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(); 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->getListFile() == "") && (globaldata->getRabundFile() == "") && (globaldata->getSabundFile() == "")) { mothurOut("You must read a list, sabund or rabund 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                         if (abort == false) {
  93
  94                                 validCalculator = new ValidCalculators();
  95                                 int i;
  96
  97                                 for (i=0; i<Estimators.size(); i++) {
  98                                         if (validCalculator->isValidCalculator("summary", Estimators[i]) == true) {
  99                                                 if(Estimators[i] == "sobs"){
 100                                                         sumCalculators.push_back(new Sobs());
 101                                                 }else if(Estimators[i] == "chao"){
 102                                                         sumCalculators.push_back(new Chao1());
 103                                                 }else if(Estimators[i] == "coverage"){
 104                                                         sumCalculators.push_back(new Coverage());
 105                                                 }else if(Estimators[i] == "geometric"){
 106                                                         sumCalculators.push_back(new Geom());
 107                                                 }else if(Estimators[i] == "logseries"){
 108                                                         sumCalculators.push_back(new LogSD());
 109                                                 }else if(Estimators[i] == "qstat"){
 110                                                         sumCalculators.push_back(new QStat());
 111                                                 }else if(Estimators[i] == "bergerparker"){
 112                                                         sumCalculators.push_back(new BergerParker());
 113                                                 }else if(Estimators[i] == "bstick"){
 114                                                         sumCalculators.push_back(new BStick());
 115                                                 }else if(Estimators[i] == "ace"){
 116                                                         if(abund < 5)
 117                                                                 abund = 10;
 118                                                         sumCalculators.push_back(new Ace(abund));
 119                                                 }else if(Estimators[i] == "jack"){
 120                                                         sumCalculators.push_back(new Jackknife());
 121                                                 }else if(Estimators[i] == "shannon"){
 122                                                         sumCalculators.push_back(new Shannon());
 123                                                 }else if(Estimators[i] == "npshannon"){
 124                                                         sumCalculators.push_back(new NPShannon());
 125                                                 }else if(Estimators[i] == "simpson"){
 126                                                         sumCalculators.push_back(new Simpson());
 127                                                 }else if(Estimators[i] == "bootstrap"){
 128                                                         sumCalculators.push_back(new Bootstrap());
 129                                                 }else if (Estimators[i] == "nseqs") {
 130                                                         sumCalculators.push_back(new NSeqs());
 131                                                 }else if (Estimators[i] == "goodscoverage") {
 132                                                         sumCalculators.push_back(new GoodsCoverage());
 133                                                 }else if (Estimators[i] == "efron") {
 134                                                         sumCalculators.push_back(new Efron(size));
 135                                                 }else if (Estimators[i] == "boneh") {
 136                                                         sumCalculators.push_back(new Boneh(size));
 137                                                 }else if (Estimators[i] == "solow") {
 138                                                         sumCalculators.push_back(new Solow(size));
 139                                                 }else if (Estimators[i] == "shen") {
 140                                                         sumCalculators.push_back(new Shen(size, abund));
 141                                                 }
 142                                         }
 143                                 }
 144                         }
 145                 }
 146
 147
 148         }
 149         catch(exception& e) {
 150                 errorOut(e, "SummaryCommand", "SummaryCommand");
 151                 exit(1);
 152         }
 153 }
 154 //**********************************************************************************************************************
 155
 156 void SummaryCommand::help(){
 157         try {
 158                 mothurOut("The summary.single command can only be executed after a successful read.otu WTIH ONE EXECEPTION.\n");
 159                 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");
 160                 mothurOut("The summary.single command parameters are label, calc, abund.  No parameters are required.\n");
 161                 mothurOut("The summary.single command should be in the following format: \n");
 162                 mothurOut("summary.single(label=yourLabel, calc=yourEstimators).\n");
 163                 mothurOut("Example summary.single(label=unique-.01-.03, calc=sobs-chao-ace-jack-bootstrap-shannon-npshannon-simpson).\n");
 164                 validCalculator->printCalc("summary", cout);
 165                 mothurOut("The default value calc is sobs-chao-ace-jack-shannon-npshannon-simpson\n");
 166                 mothurOut("The label parameter is used to analyze specific labels in your input.\n");
 167                 mothurOut("Note: No spaces between parameter labels (i.e. label), '=' and parameters (i.e.yourLabels).\n\n");
 168         }
 169         catch(exception& e) {
 170                 errorOut(e, "SummaryCommand", "help");
 171                 exit(1);
 172         }
 173 }
 174
 175 //**********************************************************************************************************************
 176
 177 SummaryCommand::~SummaryCommand(){
 178         if (abort == false) {
 179                 delete input;  globaldata->ginput = NULL;
 180                 delete read;
 181                 delete validCalculator;
 182                 globaldata->sabund = NULL;
 183         }
 184 }
 185
 186 //**********************************************************************************************************************
 187
 188 int SummaryCommand::execute(){
 189         try {
 190
 191                 if (abort == true) { return 0; }
 192
 193                 //if the users entered no valid calculators don't execute command
 194                 if (sumCalculators.size() == 0) { return 0; }
 195
 196                 outputFileName = ((getRootName(globaldata->inputFileName)) + "summary");
 197                 openOutputFile(outputFileName, outputFileHandle);
 198                 outputFileHandle << "label";
 199
 200                 read = new ReadOTUFile(globaldata->inputFileName);
 201                 read->read(&*globaldata);
 202
 203                 sabund = globaldata->sabund;
 204                 string lastLabel = sabund->getLabel();
 205                 input = globaldata->ginput;
 206
 207                 for(int i=0;i<sumCalculators.size();i++){
 208                         if(sumCalculators[i]->getCols() == 1){
 209                                 outputFileHandle << '\t' << sumCalculators[i]->getName();
 210                         }
 211                         else{
 212                                 outputFileHandle << '\t' << sumCalculators[i]->getName() << "\t" << sumCalculators[i]->getName() << "_lci\t" << sumCalculators[i]->getName() << "_hci";
 213                         }
 214                 }
 215                 outputFileHandle << endl;
 216
 217                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 218                 set<string> processedLabels;
 219                 set<string> userLabels = labels;
 220
 221                 while((sabund != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 222
 223                         if(allLines == 1 || labels.count(sabund->getLabel()) == 1){
 224
 225                                 mothurOut(sabund->getLabel()); mothurOutEndLine();
 226                                 processedLabels.insert(sabund->getLabel());
 227                                 userLabels.erase(sabund->getLabel());
 228
 229                                 outputFileHandle << sabund->getLabel();
 230                                 for(int i=0;i<sumCalculators.size();i++){
 231                                         vector<double> data = sumCalculators[i]->getValues(sabund);
 232                                         outputFileHandle << '\t';
 233                                         sumCalculators[i]->print(outputFileHandle);
 234                                 }
 235                                 outputFileHandle << endl;
 236                         }
 237
 238                         if ((anyLabelsToProcess(sabund->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 239                                 delete sabund;
 240                                 sabund = input->getSAbundVector(lastLabel);
 241
 242                                 mothurOut(sabund->getLabel()); mothurOutEndLine();
 243                                 processedLabels.insert(sabund->getLabel());
 244                                 userLabels.erase(sabund->getLabel());
 245
 246                                 outputFileHandle << sabund->getLabel();
 247                                 for(int i=0;i<sumCalculators.size();i++){
 248                                         vector<double> data = sumCalculators[i]->getValues(sabund);
 249                                         outputFileHandle << '\t';
 250                                         sumCalculators[i]->print(outputFileHandle);
 251                                 }
 252                                 outputFileHandle << endl;
 253                         }
 254
 255                         lastLabel = sabund->getLabel();
 256
 257                         delete sabund;
 258                         sabund = input->getSAbundVector();
 259                 }
 260
 261                 //output error messages about any remaining user labels
 262                 set<string>::iterator it;
 263                 bool needToRun = false;
 264                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 265                         mothurOut("Your file does not include the label " + *it);
 266                         if (processedLabels.count(lastLabel) != 1) {
 267                                 mothurOut(". I will use " + lastLabel + "."); mothurOutEndLine();
 268                                 needToRun = true;
 269                         }else {
 270                                 mothurOut(". Please refer to " + lastLabel + "."); mothurOutEndLine();
 271                         }
 272                 }
 273
 274                 //run last label if you need to
 275                 if (needToRun == true)  {
 276                         if (sabund != NULL) {   delete sabund;  }
 277                         sabund = input->getSAbundVector(lastLabel);
 278
 279                         mothurOut(sabund->getLabel()); mothurOutEndLine();
 280                         outputFileHandle << sabund->getLabel();
 281                         for(int i=0;i<sumCalculators.size();i++){
 282                                 vector<double> data = sumCalculators[i]->getValues(sabund);
 283                                 outputFileHandle << '\t';
 284                                 sumCalculators[i]->print(outputFileHandle);
 285                         }
 286                         outputFileHandle << endl;
 287                         delete sabund;
 288                 }
 289
 290                 outputFileHandle.close();
 291
 292                 return 0;
 293         }
 294         catch(exception& e) {
 295                 errorOut(e, "SummaryCommand", "execute");
 296                 exit(1);
 297         }
 298 }
 299
 300 //**********************************************************************************************************************