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