collectcommand.cpp

   1 /*
   2  *  collectcommand.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 "collectcommand.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 "qstat.h"
  22 #include "logsd.h"
  23 #include "bergerparker.h"
  24 #include "bstick.h"
  25 #include "goodscoverage.h"
  26 #include "efron.h"
  27 #include "boneh.h"
  28 #include "solow.h"
  29 #include "shen.h"
  30 #include "coverage.h"
  31
  32
  33 //**********************************************************************************************************************
  34 CollectCommand::CollectCommand(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[] =  {"freq","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 collect.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, "freq", false);                     if (temp == "not found") { temp = "100"; }
  87                         convert(temp, freq);
  88
  89                         temp = validParameter.validFile(parameters, "abund", false);            if (temp == "not found") { temp = "10"; }
  90                         convert(temp, abund);
  91
  92                         temp = validParameter.validFile(parameters, "size", false);                     if (temp == "not found") { temp = "0"; }
  93                         convert(temp, size);
  94
  95                         if (abort == false) {
  96                                 string fileNameRoot = getRootName(globaldata->inputFileName);
  97                                 int i;
  98                                 validCalculator = new ValidCalculators();
  99
 100                                 for (i=0; i<Estimators.size(); i++) {
 101                                         if (validCalculator->isValidCalculator("single", Estimators[i]) == true) {
 102                                                 if (Estimators[i] == "sobs") {
 103                                                         cDisplays.push_back(new CollectDisplay(new Sobs(), new OneColumnFile(fileNameRoot+"sobs")));
 104                                                 }else if (Estimators[i] == "chao") {
 105                                                         cDisplays.push_back(new CollectDisplay(new Chao1(), new ThreeColumnFile(fileNameRoot+"chao")));
 106                                                 }else if (Estimators[i] == "nseqs") {
 107                                                         cDisplays.push_back(new CollectDisplay(new NSeqs(), new OneColumnFile(fileNameRoot+"nseqs")));
 108                                                 }else if (Estimators[i] == "coverage") {
 109                                                         cDisplays.push_back(new CollectDisplay(new Coverage(), new OneColumnFile(fileNameRoot+"coverage")));
 110                                                 }else if (Estimators[i] == "ace") {
 111                                                         cDisplays.push_back(new CollectDisplay(new Ace(abund), new ThreeColumnFile(fileNameRoot+"ace")));
 112                                                 }else if (Estimators[i] == "jack") {
 113                                                         cDisplays.push_back(new CollectDisplay(new Jackknife(), new ThreeColumnFile(fileNameRoot+"jack")));
 114                                                 }else if (Estimators[i] == "shannon") {
 115                                                         cDisplays.push_back(new CollectDisplay(new Shannon(), new ThreeColumnFile(fileNameRoot+"shannon")));
 116                                                 }else if (Estimators[i] == "npshannon") {
 117                                                         cDisplays.push_back(new CollectDisplay(new NPShannon(), new OneColumnFile(fileNameRoot+"np_shannon")));
 118                                                 }else if (Estimators[i] == "simpson") {
 119                                                         cDisplays.push_back(new CollectDisplay(new Simpson(), new ThreeColumnFile(fileNameRoot+"simpson")));
 120                                                 }else if (Estimators[i] == "bootstrap") {
 121                                                         cDisplays.push_back(new CollectDisplay(new Bootstrap(), new OneColumnFile(fileNameRoot+"bootstrap")));
 122                                                 }else if (Estimators[i] == "geometric") {
 123                                                         cDisplays.push_back(new CollectDisplay(new Geom(), new OneColumnFile(fileNameRoot+"geometric")));
 124                                                 }else if (Estimators[i] == "qstat") {
 125                                                         cDisplays.push_back(new CollectDisplay(new QStat(), new OneColumnFile(fileNameRoot+"qstat")));
 126                                                 }else if (Estimators[i] == "logseries") {
 127                                                         cDisplays.push_back(new CollectDisplay(new LogSD(), new OneColumnFile(fileNameRoot+"logseries")));
 128                                                 }else if (Estimators[i] == "bergerparker") {
 129                                                         cDisplays.push_back(new CollectDisplay(new BergerParker(), new OneColumnFile(fileNameRoot+"bergerparker")));
 130                                                 }else if (Estimators[i] == "bstick") {
 131                                                         cDisplays.push_back(new CollectDisplay(new BStick(), new ThreeColumnFile(fileNameRoot+"bstick")));
 132                                                 }else if (Estimators[i] == "goodscoverage") {
 133                                                         cDisplays.push_back(new CollectDisplay(new GoodsCoverage(), new OneColumnFile(fileNameRoot+"goodscoverage")));
 134                                                 }else if (Estimators[i] == "efron") {
 135                                                         cDisplays.push_back(new CollectDisplay(new Efron(size), new OneColumnFile(fileNameRoot+"efron")));
 136                                                 }else if (Estimators[i] == "boneh") {
 137                                                         cDisplays.push_back(new CollectDisplay(new Boneh(size), new OneColumnFile(fileNameRoot+"boneh")));
 138                                                 }else if (Estimators[i] == "solow") {
 139                                                         cDisplays.push_back(new CollectDisplay(new Solow(size), new OneColumnFile(fileNameRoot+"solow")));
 140                                                 }else if (Estimators[i] == "shen") {
 141                                                         cDisplays.push_back(new CollectDisplay(new Shen(size, abund), new OneColumnFile(fileNameRoot+"shen")));
 142                                                 }
 143                                         }
 144                                 }
 145                         }
 146                 }
 147
 148         }
 149         catch(exception& e) {
 150                 errorOut(e, "CollectCommand", "CollectCommand");
 151                 exit(1);
 152         }
 153 }
 154 //**********************************************************************************************************************
 155
 156 void CollectCommand::help(){
 157         try {
 158                 mothurOut("The collect.single command can only be executed after a successful read.otu command. WITH ONE EXECEPTION. \n");
 159                 mothurOut("The collect.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 collect.single command parameters are label, freq, calc and abund.  No parameters are required. \n");
 161                 mothurOut("The collect.single command should be in the following format: \n");
 162                 mothurOut("collect.single(label=yourLabel, iters=yourIters, freq=yourFreq, calc=yourEstimators).\n");
 163                 mothurOut("Example collect(label=unique-.01-.03, iters=10000, freq=10, calc=sobs-chao-ace-jack).\n");
 164                 mothurOut("The default values for freq is 100, and calc are sobs-chao-ace-jack-shannon-npshannon-simpson.\n");
 165                 validCalculator->printCalc("single", cout);
 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. freq), '=' and parameters (i.e.yourFreq).\n\n");
 168         }
 169         catch(exception& e) {
 170                 errorOut(e, "CollectCommand", "help");
 171                 exit(1);
 172         }
 173 }
 174
 175 //**********************************************************************************************************************
 176
 177 CollectCommand::~CollectCommand(){
 178         if (abort == false) {
 179                 //delete order;
 180                 delete input;  globaldata->ginput = NULL;
 181                 delete read;
 182                 delete validCalculator;
 183                 globaldata->gorder = NULL;
 184         }
 185 }
 186
 187 //**********************************************************************************************************************
 188
 189 int CollectCommand::execute(){
 190         try {
 191
 192                 if (abort == true) { return 0; }
 193
 194                 //if the users entered no valid calculators don't execute command
 195                 if (cDisplays.size() == 0) { return 0; }
 196
 197                 read = new ReadOTUFile(globaldata->inputFileName);
 198                 read->read(&*globaldata);
 199
 200                 order = globaldata->gorder;
 201                 string lastLabel = order->getLabel();
 202                 input = globaldata->ginput;
 203
 204                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 205                 set<string> processedLabels;
 206                 set<string> userLabels = labels;
 207
 208                 while((order != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 209
 210                         if(allLines == 1 || labels.count(order->getLabel()) == 1){
 211
 212                                 cCurve = new Collect(order, cDisplays);
 213                                 cCurve->getCurve(freq);
 214                                 delete cCurve;
 215
 216                                 mothurOut(order->getLabel()); mothurOutEndLine();
 217                                 processedLabels.insert(order->getLabel());
 218                                 userLabels.erase(order->getLabel());
 219
 220
 221                         }
 222                         //you have a label the user want that is smaller than this label and the last label has not already been processed
 223                         if ((anyLabelsToProcess(order->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 224                                 string saveLabel = order->getLabel();
 225
 226                                 delete order;
 227                                 order = (input->getOrderVector(lastLabel));
 228
 229                                 cCurve = new Collect(order, cDisplays);
 230                                 cCurve->getCurve(freq);
 231                                 delete cCurve;
 232
 233                                 mothurOut(order->getLabel()); mothurOutEndLine();
 234                                 processedLabels.insert(order->getLabel());
 235                                 userLabels.erase(order->getLabel());
 236
 237                                 //restore real lastlabel to save below
 238                                 order->setLabel(saveLabel);
 239                         }
 240
 241                         lastLabel = order->getLabel();
 242
 243                         delete order;
 244                         order = (input->getOrderVector());
 245                 }
 246
 247                 //output error messages about any remaining user labels
 248                 set<string>::iterator it;
 249                 bool needToRun = false;
 250                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 251                         mothurOut("Your file does not include the label " + *it);
 252                         if (processedLabels.count(lastLabel) != 1) {
 253                                 mothurOut(". I will use " + lastLabel + "."); mothurOutEndLine();
 254                                 needToRun = true;
 255                         }else {
 256                                 mothurOut(". Please refer to " + lastLabel + "."); mothurOutEndLine();
 257                         }
 258                 }
 259
 260                 //run last label if you need to
 261                 if (needToRun == true)  {
 262                         if (order != NULL) {    delete order;   }
 263                         order = (input->getOrderVector(lastLabel));
 264
 265                         mothurOut(order->getLabel()); mothurOutEndLine();
 266
 267                         cCurve = new Collect(order, cDisplays);
 268                         cCurve->getCurve(freq);
 269                         delete cCurve;
 270                         delete order;
 271                 }
 272
 273
 274                 for(int i=0;i<cDisplays.size();i++){    delete cDisplays[i];    }
 275                 return 0;
 276         }
 277         catch(exception& e) {
 278                 errorOut(e, "CollectCommand", "execute");
 279                 exit(1);
 280         }
 281 }
 282
 283 //**********************************************************************************************************************