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