summarycommand.cpp

   1 /*
   2  *  summarycommand.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 "summarycommand.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 "logsd.h"
  22 #include "qstat.h"
  23 #include "bergerparker.h"
  24 #include "bstick.h"
  25 #include "goodscoverage.h"
  26 #include "coverage.h"
  27 #include "efron.h"
  28 #include "boneh.h"
  29 #include "solow.h"
  30 #include "shen.h"
  31
  32 //**********************************************************************************************************************
  33
  34 SummaryCommand::SummaryCommand(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[] =  {"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 summary.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, "abund", false);            if (temp == "not found") { temp = "10"; }
  98                         convert(temp, abund);
  99
 100                         temp = validParameter.validFile(parameters, "size", false);                     if (temp == "not found") { temp = "0"; }
 101                         convert(temp, size);
 102
 103                         if (abort == false) {
 104
 105                                 validCalculator = new ValidCalculators();
 106                                 int i;
 107
 108                                 for (i=0; i<Estimators.size(); i++) {
 109                                         if (validCalculator->isValidCalculator("summary", Estimators[i]) == true) {
 110                                                 if(Estimators[i] == "sobs"){
 111                                                         sumCalculators.push_back(new Sobs());
 112                                                 }else if(Estimators[i] == "chao"){
 113                                                         sumCalculators.push_back(new Chao1());
 114                                                 }else if(Estimators[i] == "coverage"){
 115                                                         sumCalculators.push_back(new Coverage());
 116                                                 }else if(Estimators[i] == "geometric"){
 117                                                         sumCalculators.push_back(new Geom());
 118                                                 }else if(Estimators[i] == "logseries"){
 119                                                         sumCalculators.push_back(new LogSD());
 120                                                 }else if(Estimators[i] == "qstat"){
 121                                                         sumCalculators.push_back(new QStat());
 122                                                 }else if(Estimators[i] == "bergerparker"){
 123                                                         sumCalculators.push_back(new BergerParker());
 124                                                 }else if(Estimators[i] == "bstick"){
 125                                                         sumCalculators.push_back(new BStick());
 126                                                 }else if(Estimators[i] == "ace"){
 127                                                         if(abund < 5)
 128                                                                 abund = 10;
 129                                                         sumCalculators.push_back(new Ace(abund));
 130                                                 }else if(Estimators[i] == "jack"){
 131                                                         sumCalculators.push_back(new Jackknife());
 132                                                 }else if(Estimators[i] == "shannon"){
 133                                                         sumCalculators.push_back(new Shannon());
 134                                                 }else if(Estimators[i] == "npshannon"){
 135                                                         sumCalculators.push_back(new NPShannon());
 136                                                 }else if(Estimators[i] == "simpson"){
 137                                                         sumCalculators.push_back(new Simpson());
 138                                                 }else if(Estimators[i] == "bootstrap"){
 139                                                         sumCalculators.push_back(new Bootstrap());
 140                                                 }else if (Estimators[i] == "nseqs") {
 141                                                         sumCalculators.push_back(new NSeqs());
 142                                                 }else if (Estimators[i] == "goodscoverage") {
 143                                                         sumCalculators.push_back(new GoodsCoverage());
 144                                                 }else if (Estimators[i] == "efron") {
 145                                                         sumCalculators.push_back(new Efron(size));
 146                                                 }else if (Estimators[i] == "boneh") {
 147                                                         sumCalculators.push_back(new Boneh(size));
 148                                                 }else if (Estimators[i] == "solow") {
 149                                                         sumCalculators.push_back(new Solow(size));
 150                                                 }else if (Estimators[i] == "shen") {
 151                                                         sumCalculators.push_back(new Shen(size, abund));
 152                                                 }
 153                                         }
 154                                 }
 155                         }
 156                 }
 157
 158
 159         }
 160         catch(exception& e) {
 161                 cout << "Standard Error: " << e.what() << " has occurred in the SummaryCommand class Function SummaryCommand. 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 SummaryCommand class function SummaryCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 166                 exit(1);
 167         }
 168 }
 169 //**********************************************************************************************************************
 170
 171 void SummaryCommand::help(){
 172         try {
 173                 cout << "The summary.single command can only be executed after a successful read.otu WTIH ONE EXECEPTION." << "\n";
 174                 cout << "The summary.single command can be executed after a successful cluster command.  It will use the .list file from the output of the cluster." << "\n";
 175                 cout << "The summary.single command parameters are label, line, calc, abund.  No parameters are required, but you may not use " << "\n";
 176                 cout << "both the line and label parameters at the same time. The summary.single command should be in the following format: " << "\n";
 177                 cout << "summary.single(label=yourLabel, line=yourLines, calc=yourEstimators)." << "\n";
 178                 cout << "Example summary.single(label=unique-.01-.03, line=0,5,10, calc=sobs-chao-ace-jack-bootstrap-shannon-npshannon-simpson)." << "\n";
 179                 validCalculator->printCalc("summary", cout);
 180                 cout << "The default value calc is sobs-chao-ace-jack-shannon-npshannon-simpson" << "\n";
 181                 cout << "The label and line parameters are used to analyze specific lines in your input." << "\n";
 182                 cout << "Note: No spaces between parameter labels (i.e. line), '=' and parameters (i.e.yourLines)." << "\n" << "\n";
 183         }
 184         catch(exception& e) {
 185                 cout << "Standard Error: " << e.what() << " has occurred in the SummaryCommand class Function help. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 186                 exit(1);
 187         }
 188         catch(...) {
 189                 cout << "An unknown error has occurred in the SummaryCommand class function help. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 190                 exit(1);
 191         }
 192 }
 193
 194 //**********************************************************************************************************************
 195
 196 SummaryCommand::~SummaryCommand(){
 197         delete sabund;
 198         delete input;
 199         delete read;
 200         delete validCalculator;
 201 }
 202
 203 //**********************************************************************************************************************
 204
 205 int SummaryCommand::execute(){
 206         try {
 207
 208                 if (abort == true) { return 0; }
 209
 210                 int count = 1;
 211
 212                 //if the users entered no valid calculators don't execute command
 213                 if (sumCalculators.size() == 0) { return 0; }
 214
 215                 outputFileName = ((getRootName(globaldata->inputFileName)) + "summary");
 216                 openOutputFile(outputFileName, outputFileHandle);
 217                 outputFileHandle << "label";
 218
 219                 read = new ReadOTUFile(globaldata->inputFileName);
 220                 read->read(&*globaldata);
 221
 222                 sabund = globaldata->sabund;
 223                 SAbundVector* lastSAbund = sabund;
 224                 input = globaldata->ginput;
 225
 226                 for(int i=0;i<sumCalculators.size();i++){
 227                         if(sumCalculators[i]->getCols() == 1){
 228                                 outputFileHandle << '\t' << sumCalculators[i]->getName();
 229                         }
 230                         else{
 231                                 outputFileHandle << '\t' << sumCalculators[i]->getName() << "\t" << sumCalculators[i]->getName() << "_lci\t" << sumCalculators[i]->getName() << "_hci";
 232                         }
 233                 }
 234                 outputFileHandle << endl;
 235
 236                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 237                 set<string> processedLabels;
 238                 set<string> userLabels = labels;
 239                 set<int> userLines = lines;
 240
 241                 while((sabund != NULL) && ((allLines == 1) || (userLabels.size() != 0) || (userLines.size() != 0))) {
 242
 243                         if(allLines == 1 || lines.count(count) == 1 || labels.count(sabund->getLabel()) == 1){
 244
 245                                 cout << sabund->getLabel() << '\t' << count << endl;
 246                                 processedLabels.insert(sabund->getLabel());
 247                                 userLabels.erase(sabund->getLabel());
 248                                 userLines.erase(count);
 249
 250
 251                                 outputFileHandle << sabund->getLabel();
 252                                 for(int i=0;i<sumCalculators.size();i++){
 253                                         vector<double> data = sumCalculators[i]->getValues(sabund);
 254                                         outputFileHandle << '\t';
 255                                         sumCalculators[i]->print(outputFileHandle);
 256                                 }
 257                                 outputFileHandle << endl;
 258                         }
 259
 260                         if ((anyLabelsToProcess(sabund->getLabel(), userLabels, "") == true) && (processedLabels.count(lastSAbund->getLabel()) != 1)) {
 261
 262                                 cout << lastSAbund->getLabel() << '\t' << count << endl;
 263                                 processedLabels.insert(lastSAbund->getLabel());
 264                                 userLabels.erase(lastSAbund->getLabel());
 265
 266                                 outputFileHandle << lastSAbund->getLabel();
 267                                 for(int i=0;i<sumCalculators.size();i++){
 268                                         vector<double> data = sumCalculators[i]->getValues(lastSAbund);
 269                                         outputFileHandle << '\t';
 270                                         sumCalculators[i]->print(outputFileHandle);
 271                                 }
 272                                 outputFileHandle << endl;
 273                         }
 274
 275                         if (count != 1) { delete lastSAbund; }
 276                         lastSAbund = sabund;
 277
 278                         sabund = input->getSAbundVector();
 279                         count++;
 280                 }
 281
 282                 //output error messages about any remaining user labels
 283                 set<string>::iterator it;
 284                 bool needToRun = false;
 285                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 286                         cout << "Your file does not include the label "<< *it;
 287                         if (processedLabels.count(lastSAbund->getLabel()) != 1) {
 288                                 cout << ". I will use " << lastSAbund->getLabel() << "." << endl;
 289                                 needToRun = true;
 290                         }else {
 291                                 cout << ". Please refer to " << lastSAbund->getLabel() << "." << endl;
 292                         }
 293                 }
 294
 295                 //run last line if you need to
 296                 if (needToRun == true)  {
 297                         cout << lastSAbund->getLabel() << '\t' << count << endl;
 298                         outputFileHandle << lastSAbund->getLabel();
 299                         for(int i=0;i<sumCalculators.size();i++){
 300                                 vector<double> data = sumCalculators[i]->getValues(lastSAbund);
 301                                 outputFileHandle << '\t';
 302                                 sumCalculators[i]->print(outputFileHandle);
 303                         }
 304                         outputFileHandle << endl;
 305                 }
 306
 307                 delete lastSAbund;
 308                 return 0;
 309         }
 310         catch(exception& e) {
 311                 cout << "Standard Error: " << e.what() << " has occurred in the SummaryCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 312                 exit(1);
 313         }
 314         catch(...) {
 315                 cout << "An unknown error has occurred in the SummaryCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 316                 exit(1);
 317         }
 318 }
 319
 320 //**********************************************************************************************************************