rarefactcommand.cpp

   1 /*
   2  *  rarefactcommand.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 "rarefactcommand.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 "coverage.h"
  21
  22 //**********************************************************************************************************************
  23
  24
  25 RareFactCommand::RareFactCommand(string option){
  26         try {
  27                 globaldata = GlobalData::getInstance();
  28                 abort = false;
  29                 allLines = 1;
  30                 lines.clear();
  31                 labels.clear();
  32                 Estimators.clear();
  33
  34                 //allow user to run help
  35                 if(option == "help") { validCalculator = new ValidCalculators(); help(); abort = true; }
  36
  37                 else {
  38                         //valid paramters for this command
  39                         string Array[] =  {"iters","freq","line","label","calc","abund"};
  40                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  41
  42                         parser = new OptionParser();
  43                         parser->parse(option, parameters);  delete parser;
  44
  45                         ValidParameters* validParameter = new ValidParameters();
  46
  47                         //check to make sure all parameters are valid for command
  48                         for (it = parameters.begin(); it != parameters.end(); it++) {
  49                                 if (validParameter->isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  50                         }
  51
  52                         //make sure the user has already run the read.otu command
  53                         if ((globaldata->getListFile() == "") && (globaldata->getRabundFile() == "") && (globaldata->getSabundFile() == "")) { cout << "You must read a list, sabund or rabund before you can use the rarefaction.single command." << endl; abort = true; }
  54
  55                         //check for optional parameter and set defaults
  56                         // ...at some point should added some additional type checking...
  57                         line = validParameter->validFile(parameters, "line", false);
  58                         if (line == "not found") { line = "";  }
  59                         else {
  60                                 if(line != "all") {  splitAtDash(line, lines);  allLines = 0;  }
  61                                 else { allLines = 1;  }
  62                         }
  63
  64                         label = validParameter->validFile(parameters, "label", false);
  65                         if (label == "not found") { label = ""; }
  66                         else {
  67                                 if(label != "all") {  splitAtDash(label, labels);  allLines = 0;  }
  68                                 else { allLines = 1;  }
  69                         }
  70
  71                         //make sure user did not use both the line and label parameters
  72                         if ((line != "") && (label != "")) { cout << "You cannot use both the line and label parameters at the same time. " << endl; abort = true; }
  73                         //if the user has not specified any line or labels use the ones from read.otu
  74                         else if((line == "") && (label == "")) {
  75                                 allLines = globaldata->allLines;
  76                                 labels = globaldata->labels;
  77                                 lines = globaldata->lines;
  78                         }
  79
  80                         calc = validParameter->validFile(parameters, "calc", false);
  81                         if (calc == "not found") { calc = "sobs";  }
  82                         else {
  83                                  if (calc == "default")  {  calc = "sobs";  }
  84                         }
  85                         splitAtDash(calc, Estimators);
  86
  87                         string temp;
  88                         temp = validParameter->validFile(parameters, "freq", false);                    if (temp == "not found") { temp = "100"; }
  89                         convert(temp, freq);
  90
  91                         temp = validParameter->validFile(parameters, "abund", false);                   if (temp == "not found") { temp = "10"; }
  92                         convert(temp, abund);
  93
  94                         temp = validParameter->validFile(parameters, "iters", false);                   if (temp == "not found") { temp = "1000"; }
  95                         convert(temp, nIters);
  96
  97                         delete validParameter;
  98
  99                         if (abort == false) {
 100
 101                                 string fileNameRoot = getRootName(globaldata->inputFileName);
 102                                 int i;
 103                                 validCalculator = new ValidCalculators();
 104
 105
 106                                 for (i=0; i<Estimators.size(); i++) {
 107                                         if (validCalculator->isValidCalculator("rarefaction", Estimators[i]) == true) {
 108                                                 if (Estimators[i] == "sobs") {
 109                                                         rDisplays.push_back(new RareDisplay(new Sobs(), new ThreeColumnFile(fileNameRoot+"rarefaction")));
 110                                                 }else if (Estimators[i] == "chao") {
 111                                                         rDisplays.push_back(new RareDisplay(new Chao1(), new ThreeColumnFile(fileNameRoot+"r_chao")));
 112                                                 }else if (Estimators[i] == "ace") {
 113                                                         if(abund < 5)
 114                                                                 abund = 10;
 115                                                         rDisplays.push_back(new RareDisplay(new Ace(abund), new ThreeColumnFile(fileNameRoot+"r_ace")));
 116                                                 }else if (Estimators[i] == "jack") {
 117                                                         rDisplays.push_back(new RareDisplay(new Jackknife(), new ThreeColumnFile(fileNameRoot+"r_jack")));
 118                                                 }else if (Estimators[i] == "shannon") {
 119                                                         rDisplays.push_back(new RareDisplay(new Shannon(), new ThreeColumnFile(fileNameRoot+"r_shannon")));
 120                                                 }else if (Estimators[i] == "npshannon") {
 121                                                         rDisplays.push_back(new RareDisplay(new NPShannon(), new ThreeColumnFile(fileNameRoot+"r_npshannon")));
 122                                                 }else if (Estimators[i] == "simpson") {
 123                                                         rDisplays.push_back(new RareDisplay(new Simpson(), new ThreeColumnFile(fileNameRoot+"r_simpson")));
 124                                                 }else if (Estimators[i] == "bootstrap") {
 125                                                         rDisplays.push_back(new RareDisplay(new Bootstrap(), new ThreeColumnFile(fileNameRoot+"r_bootstrap")));
 126                                                 }else if (Estimators[i] == "coverage") {
 127                                                         rDisplays.push_back(new RareDisplay(new Coverage(), new ThreeColumnFile(fileNameRoot+"r_coverage")));
 128                                                 }else if (Estimators[i] == "nseqs") {
 129                                                         rDisplays.push_back(new RareDisplay(new NSeqs(), new ThreeColumnFile(fileNameRoot+"r_nseqs")));
 130                                                 }
 131                                         }
 132                                 }
 133                         }
 134
 135                 }
 136
 137         }
 138         catch(exception& e) {
 139                 cout << "Standard Error: " << e.what() << " has occurred in the RareFactCommand class Function RareFactCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 140                 exit(1);
 141         }
 142         catch(...) {
 143                 cout << "An unknown error has occurred in the RareFactCommand class function RareFactCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 144                 exit(1);
 145         }
 146
 147 }
 148 //**********************************************************************************************************************
 149
 150 void RareFactCommand::help(){
 151         try {
 152                 cout << "The rarefaction.single command can only be executed after a successful read.otu WTIH ONE EXECEPTION." << "\n";
 153                 cout << "The rarefaction.single command can be executed after a successful cluster command.  It will use the .list file from the output of the cluster." << "\n";
 154                 cout << "The rarefaction.single command parameters are label, line, iters, freq, calc and abund.  No parameters are required, but you may not use " << "\n";
 155                 cout << "both the line and label parameters at the same time. The rarefaction.single command should be in the following format: " << "\n";
 156                 cout << "rarefaction.single(label=yourLabel, line=yourLines, iters=yourIters, freq=yourFreq, calc=yourEstimators)." << "\n";
 157                 cout << "Example rarefaction.single(label=unique-.01-.03, line=0-5-10, iters=10000, freq=10, calc=sobs-rchao-race-rjack-rbootstrap-rshannon-rnpshannon-rsimpson)." << "\n";
 158                 cout << "The default values for iters is 1000, freq is 100, and calc is rarefaction which calculates the rarefaction curve for the observed richness." << "\n";
 159                 validCalculator->printCalc("rarefaction", cout);
 160                 cout << "The label and line parameters are used to analyze specific lines in your input." << "\n";
 161                 cout << "Note: No spaces between parameter labels (i.e. freq), '=' and parameters (i.e.yourFreq)." << "\n" << "\n";
 162         }
 163         catch(exception& e) {
 164                 cout << "Standard Error: " << e.what() << " has occurred in the RareFactCommand class Function help. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 165                 exit(1);
 166         }
 167         catch(...) {
 168                 cout << "An unknown error has occurred in the RareFactCommand class function help. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 169                 exit(1);
 170         }
 171 }
 172
 173 //**********************************************************************************************************************
 174
 175 RareFactCommand::~RareFactCommand(){
 176         delete order;
 177         delete input;
 178         delete rCurve;
 179         delete read;
 180         delete validCalculator;
 181 }
 182
 183 //**********************************************************************************************************************
 184
 185 int RareFactCommand::execute(){
 186         try {
 187
 188                 if (abort == true) { return 0; }
 189
 190                 int count = 1;
 191
 192                 //if the users entered no valid calculators don't execute command
 193                 if (rDisplays.size() == 0) { return 0; }
 194
 195                 read = new ReadOTUFile(globaldata->inputFileName);
 196                 read->read(&*globaldata);
 197
 198                 order = globaldata->gorder;
 199                 OrderVector* lastOrder = order;
 200                 input = globaldata->ginput;
 201
 202                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 203                 set<string> processedLabels;
 204                 set<string> userLabels = labels;
 205                 set<int> userLines = lines;
 206
 207                 //as long as you are not at the end of the file or done wih the lines you want
 208                 while((order != NULL) && ((allLines == 1) || (userLabels.size() != 0) || (userLines.size() != 0))) {
 209
 210                         if(allLines == 1 || lines.count(count) == 1 || labels.count(order->getLabel()) == 1){
 211
 212                                 rCurve = new Rarefact(order, rDisplays);
 213                                 rCurve->getCurve(freq, nIters);
 214                                 delete rCurve;
 215
 216                                 cout << order->getLabel() << '\t' << count << endl;
 217                                 processedLabels.insert(order->getLabel());
 218                                 userLabels.erase(order->getLabel());
 219                                 userLines.erase(count);
 220                         }
 221
 222                         if ((anyLabelsToProcess(order->getLabel(), userLabels, "") == true) && (processedLabels.count(lastOrder->getLabel()) != 1)) {
 223                                 rCurve = new Rarefact(lastOrder, rDisplays);
 224                                 rCurve->getCurve(freq, nIters);
 225                                 delete rCurve;
 226
 227                                 cout << lastOrder->getLabel() << '\t' << count << endl;
 228                                 processedLabels.insert(lastOrder->getLabel());
 229                                 userLabels.erase(lastOrder->getLabel());
 230                         }
 231
 232                         if (count != 1) { delete lastOrder; }
 233                         lastOrder = order;
 234
 235                         order = (input->getOrderVector());
 236                         count++;
 237                 }
 238
 239                 //output error messages about any remaining user labels
 240                 set<string>::iterator it;
 241                 bool needToRun = false;
 242                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 243                         cout << "Your file does not include the label "<< *it;
 244                         if (processedLabels.count(lastOrder->getLabel()) != 1) {
 245                                 cout << ". I will use " << lastOrder->getLabel() << "." << endl;
 246                                 needToRun = true;
 247                         }else {
 248                                 cout << ". Please refer to " << lastOrder->getLabel() << "." << endl;
 249                         }
 250                 }
 251
 252                 //run last line if you need to
 253                 if (needToRun == true)  {
 254                         rCurve = new Rarefact(lastOrder, rDisplays);
 255                         rCurve->getCurve(freq, nIters);
 256                         delete rCurve;
 257
 258                         cout << lastOrder->getLabel() << '\t' << count << endl;
 259                 }
 260
 261                 delete lastOrder;
 262
 263                 for(int i=0;i<rDisplays.size();i++){    delete rDisplays[i];    }
 264                 return 0;
 265         }
 266         catch(exception& e) {
 267                 cout << "Standard Error: " << e.what() << " has occurred in the RareFactCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 268                 exit(1);
 269         }
 270         catch(...) {
 271                 cout << "An unknown error has occurred in the RareFactCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 272                 exit(1);
 273         }
 274 }
 275
 276 //**********************************************************************************************************************