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                 lines.clear();
  40                 labels.clear();
  41                 Estimators.clear();
  42
  43                 //allow user to run help
  44                 if(option == "help") { validCalculator = new ValidCalculators(); help(); abort = true; }
  45
  46                 else {
  47                         //valid paramters for this command
  48                         string Array[] =  {"freq","line","label","calc","abund","size"};
  49                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  50
  51                         OptionParser parser(option);
  52                         map<string,string> parameters = parser.getParameters();
  53
  54                         ValidParameters validParameter;
  55
  56                         //check to make sure all parameters are valid for command
  57                         for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
  58                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  59                         }
  60
  61                         //make sure the user has already run the read.otu command
  62                         if ((globaldata->getListFile() == "") && (globaldata->getRabundFile() == "") && (globaldata->getSabundFile() == "")) { cout << "You must read a list, sabund or rabund before you can use the collect.single command." << endl; abort = true; }
  63
  64                         //check for optional parameter and set defaults
  65                         // ...at some point should added some additional type checking...
  66                         line = validParameter.validFile(parameters, "line", false);
  67                         if (line == "not found") { line = "";  }
  68                         else {
  69                                 if(line != "all") {  splitAtDash(line, lines);  allLines = 0;  }
  70                                 else { allLines = 1;  }
  71                         }
  72
  73                         label = validParameter.validFile(parameters, "label", false);
  74                         if (label == "not found") { label = ""; }
  75                         else {
  76                                 if(label != "all") {  splitAtDash(label, labels);  allLines = 0;  }
  77                                 else { allLines = 1;  }
  78                         }
  79
  80                         //make sure user did not use both the line and label parameters
  81                         if ((line != "") && (label != "")) { cout << "You cannot use both the line and label parameters at the same time. " << endl; abort = true; }
  82                         //if the user has not specified any line or labels use the ones from read.otu
  83                         else if((line == "") && (label == "")) {
  84                                 allLines = globaldata->allLines;
  85                                 labels = globaldata->labels;
  86                                 lines = globaldata->lines;
  87                         }
  88
  89                         calc = validParameter.validFile(parameters, "calc", false);
  90                         if (calc == "not found") { calc = "sobs-chao-ace-jack-shannon-npshannon-simpson";  }
  91                         else {
  92                                  if (calc == "default")  {  calc = "sobs-chao-ace-jack-shannon-npshannon-simpson";  }
  93                         }
  94                         splitAtDash(calc, Estimators);
  95
  96                         string temp;
  97                         temp = validParameter.validFile(parameters, "freq", false);                     if (temp == "not found") { temp = "100"; }
  98                         convert(temp, freq);
  99
 100                         temp = validParameter.validFile(parameters, "abund", false);            if (temp == "not found") { temp = "10"; }
 101                         convert(temp, abund);
 102
 103                         temp = validParameter.validFile(parameters, "size", false);                     if (temp == "not found") { temp = "0"; }
 104                         convert(temp, size);
 105
 106                         if (abort == false) {
 107                                 string fileNameRoot = getRootName(globaldata->inputFileName);
 108                                 int i;
 109                                 validCalculator = new ValidCalculators();
 110
 111                                 for (i=0; i<Estimators.size(); i++) {
 112                                         if (validCalculator->isValidCalculator("single", Estimators[i]) == true) {
 113                                                 if (Estimators[i] == "sobs") {
 114                                                         cDisplays.push_back(new CollectDisplay(new Sobs(), new OneColumnFile(fileNameRoot+"sobs")));
 115                                                 }else if (Estimators[i] == "chao") {
 116                                                         cDisplays.push_back(new CollectDisplay(new Chao1(), new ThreeColumnFile(fileNameRoot+"chao")));
 117                                                 }else if (Estimators[i] == "nseqs") {
 118                                                         cDisplays.push_back(new CollectDisplay(new NSeqs(), new OneColumnFile(fileNameRoot+"nseqs")));
 119                                                 }else if (Estimators[i] == "coverage") {
 120                                                         cDisplays.push_back(new CollectDisplay(new Coverage(), new OneColumnFile(fileNameRoot+"coverage")));
 121                                                 }else if (Estimators[i] == "ace") {
 122                                                         cDisplays.push_back(new CollectDisplay(new Ace(abund), new ThreeColumnFile(fileNameRoot+"ace")));
 123                                                 }else if (Estimators[i] == "jack") {
 124                                                         cDisplays.push_back(new CollectDisplay(new Jackknife(), new ThreeColumnFile(fileNameRoot+"jack")));
 125                                                 }else if (Estimators[i] == "shannon") {
 126                                                         cDisplays.push_back(new CollectDisplay(new Shannon(), new ThreeColumnFile(fileNameRoot+"shannon")));
 127                                                 }else if (Estimators[i] == "npshannon") {
 128                                                         cDisplays.push_back(new CollectDisplay(new NPShannon(), new OneColumnFile(fileNameRoot+"np_shannon")));
 129                                                 }else if (Estimators[i] == "simpson") {
 130                                                         cDisplays.push_back(new CollectDisplay(new Simpson(), new ThreeColumnFile(fileNameRoot+"simpson")));
 131                                                 }else if (Estimators[i] == "bootstrap") {
 132                                                         cDisplays.push_back(new CollectDisplay(new Bootstrap(), new OneColumnFile(fileNameRoot+"bootstrap")));
 133                                                 }else if (Estimators[i] == "geometric") {
 134                                                         cDisplays.push_back(new CollectDisplay(new Geom(), new OneColumnFile(fileNameRoot+"geometric")));
 135                                                 }else if (Estimators[i] == "qstat") {
 136                                                         cDisplays.push_back(new CollectDisplay(new QStat(), new OneColumnFile(fileNameRoot+"qstat")));
 137                                                 }else if (Estimators[i] == "logseries") {
 138                                                         cDisplays.push_back(new CollectDisplay(new LogSD(), new OneColumnFile(fileNameRoot+"logseries")));
 139                                                 }else if (Estimators[i] == "bergerparker") {
 140                                                         cDisplays.push_back(new CollectDisplay(new BergerParker(), new OneColumnFile(fileNameRoot+"bergerparker")));
 141                                                 }else if (Estimators[i] == "bstick") {
 142                                                         cDisplays.push_back(new CollectDisplay(new BStick(), new ThreeColumnFile(fileNameRoot+"bstick")));
 143                                                 }else if (Estimators[i] == "goodscoverage") {
 144                                                         cDisplays.push_back(new CollectDisplay(new GoodsCoverage(), new OneColumnFile(fileNameRoot+"goodscoverage")));
 145                                                 }else if (Estimators[i] == "efron") {
 146                                                         cDisplays.push_back(new CollectDisplay(new Efron(size), new OneColumnFile(fileNameRoot+"efron")));
 147                                                 }else if (Estimators[i] == "boneh") {
 148                                                         cDisplays.push_back(new CollectDisplay(new Boneh(size), new OneColumnFile(fileNameRoot+"boneh")));
 149                                                 }else if (Estimators[i] == "solow") {
 150                                                         cDisplays.push_back(new CollectDisplay(new Solow(size), new OneColumnFile(fileNameRoot+"solow")));
 151                                                 }else if (Estimators[i] == "shen") {
 152                                                         cDisplays.push_back(new CollectDisplay(new Shen(size, abund), new OneColumnFile(fileNameRoot+"shen")));
 153                                                 }
 154                                         }
 155                                 }
 156                         }
 157                 }
 158
 159         }
 160         catch(exception& e) {
 161                 cout << "Standard Error: " << e.what() << " has occurred in the CollectCommand class Function CollectCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 162                 exit(1);
 163         }
 164         catch(...) {
 165                 cout << "An unknown error has occurred in the CollectCommand class function CollectCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 166                 exit(1);
 167         }
 168
 169 }
 170 //**********************************************************************************************************************
 171
 172 void CollectCommand::help(){
 173         try {
 174                 cout << "The collect.single command can only be executed after a successful read.otu command. WITH ONE EXECEPTION. " << "\n";
 175                 cout << "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";
 176                 cout << "The collect.single command parameters are label, line, freq, calc and abund.  No parameters are required, but you may not use " << "\n";
 177                 cout << "both the line and label parameters at the same time. The collect.single command should be in the following format: " << "\n";
 178                 cout << "collect.single(label=yourLabel, line=yourLines, iters=yourIters, freq=yourFreq, calc=yourEstimators)." << "\n";
 179                 cout << "Example collect(label=unique-.01-.03, line=0-5-10, iters=10000, freq=10, calc=sobs-chao-ace-jack)." << "\n";
 180                 cout << "The default values for freq is 100, and calc are sobs-chao-ace-jack-shannon-npshannon-simpson." << "\n";
 181                 validCalculator->printCalc("single", cout);
 182                 cout << "The label and line parameters are used to analyze specific lines in your input." << "\n";
 183                 cout << "Note: No spaces between parameter labels (i.e. freq), '=' and parameters (i.e.yourFreq)." << "\n" << "\n";
 184         }
 185         catch(exception& e) {
 186                 cout << "Standard Error: " << e.what() << " has occurred in the CollectCommand class Function help. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 187                 exit(1);
 188         }
 189         catch(...) {
 190                 cout << "An unknown error has occurred in the CollectCommand class function help. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 191                 exit(1);
 192         }
 193 }
 194
 195 //**********************************************************************************************************************
 196
 197 CollectCommand::~CollectCommand(){
 198         delete order;
 199         delete input;
 200         delete cCurve;
 201         delete read;
 202         delete validCalculator;
 203 }
 204
 205 //**********************************************************************************************************************
 206
 207 int CollectCommand::execute(){
 208         try {
 209
 210                 if (abort == true) { return 0; }
 211
 212                 int count = 1;
 213
 214                 //if the users entered no valid calculators don't execute command
 215                 if (cDisplays.size() == 0) { return 0; }
 216
 217                 read = new ReadOTUFile(globaldata->inputFileName);
 218                 read->read(&*globaldata);
 219
 220                 order = globaldata->gorder;
 221                 lastOrder = order;
 222                 input = globaldata->ginput;
 223
 224                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 225                 set<string> processedLabels;
 226                 set<string> userLabels = labels;
 227                 set<int> userLines = lines;
 228
 229                 while((order != NULL) && ((allLines == 1) || (userLabels.size() != 0) || (userLines.size() != 0))) {
 230
 231                         if(allLines == 1 || lines.count(count) == 1 || labels.count(order->getLabel()) == 1){
 232
 233                                 cCurve = new Collect(order, cDisplays);
 234                                 cCurve->getCurve(freq);
 235                                 delete cCurve;
 236
 237                                 cout << order->getLabel() << '\t' << count << endl;
 238                                 processedLabels.insert(order->getLabel());
 239                                 userLabels.erase(order->getLabel());
 240                                 userLines.erase(count);
 241
 242                         //you have a label the user want that is smaller than this line and the last line has not already been processed
 243                         }
 244
 245                         if ((anyLabelsToProcess(order->getLabel(), userLabels, "") == true) && (processedLabels.count(lastOrder->getLabel()) != 1)) {
 246                                 cCurve = new Collect(lastOrder, cDisplays);
 247                                 cCurve->getCurve(freq);
 248                                 delete cCurve;
 249
 250                                 cout << lastOrder->getLabel() << '\t' << count << endl;
 251                                 processedLabels.insert(lastOrder->getLabel());
 252                                 userLabels.erase(lastOrder->getLabel());
 253                         }
 254
 255                         if (count != 1) { delete lastOrder; }
 256                         lastOrder = order;
 257                         order = (input->getOrderVector());
 258                         count++;
 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                         cout << "Your file does not include the label "<< *it;
 266                         if (processedLabels.count(lastOrder->getLabel()) != 1) {
 267                                 cout << ". I will use " << lastOrder->getLabel() << "." << endl;
 268                                 needToRun = true;
 269                         }else {
 270                                 cout << ". Please refer to " << lastOrder->getLabel() << "." << endl;
 271                         }
 272                 }
 273
 274                 //run last line if you need to
 275                 if (needToRun == true)  {
 276                         cCurve = new Collect(lastOrder, cDisplays);
 277                         cCurve->getCurve(freq);
 278                         delete cCurve;
 279
 280                         cout << lastOrder->getLabel() << '\t' << count << endl;
 281                 }
 282
 283                 delete lastOrder;
 284                 for(int i=0;i<cDisplays.size();i++){    delete cDisplays[i];    }
 285                 return 0;
 286         }
 287         catch(exception& e) {
 288                 cout << "Standard Error: " << e.what() << " has occurred in the CollectCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 289                 exit(1);
 290         }
 291         catch(...) {
 292                 cout << "An unknown error has occurred in the CollectCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 293                 exit(1);
 294         }
 295 }
 296
 297 //**********************************************************************************************************************