collectsharedcommand.cpp

   1 /*
   2  *  collectsharedcommand.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 "collectsharedcommand.h"
  11 #include "sharedsobscollectsummary.h"
  12 #include "sharedchao1.h"
  13 #include "sharedace.h"
  14 #include "sharedjabund.h"
  15 #include "sharedsorabund.h"
  16 #include "sharedjclass.h"
  17 #include "sharedsorclass.h"
  18 #include "sharedjest.h"
  19 #include "sharedsorest.h"
  20 #include "sharedthetayc.h"
  21 #include "sharedthetan.h"
  22 #include "sharedkstest.h"
  23 #include "whittaker.h"
  24 #include "sharednseqs.h"
  25 #include "sharedochiai.h"
  26 #include "sharedanderbergs.h"
  27 #include "sharedkulczynski.h"
  28 #include "sharedkulczynskicody.h"
  29 #include "sharedlennon.h"
  30 #include "sharedmorisitahorn.h"
  31 #include "sharedbraycurtis.h"
  32 #include "sharedjackknife.h"
  33 #include "whittaker.h"
  34
  35
  36
  37 //**********************************************************************************************************************
  38
  39 CollectSharedCommand::CollectSharedCommand(string option){
  40         try {
  41                 globaldata = GlobalData::getInstance();
  42                 abort = false;
  43                 allLines = 1;
  44                 lines.clear();
  45                 labels.clear();
  46                 Estimators.clear();
  47
  48                 //allow user to run help
  49                 if(option == "help") { validCalculator = new ValidCalculators(); help(); abort = true; }
  50
  51                 else {
  52                         //valid paramters for this command
  53                         string Array[] =  {"freq","line","label","calc","groups"};
  54                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  55
  56                         parser = new OptionParser();
  57                         parser->parse(option, parameters);  delete parser;
  58
  59                         ValidParameters* validParameter = new ValidParameters();
  60
  61                         //check to make sure all parameters are valid for command
  62                         for (it = parameters.begin(); it != parameters.end(); it++) {
  63                                 if (validParameter->isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  64                         }
  65
  66                         //make sure the user has already run the read.otu command
  67                         if (globaldata->getSharedFile() == "") {
  68                                 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; }
  69                                 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; }
  70                         }
  71
  72
  73                         //check for optional parameter and set defaults
  74                         // ...at some point should added some additional type checking...
  75                         line = validParameter->validFile(parameters, "line", false);
  76                         if (line == "not found") { line = "";  }
  77                         else {
  78                                 if(line != "all") {  splitAtDash(line, lines);  allLines = 0;  }
  79                                 else { allLines = 1;  }
  80                         }
  81
  82                         label = validParameter->validFile(parameters, "label", false);
  83                         if (label == "not found") { label = ""; }
  84                         else {
  85                                 if(label != "all") {  splitAtDash(label, labels);  allLines = 0;  }
  86                                 else { allLines = 1;  }
  87                         }
  88
  89                         //make sure user did not use both the line and label parameters
  90                         if ((line != "") && (label != "")) { cout << "You cannot use both the line and label parameters at the same time. " << endl; abort = true; }
  91                         //if the user has not specified any line or labels use the ones from read.otu
  92                         else if((line == "") && (label == "")) {
  93                                 allLines = globaldata->allLines;
  94                                 labels = globaldata->labels;
  95                                 lines = globaldata->lines;
  96                         }
  97
  98                         calc = validParameter->validFile(parameters, "calc", false);
  99                         if (calc == "not found") { calc = "sharedsobs-sharedchao-sharedace-jabund-sorabund-jclass-sorclass-jest-sorest-thetayc-thetan";  }
 100                         else {
 101                                  if (calc == "default")  {  calc = "sharedsobs-sharedchao-sharedace-jabund-sorabund-jclass-sorclass-jest-sorest-thetayc-thetan";  }
 102                         }
 103                         splitAtDash(calc, Estimators);
 104
 105                         groups = validParameter->validFile(parameters, "groups", false);
 106                         if (groups == "not found") { groups = ""; }
 107                         else {
 108                                 splitAtDash(groups, Groups);
 109                                 globaldata->Groups = Groups;
 110                         }
 111
 112                         string temp;
 113                         temp = validParameter->validFile(parameters, "freq", false);                    if (temp == "not found") { temp = "100"; }
 114                         convert(temp, freq);
 115
 116                         delete validParameter;
 117
 118                         if (abort == false) {
 119
 120                                 string fileNameRoot = getRootName(globaldata->inputFileName);
 121                                 format = globaldata->getFormat();
 122                                 int i;
 123
 124                                 validCalculator = new ValidCalculators();
 125                                 util = new SharedUtil();
 126
 127                                 for (i=0; i<Estimators.size(); i++) {
 128                                         if (validCalculator->isValidCalculator("shared", Estimators[i]) == true) {
 129                                                 if (Estimators[i] == "sharedchao") {
 130                                                         cDisplays.push_back(new CollectDisplay(new SharedChao1(), new SharedOneColumnFile(fileNameRoot+"shared.chao")));
 131                                                 }else if (Estimators[i] == "sharedsobs") {
 132                                                         cDisplays.push_back(new CollectDisplay(new SharedSobsCS(), new SharedOneColumnFile(fileNameRoot+"shared.sobs")));
 133                                                 }else if (Estimators[i] == "sharedace") {
 134                                                         cDisplays.push_back(new CollectDisplay(new SharedAce(), new SharedOneColumnFile(fileNameRoot+"shared.ace")));
 135                                                 }else if (Estimators[i] == "jabund") {
 136                                                         cDisplays.push_back(new CollectDisplay(new JAbund(), new SharedOneColumnFile(fileNameRoot+"jabund")));
 137                                                 }else if (Estimators[i] == "sorabund") {
 138                                                         cDisplays.push_back(new CollectDisplay(new SorAbund(), new SharedOneColumnFile(fileNameRoot+"sorabund")));
 139                                                 }else if (Estimators[i] == "jclass") {
 140                                                         cDisplays.push_back(new CollectDisplay(new Jclass(), new SharedOneColumnFile(fileNameRoot+"jclass")));
 141                                                 }else if (Estimators[i] == "sorclass") {
 142                                                         cDisplays.push_back(new CollectDisplay(new SorClass(), new SharedOneColumnFile(fileNameRoot+"sorclass")));
 143                                                 }else if (Estimators[i] == "jest") {
 144                                                         cDisplays.push_back(new CollectDisplay(new Jest(), new SharedOneColumnFile(fileNameRoot+"jest")));
 145                                                 }else if (Estimators[i] == "sorest") {
 146                                                         cDisplays.push_back(new CollectDisplay(new SorEst(), new SharedOneColumnFile(fileNameRoot+"sorest")));
 147                                                 }else if (Estimators[i] == "thetayc") {
 148                                                         cDisplays.push_back(new CollectDisplay(new ThetaYC(), new SharedOneColumnFile(fileNameRoot+"thetayc")));
 149                                                 }else if (Estimators[i] == "thetan") {
 150                                                         cDisplays.push_back(new CollectDisplay(new ThetaN(), new SharedOneColumnFile(fileNameRoot+"thetan")));
 151                                                 }else if (Estimators[i] == "kstest") {
 152                                                         cDisplays.push_back(new CollectDisplay(new KSTest(), new SharedOneColumnFile(fileNameRoot+"kstest")));
 153                                                 }else if (Estimators[i] == "whittaker") {
 154                                                         cDisplays.push_back(new CollectDisplay(new Whittaker(), new SharedOneColumnFile(fileNameRoot+"whittaker")));
 155                                                 }else if (Estimators[i] == "sharednseqs") {
 156                                                         cDisplays.push_back(new CollectDisplay(new SharedNSeqs(), new SharedOneColumnFile(fileNameRoot+"shared.nseqs")));
 157                                                 }else if (Estimators[i] == "ochiai") {
 158                                                         cDisplays.push_back(new CollectDisplay(new Ochiai(), new SharedOneColumnFile(fileNameRoot+"ochiai")));
 159                                                 }else if (Estimators[i] == "anderberg") {
 160                                                         cDisplays.push_back(new CollectDisplay(new Anderberg(), new SharedOneColumnFile(fileNameRoot+"anderberg")));
 161                                                 }else if (Estimators[i] == "skulczynski") {
 162                                                         cDisplays.push_back(new CollectDisplay(new Kulczynski(), new SharedOneColumnFile(fileNameRoot+"kulczynski")));
 163                                                 }else if (Estimators[i] == "kulczynskicody") {
 164                                                         cDisplays.push_back(new CollectDisplay(new KulczynskiCody(), new SharedOneColumnFile(fileNameRoot+"kulczynskicody")));
 165                                                 }else if (Estimators[i] == "lennon") {
 166                                                         cDisplays.push_back(new CollectDisplay(new Lennon(), new SharedOneColumnFile(fileNameRoot+"lennon")));
 167                                                 }else if (Estimators[i] == "morisitahorn") {
 168                                                         cDisplays.push_back(new CollectDisplay(new MorHorn(), new SharedOneColumnFile(fileNameRoot+"morisitahorn")));
 169                                                 }else if (Estimators[i] == "braycurtis") {
 170                                                         cDisplays.push_back(new CollectDisplay(new BrayCurtis(), new SharedOneColumnFile(fileNameRoot+"braycurtis")));
 171                                                 }
 172                                         }
 173                                 }
 174                         }
 175                 }
 176
 177         }
 178         catch(exception& e) {
 179                 cout << "Standard Error: " << e.what() << " has occurred in the CollectSharedCommand class Function CollectSharedCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 180                 exit(1);
 181         }
 182         catch(...) {
 183                 cout << "An unknown error has occurred in the CollectSharedCommand class function CollectSharedCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 184                 exit(1);
 185         }
 186
 187 }
 188 //**********************************************************************************************************************
 189
 190 void CollectSharedCommand::help(){
 191         try {
 192                 cout << "The collect.shared command can only be executed after a successful read.otu command." << "\n";
 193                 cout << "The collect.shared command parameters are label, line, freq, calc and groups.  No parameters are required, but you may not use " << "\n";
 194                 cout << "both the line and label parameters at the same time. The collect.shared command should be in the following format: " << "\n";
 195                 cout << "collect.shared(label=yourLabel, line=yourLines, freq=yourFreq, calc=yourEstimators, groups=yourGroups)." << "\n";
 196                 cout << "Example collect.shared(label=unique-.01-.03, line=0-5-10, freq=10, groups=B-C, calc=sharedchao-sharedace-jabund-sorensonabund-jclass-sorclass-jest-sorest-thetayc-thetan)." << "\n";
 197                 cout << "The default values for freq is 100 and calc are sharedsobs-sharedchao-sharedace-jabund-sorensonabund-jclass-sorclass-jest-sorest-thetayc-thetan." << "\n";
 198                 cout << "The default value for groups is all the groups in your groupfile." << "\n";
 199                 validCalculator->printCalc("shared", cout);
 200                 cout << "The label and line parameters are used to analyze specific lines in your input." << "\n";
 201                 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";
 202                 cout << "Note: No spaces between parameter labels (i.e. list), '=' and parameters (i.e.yourListfile)." << "\n" << "\n";
 203
 204         }
 205         catch(exception& e) {
 206                 cout << "Standard Error: " << e.what() << " has occurred in the CollectSharedCommand class Function help. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 207                 exit(1);
 208         }
 209         catch(...) {
 210                 cout << "An unknown error has occurred in the CollectSharedCommand class function help. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 211                 exit(1);
 212         }
 213 }
 214
 215 //**********************************************************************************************************************
 216
 217 CollectSharedCommand::~CollectSharedCommand(){
 218         delete order;
 219         delete input;
 220         delete cCurve;
 221         delete read;
 222         delete util;
 223         delete validCalculator;
 224 }
 225
 226 //**********************************************************************************************************************
 227
 228 int CollectSharedCommand::execute(){
 229         try {
 230
 231                 if (abort == true) {    return 0;       }
 232
 233                 int count = 1;
 234
 235                 //if the users entered no valid calculators don't execute command
 236                 if (cDisplays.size() == 0) { return 0; }
 237
 238                 read = new ReadOTUFile(globaldata->inputFileName);
 239                 read->read(&*globaldata);
 240
 241                 input = globaldata->ginput;
 242                 order = input->getSharedOrderVector();
 243                 SharedOrderVector* lastOrder = order;
 244
 245                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 246                 set<string> processedLabels;
 247                 set<string> userLabels = globaldata->labels;
 248                 set<int> userLines = globaldata->lines;
 249
 250                 //set users groups
 251                 util->setGroups(globaldata->Groups, globaldata->gGroupmap->namesOfGroups, "collect");
 252                 util->updateGroupIndex(globaldata->Groups, globaldata->gGroupmap->groupIndex);
 253
 254                 while((order != NULL) && ((globaldata->allLines == 1) || (userLabels.size() != 0) || (userLines.size() != 0))) {
 255
 256                         if(globaldata->allLines == 1 || globaldata->lines.count(count) == 1 || globaldata->labels.count(order->getLabel()) == 1){
 257
 258                                 //create collectors curve
 259                                 cCurve = new Collect(order, cDisplays);
 260                                 cCurve->getSharedCurve(freq);
 261                                 delete cCurve;
 262
 263                                 cout << order->getLabel() << '\t' << count << endl;
 264                                 processedLabels.insert(order->getLabel());
 265                                 userLabels.erase(order->getLabel());
 266                                 userLines.erase(count);
 267
 268                         //you have a label the user want that is smaller than this line and the last line has not already been processed
 269                         }
 270
 271                         if ((anyLabelsToProcess(order->getLabel(), userLabels, "") == true) && (processedLabels.count(lastOrder->getLabel()) != 1)) {
 272                                 //create collectors curve
 273                                 cCurve = new Collect(lastOrder, cDisplays);
 274                                 cCurve->getSharedCurve(freq);
 275                                 delete cCurve;
 276
 277                                 cout << lastOrder->getLabel() << '\t' << count << endl;
 278                                 processedLabels.insert(lastOrder->getLabel());
 279                                 userLabels.erase(lastOrder->getLabel());
 280                         }
 281
 282                         if (count != 1) { delete lastOrder; }
 283                         lastOrder = order;
 284
 285                         //get next line to process
 286                         order = input->getSharedOrderVector();
 287                         count++;
 288                 }
 289
 290                 //output error messages about any remaining user labels
 291                 set<string>::iterator it;
 292                 bool needToRun = false;
 293                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 294                         cout << "Your file does not include the label "<< *it;
 295                         if (processedLabels.count(lastOrder->getLabel()) != 1) {
 296                                 cout << ". I will use " << lastOrder->getLabel() << "." << endl;
 297                                 needToRun = true;
 298                         }else {
 299                                 cout << ". Please refer to " << lastOrder->getLabel() << "." << endl;
 300                         }
 301                 }
 302
 303                 //run last line if you need to
 304                 if (needToRun == true)  {
 305                         cCurve = new Collect(lastOrder, cDisplays);
 306                         cCurve->getCurve(freq);
 307                         delete cCurve;
 308
 309                         cout << lastOrder->getLabel() << '\t' << count << endl;
 310                 }
 311
 312                 delete lastOrder;
 313                 for(int i=0;i<cDisplays.size();i++){    delete cDisplays[i];    }
 314
 315                 //reset groups parameter
 316                 globaldata->Groups.clear();
 317
 318                 return 0;
 319         }
 320         catch(exception& e) {
 321                 cout << "Standard Error: " << e.what() << " has occurred in the CollectSharedCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 322                 exit(1);
 323         }
 324         catch(...) {
 325                 cout << "An unknown error has occurred in the CollectSharedCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 326                 exit(1);
 327         }
 328 }
 329
 330 /***********************************************************/
 331