rarefactsharedcommand.cpp

   1 /*
   2  *  rarefactsharedcommand.cpp
   3  *  Dotur
   4  *
   5  *  Created by Sarah Westcott on 1/6/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10 #include "rarefactsharedcommand.h"
  11 #include "sharedsobs.h"
  12 #include "sharednseqs.h"
  13
  14 //**********************************************************************************************************************
  15
  16 RareFactSharedCommand::RareFactSharedCommand(string option){
  17         try {
  18                 globaldata = GlobalData::getInstance();
  19
  20                 abort = false;
  21                 allLines = 1;
  22                 lines.clear();
  23                 labels.clear();
  24                 Estimators.clear();
  25                 Groups.clear();
  26
  27                 //allow user to run help
  28                 if(option == "help") { validCalculator = new ValidCalculators(); help(); abort = true; }
  29
  30                 else {
  31                         //valid paramters for this command
  32                         string Array[] =  {"iters","line","label","calc","groups"};
  33                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  34
  35                         OptionParser parser(option);
  36                         map<string,string> parameters = parser.getParameters();
  37
  38                         ValidParameters validParameter;
  39
  40                         //check to make sure all parameters are valid for command
  41                         for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
  42                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  43                         }
  44
  45                         //make sure the user has already run the read.otu command
  46                         if (globaldata->getSharedFile() == "") {
  47                                 if (globaldata->getListFile() == "") { cout << "You must read a list and a group, or a shared before you can use the collect.shared command." << endl; abort = true; }
  48                                 else if (globaldata->getGroupFile() == "") { cout << "You must read a list and a group, or a shared before you can use the collect.shared command." << endl; abort = true; }
  49                         }
  50
  51
  52                         //check for optional parameter and set defaults
  53                         // ...at some point should added some additional type checking...
  54                         line = validParameter.validFile(parameters, "line", false);
  55                         if (line == "not found") { line = "";  }
  56                         else {
  57                                 if(line != "all") {  splitAtDash(line, lines);  allLines = 0;  }
  58                                 else { allLines = 1;  }
  59                         }
  60
  61                         label = validParameter.validFile(parameters, "label", false);
  62                         if (label == "not found") { label = ""; }
  63                         else {
  64                                 if(label != "all") {  splitAtDash(label, labels);  allLines = 0;  }
  65                                 else { allLines = 1;  }
  66                         }
  67
  68                         //make sure user did not use both the line and label parameters
  69                         if ((line != "") && (label != "")) { cout << "You cannot use both the line and label parameters at the same time. " << endl; abort = true; }
  70                         //if the user has not specified any line or labels use the ones from read.otu
  71                         else if((line == "") && (label == "")) {
  72                                 allLines = globaldata->allLines;
  73                                 labels = globaldata->labels;
  74                                 lines = globaldata->lines;
  75                         }
  76
  77                         calc = validParameter.validFile(parameters, "calc", false);
  78                         if (calc == "not found") { calc = "sharedobserved";  }
  79                         else {
  80                                  if (calc == "default")  {  calc = "sharedobserved";  }
  81                         }
  82                         splitAtDash(calc, Estimators);
  83
  84                         groups = validParameter.validFile(parameters, "groups", false);
  85                         if (groups == "not found") { groups = ""; }
  86                         else {
  87                                 splitAtDash(groups, Groups);
  88                         }
  89                         globaldata->Groups = Groups;
  90
  91                         string temp;
  92                         temp = validParameter.validFile(parameters, "iters", false);                    if (temp == "not found") { temp = "1000"; }
  93                         convert(temp, nIters);
  94
  95                         if (abort == false) {
  96
  97                                 string fileNameRoot = getRootName(globaldata->inputFileName);
  98 //                              format = globaldata->getFormat();
  99
 100
 101                                 validCalculator = new ValidCalculators();
 102
 103                                 for (int i=0; i<Estimators.size(); i++) {
 104                                         if (validCalculator->isValidCalculator("sharedrarefaction", Estimators[i]) == true) {
 105                                                 if (Estimators[i] == "sharedobserved") {
 106                                                         rDisplays.push_back(new RareDisplay(new SharedSobs(), new SharedThreeColumnFile(fileNameRoot+"shared.rarefaction", "")));
 107                                                 }else if (Estimators[i] == "sharednseqs") {
 108                                                         rDisplays.push_back(new RareDisplay(new SharedNSeqs(), new SharedThreeColumnFile(fileNameRoot+"shared.r_nseqs", "")));
 109                                                 }
 110                                         }
 111                                 }
 112                         }
 113
 114                 }
 115
 116         }
 117         catch(exception& e) {
 118                 cout << "Standard Error: " << e.what() << " has occurred in the RareFactSharedCommand class Function RareFactSharedCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 119                 exit(1);
 120         }
 121         catch(...) {
 122                 cout << "An unknown error has occurred in the RareFactSharedCommand class function RareFactSharedCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 123                 exit(1);
 124         }
 125
 126 }
 127
 128 //**********************************************************************************************************************
 129
 130 void RareFactSharedCommand::help(){
 131         try {
 132                 cout << "The rarefaction.shared command can only be executed after a successful read.otu command." << "\n";
 133                 cout << "The rarefaction.shared command parameters are label, line, iters, groups and calc.  No parameters are required, but you may not use " << "\n";
 134                 cout << "both the line and label parameters at the same time. The rarefaction command should be in the following format: " << "\n";
 135                 cout << "rarefaction.shared(label=yourLabel, line=yourLines, iters=yourIters, calc=yourEstimators, groups=yourGroups)." << "\n";
 136                 cout << "Example rarefaction.shared(label=unique-.01-.03, line=0-5-10, iters=10000, groups=B-C, calc=sharedobserved)." << "\n";
 137                 cout << "The default values for iters is 1000, freq is 100, and calc is sharedobserved which calculates the shared rarefaction curve for the observed richness." << "\n";
 138                 cout << "The default value for groups is all the groups in your groupfile." << "\n";
 139                 validCalculator->printCalc("sharedrarefaction", cout);
 140                 cout << "The label and line parameters are used to analyze specific lines in your input." << "\n";
 141                 cout << "The groups parameter allows you to specify which of the groups in your groupfile you would like analyzed.  You must enter at least 2 valid groups." << "\n";
 142                 cout << "Note: No spaces between parameter labels (i.e. freq), '=' and parameters (i.e.yourFreq)." << "\n" << "\n";
 143         }
 144         catch(exception& e) {
 145                 cout << "Standard Error: " << e.what() << " has occurred in the RareFactSharedCommand class Function help. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 146                 exit(1);
 147         }
 148         catch(...) {
 149                 cout << "An unknown error has occurred in the RareFactSharedCommand class function help. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 150                 exit(1);
 151         }
 152 }
 153
 154 //**********************************************************************************************************************
 155
 156 RareFactSharedCommand::~RareFactSharedCommand(){
 157         if (abort == false) {
 158                 delete input;   globaldata->ginput = NULL;
 159                 delete read;
 160                 delete validCalculator;
 161         }
 162 }
 163
 164 //**********************************************************************************************************************
 165
 166 int RareFactSharedCommand::execute(){
 167         try {
 168
 169                 if (abort == true) { return 0; }
 170
 171                 int count = 1;
 172
 173                 //if the users entered no valid calculators don't execute command
 174                 if (rDisplays.size() == 0) { return 0; }
 175
 176                 read = new ReadOTUFile(globaldata->inputFileName);
 177                 read->read(&*globaldata);
 178
 179                 input = globaldata->ginput;
 180                 lookup = input->getSharedRAbundVectors();
 181                 string lastLabel = lookup[0]->getLabel();
 182
 183                 if (lookup.size() < 2) {
 184                         cout << "I cannot run the command without at least 2 valid groups.";
 185                         for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; }
 186                         return 0;
 187                 }
 188
 189                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 190                 set<string> processedLabels;
 191                 set<string> userLabels = labels;
 192                 set<int> userLines = lines;
 193
 194                 //as long as you are not at the end of the file or done wih the lines you want
 195                 while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0) || (userLines.size() != 0))) {
 196
 197                         if(allLines == 1 || lines.count(count) == 1 || labels.count(lookup[0]->getLabel()) == 1){
 198
 199                                 rCurve = new Rarefact(lookup, rDisplays);
 200                                 rCurve->getSharedCurve(freq, nIters);
 201                                 delete rCurve;
 202
 203                                 cout << lookup[0]->getLabel() << '\t' << count << endl;
 204                                 processedLabels.insert(lookup[0]->getLabel());
 205                                 userLabels.erase(lookup[0]->getLabel());
 206                                 userLines.erase(count);
 207                         }
 208
 209                         if ((anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 210                                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 211                                         lookup = input->getSharedRAbundVectors(lastLabel);
 212
 213                                         cout << lookup[0]->getLabel() << '\t' << count << endl;
 214                                         rCurve = new Rarefact(lookup, rDisplays);
 215                                         rCurve->getSharedCurve(freq, nIters);
 216                                         delete rCurve;
 217
 218                                         processedLabels.insert(lookup[0]->getLabel());
 219                                         userLabels.erase(lookup[0]->getLabel());
 220                         }
 221
 222
 223                         lastLabel = lookup[0]->getLabel();
 224
 225                         //get next line to process
 226                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 227                         lookup = input->getSharedRAbundVectors();
 228                         count++;
 229                 }
 230
 231                 //output error messages about any remaining user labels
 232                 set<string>::iterator it;
 233                 bool needToRun = false;
 234                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 235                         cout << "Your file does not include the label "<< *it;
 236                         if (processedLabels.count(lastLabel) != 1) {
 237                                 cout << ". I will use " << lastLabel << "." << endl;
 238                                 needToRun = true;
 239                         }else {
 240                                 cout << ". Please refer to " << lastLabel << "." << endl;
 241                         }
 242                 }
 243
 244                 //run last line if you need to
 245                 if (needToRun == true)  {
 246                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 247                         lookup = input->getSharedRAbundVectors(lastLabel);
 248
 249                         cout << lookup[0]->getLabel() << '\t' << count << endl;
 250                         rCurve = new Rarefact(lookup, rDisplays);
 251                         rCurve->getSharedCurve(freq, nIters);
 252                         delete rCurve;
 253                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 254                 }
 255
 256                 for(int i=0;i<rDisplays.size();i++){    delete rDisplays[i];    }
 257
 258                 //reset groups parameter
 259                 globaldata->Groups.clear();
 260
 261                 return 0;
 262         }
 263         catch(exception& e) {
 264                 cout << "Standard Error: " << e.what() << " has occurred in the RareFactSharedCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 265                 exit(1);
 266         }
 267         catch(...) {
 268                 cout << "An unknown error has occurred in the RareFactSharedCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 269                 exit(1);
 270         }
 271 }
 272
 273
 274 //**********************************************************************************************************************