == 1){
 225
 226                                         mothurOut(sabund->getLabel()); mothurOutEndLine();
 227                                         processedLabels.insert(sabund->getLabel());
 228                                         userLabels.erase(sabund->getLabel());
 229
 230                                         outputFileHandle << sabund->getLabel();
 231                                         for(int i=0;i<sumCalculators.size();i++){
 232                                                 vector<double> data = sumCalculators[i]->getValues(sabund);
 233                                                 outputFileHandle << '\t';
 234                                                 sumCalculators[i]->print(outputFileHandle);
 235                                         }
 236                                         outputFileHandle << endl;
 237                                 }
 238
 239                                 if ((anyLabelsToProcess(sabund->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 240                                         string saveLabel = sabund->getLabel();
 241
 242                                         delete sabund;
 243                                         sabund = input->getSAbundVector(lastLabel);
 244
 245                                         mothurOut(sabund->getLabel()); mothurOutEndLine();
 246                                         processedLabels.insert(sabund->getLabel());
 247                                         userLabels.erase(sabund->getLabel());
 248
 249                                         outputFileHandle << sabund->getLabel();
 250                                         for(int i=0;i<sumCalculators.size();i++){
 251                                                 vector<double> data = sumCalculators[i]->getValues(sabund);
 252                                                 outputFileHandle << '\t';
 253                                                 sumCalculators[i]->print(outputFileHandle);
 254                                         }
 255                                         outputFileHandle << endl;
 256
 257                                         //restore real lastlabel to save below
 258                                         sabund->setLabel(saveLabel);
 259                                 }
 260
 261                                 lastLabel = sabund->getLabel();
 262
 263                                 delete sabund;
 264                                 sabund = input->getSAbundVector();
 265                         }
 266
 267                         //output error messages about any remaining user labels
 268                         set<string>::iterator it;
 269                         bool needToRun = false;
 270                         for (it = userLabels.begin(); it != userLabels.end(); it++) {
 271                                 mothurOut("Your file does not include the label " + *it);
 272                                 if (processedLabels.count(lastLabel) != 1) {
 273                                         mothurOut(". I will use " + lastLabel + "."); mothurOutEndLine();
 274                                         needToRun = true;
 275                                 }else {
 276                                         mothurOut(". Please refer to " + lastLabel + "."); mothurOutEndLine();
 277                                 }
 278                         }
 279
 280                         //run last label if you need to
 281                         if (needToRun == true)  {
 282                                 if (sabund != NULL) {   delete sabund;  }
 283                                 sabund = input->getSAbundVector(lastLabel);
 284
 285                                 mothurOut(sabund->getLabel()); mothurOutEndLine();
 286                                 outputFileHandle << sabund->getLabel();
 287                                 for(int i=0;i<sumCalculators.size();i++){
 288                                         vector<double> data = sumCalculators[i]->getValues(sabund);
 289                                         outputFileHandle << '\t';
 290                                         sumCalculators[i]->print(outputFileHandle);
 291                                 }
 292                                 outputFileHandle << endl;
 293                                 delete sabund;
 294                         }
 295
 296                         outputFileHandle.close();
 297
 298                         delete input;  globaldata->ginput = NULL;
 299                         delete read;
 300                         delete validCalculator;
 301                         globaldata->sabund = NULL;
 302                 }
 303
 304                 return 0;
 305         }
 306         catch(exception& e) {
 307                 errorOut(e, "SummaryCommand", "execute");
 308                 exit(1);
 309         }
 310 }
 311 //**********************************************************************************************************************
 312 vector<string> SummaryCommand::parseSharedFile(string filename) {
 313         try {
 314                 vector<string> filenames;
 315
 316                 map<string, ofstream*> filehandles;
 317                 map<string, ofstream*>::iterator it3;
 318
 319
 320                 //read first line
 321                 read = new ReadOTUFile(filename);
 322                 read->read(&*globaldata);
 323
 324                 input = globaldata->ginput;
 325                 vector<SharedRAbundVector*> lookup = input->getSharedRAbundVectors();
 326
 327                 string sharedFileRoot = getRootName(filename);
 328
 329                 //clears file before we start to write to it below
 330                 for (int i=0; i<lookup.size(); i++) {
 331                         remove((sharedFileRoot + lookup[i]->getGroup() + ".rabund").c_str());
 332                         filenames.push_back((sharedFileRoot + lookup[i]->getGroup() + ".rabund"));
 333                 }
 334
 335                 ofstream* temp;
 336                 for (int i=0; i<lookup.size(); i++) {
 337                         temp = new ofstream;
 338                         filehandles[lookup[i]->getGroup()] = temp;
 339                         groups.push_back(lookup[i]->getGroup());
 340                 }
 341
 342                 while(lookup[0] != NULL) {
 343
 344                         for (int i = 0; i < lookup.size(); i++) {
 345                                 RAbundVector rav = lookup[i]->getRAbundVector();
 346                                 openOutputFileAppend(sharedFileRoot + lookup[i]->getGroup() + ".rabund", *(filehandles[lookup[i]->getGroup()]));
 347                                 rav.print(*(filehandles[lookup[i]->getGroup()]));
 348                                 (*(filehandles[lookup[i]->getGroup()])).close();
 349                         }
 350
 351                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 352                         lookup = input->getSharedRAbundVectors();
 353                 }
 354
 355                 //free memory
 356                 for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 357                         delete it3->second;
 358                 }
 359                 delete read;
 360                 delete input;
 361                 globaldata->ginput = NULL;
 362
 363                 return filenames;
 364         }
 365         catch(exception& e) {
 366                 errorOut(e, "SummaryCommand", "parseSharedFile");
 367                 exit(1);
 368         }
 369 }
 370 //**********************************************************************************************************************