summarysharedcommand.cpp

   1 /*
   2  *  summarysharedcommand.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 "summarysharedcommand.h"
  11 #include "sharedsobscollectsummary.h"
  12 #include "sharedchao1.h"
  13 #include "sharedace.h"
  14 #include "sharednseqs.h"
  15 #include "sharedjabund.h"
  16 #include "sharedsorabund.h"
  17 #include "sharedjclass.h"
  18 #include "sharedsorclass.h"
  19 #include "sharedjest.h"
  20 #include "sharedsorest.h"
  21 #include "sharedthetayc.h"
  22 #include "sharedthetan.h"
  23 #include "sharedkstest.h"
  24 #include "whittaker.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 SummarySharedCommand::SummarySharedCommand(string option){
  39         try {
  40                 globaldata = GlobalData::getInstance();
  41                 abort = false;
  42                 allLines = 1;
  43                 labels.clear();
  44                 Estimators.clear();
  45
  46                 //allow user to run help
  47                 if(option == "help") { validCalculator = new ValidCalculators(); help(); abort = true; }
  48
  49                 else {
  50                         //valid paramters for this command
  51                         string Array[] =  {"label","calc","groups"};
  52                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  53
  54                         OptionParser parser(option);
  55                         map<string, string> parameters = parser.getParameters();
  56
  57                         ValidParameters validParameter;
  58
  59                         //check to make sure all parameters are valid for command
  60                         for (map<string, string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
  61                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  62                         }
  63
  64                         //make sure the user has already run the read.otu command
  65                         if (globaldata->getSharedFile() == "") {
  66                                  mothurOut("You must read a list and a group, or a shared before you can use the summary.shared command."); mothurOutEndLine(); abort = true;
  67                         }
  68
  69                         //check for optional parameter and set defaults
  70                         // ...at some point should added some additional type checking...
  71                         label = validParameter.validFile(parameters, "label", false);
  72                         if (label == "not found") { label = ""; }
  73                         else {
  74                                 if(label != "all") {  splitAtDash(label, labels);  allLines = 0;  }
  75                                 else { allLines = 1;  }
  76                         }
  77
  78                         //if the user has not specified any labels use the ones from read.otu
  79                         if(label == "") {
  80                                 allLines = globaldata->allLines;
  81                                 labels = globaldata->labels;
  82                         }
  83
  84                         calc = validParameter.validFile(parameters, "calc", false);
  85                         if (calc == "not found") { calc = "sharedsobs-sharedchao-sharedace-jabund-sorabund-jclass-sorclass-jest-sorest-thetayc-thetan";  }
  86                         else {
  87                                  if (calc == "default")  {  calc = "sharedsobs-sharedchao-sharedace-jabund-sorabund-jclass-sorclass-jest-sorest-thetayc-thetan";  }
  88                         }
  89                         splitAtDash(calc, Estimators);
  90
  91                         groups = validParameter.validFile(parameters, "groups", false);
  92                         if (groups == "not found") { groups = ""; }
  93                         else {
  94                                 splitAtDash(groups, Groups);
  95                                 globaldata->Groups = Groups;
  96                         }
  97
  98                         if (abort == false) {
  99
 100                                 validCalculator = new ValidCalculators();
 101                                 int i;
 102
 103                                 for (i=0; i<Estimators.size(); i++) {
 104                                         if (validCalculator->isValidCalculator("sharedsummary", Estimators[i]) == true) {
 105                                                 if (Estimators[i] == "sharedsobs") {
 106                                                         sumCalculators.push_back(new SharedSobsCS());
 107                                                 }else if (Estimators[i] == "sharedchao") {
 108                                                         sumCalculators.push_back(new SharedChao1());
 109                                                 }else if (Estimators[i] == "sharedace") {
 110                                                         sumCalculators.push_back(new SharedAce());
 111                                                 }else if (Estimators[i] == "jabund") {
 112                                                         sumCalculators.push_back(new JAbund());
 113                                                 }else if (Estimators[i] == "sorabund") {
 114                                                         sumCalculators.push_back(new SorAbund());
 115                                                 }else if (Estimators[i] == "jclass") {
 116                                                         sumCalculators.push_back(new Jclass());
 117                                                 }else if (Estimators[i] == "sorclass") {
 118                                                         sumCalculators.push_back(new SorClass());
 119                                                 }else if (Estimators[i] == "jest") {
 120                                                         sumCalculators.push_back(new Jest());
 121                                                 }else if (Estimators[i] == "sorest") {
 122                                                         sumCalculators.push_back(new SorEst());
 123                                                 }else if (Estimators[i] == "thetayc") {
 124                                                         sumCalculators.push_back(new ThetaYC());
 125                                                 }else if (Estimators[i] == "thetan") {
 126                                                         sumCalculators.push_back(new ThetaN());
 127                                                 }else if (Estimators[i] == "kstest") {
 128                                                         sumCalculators.push_back(new KSTest());
 129                                                 }else if (Estimators[i] == "sharednseqs") {
 130                                                         sumCalculators.push_back(new SharedNSeqs());
 131                                                 }else if (Estimators[i] == "ochiai") {
 132                                                         sumCalculators.push_back(new Ochiai());
 133                                                 }else if (Estimators[i] == "anderberg") {
 134                                                         sumCalculators.push_back(new Anderberg());
 135                                                 }else if (Estimators[i] == "kulczynski") {
 136                                                         sumCalculators.push_back(new Kulczynski());
 137                                                 }else if (Estimators[i] == "kulczynskicody") {
 138                                                         sumCalculators.push_back(new KulczynskiCody());
 139                                                 }else if (Estimators[i] == "lennon") {
 140                                                         sumCalculators.push_back(new Lennon());
 141                                                 }else if (Estimators[i] == "morisitahorn") {
 142                                                         sumCalculators.push_back(new MorHorn());
 143                                                 }else if (Estimators[i] == "braycurtis") {
 144                                                         sumCalculators.push_back(new BrayCurtis());
 145                                                 }else if (Estimators[i] == "whittaker") {
 146                                                         sumCalculators.push_back(new Whittaker());
 147                                                 }
 148                                         }
 149                                 }
 150
 151                                 outputFileName = ((getRootName(globaldata->inputFileName)) + "shared.summary");
 152                                 openOutputFile(outputFileName, outputFileHandle);
 153                                 mult = false;
 154                         }
 155                 }
 156         }
 157         catch(exception& e) {
 158                 errorOut(e, "SummarySharedCommand", "SummarySharedCommand");
 159                 exit(1);
 160         }
 161 }
 162
 163 //**********************************************************************************************************************
 164
 165 void SummarySharedCommand::help(){
 166         try {
 167                 mothurOut("The summary.shared command can only be executed after a successful read.otu command.\n");
 168                 mothurOut("The summary.shared command parameters are label and calc.  No parameters are required.\n");
 169                 mothurOut("The summary.shared command should be in the following format: \n");
 170                 mothurOut("summary.shared(label=yourLabel, calc=yourEstimators, groups=yourGroups).\n");
 171                 mothurOut("Example summary.shared(label=unique-.01-.03, groups=B-C, calc=sharedchao-sharedace-jabund-sorensonabund-jclass-sorclass-jest-sorest-thetayc-thetan).\n");
 172                 validCalculator->printCalc("sharedsummary", cout);
 173                 mothurOut("The default value for calc is sharedsobs-sharedchao-sharedace-jabund-sorensonabund-jclass-sorclass-jest-sorest-thetayc-thetan\n");
 174                 mothurOut("The default value for groups is all the groups in your groupfile.\n");
 175                 mothurOut("The label parameter is used to analyze specific labels in your input.\n");
 176                 mothurOut("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");
 177                 mothurOut("Note: No spaces between parameter labels (i.e. label), '=' and parameters (i.e.yourLabel).\n\n");
 178         }
 179         catch(exception& e) {
 180                 errorOut(e, "SummarySharedCommand", "help");
 181                 exit(1);
 182         }
 183 }
 184
 185 //**********************************************************************************************************************
 186
 187 SummarySharedCommand::~SummarySharedCommand(){
 188         if (abort == false) {
 189                 delete read;
 190                 delete validCalculator;
 191         }
 192 }
 193
 194 //**********************************************************************************************************************
 195
 196 int SummarySharedCommand::execute(){
 197         try {
 198
 199                 if (abort == true) { return 0; }
 200
 201                 //if the users entered no valid calculators don't execute command
 202                 if (sumCalculators.size() == 0) { return 0; }
 203                 //check if any calcs can do multiples
 204                 else{
 205                         for (int i = 0; i < sumCalculators.size(); i++) {
 206                                 if (sumCalculators[i]->getMultiple() == true) { mult = true; }
 207                         }
 208                 }
 209
 210                 //read first line
 211                 read = new ReadOTUFile(globaldata->inputFileName);
 212                 read->read(&*globaldata);
 213
 214                 input = globaldata->ginput;
 215                 lookup = input->getSharedRAbundVectors();
 216                 string lastLabel = lookup[0]->getLabel();
 217
 218                 //output estimator names as column headers
 219                 outputFileHandle << "label" <<'\t' << "comparison" << '\t';
 220                 for(int i=0;i<sumCalculators.size();i++){
 221                         outputFileHandle << '\t' << sumCalculators[i]->getName();
 222                 }
 223                 outputFileHandle << endl;
 224
 225                 //create file and put column headers for multiple groups file
 226                 if (mult == true) {
 227                         outAllFileName = ((getRootName(globaldata->inputFileName)) + "sharedmultiple.summary");
 228                         openOutputFile(outAllFileName, outAll);
 229
 230                         outAll << "label" <<'\t' << "comparison" << '\t';
 231                         for(int i=0;i<sumCalculators.size();i++){
 232                                 if (sumCalculators[i]->getMultiple() == true) {
 233                                         outAll << '\t' << sumCalculators[i]->getName();
 234                                 }
 235                         }
 236                         outAll << endl;
 237                 }
 238
 239                 if (lookup.size() < 2) {
 240                         mothurOut("I cannot run the command without at least 2 valid groups.");
 241                         for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; }
 242
 243                         //close files and clean up
 244                         outputFileHandle.close();  remove(outputFileName.c_str());
 245                         if (mult == true) {  outAll.close();  remove(outAllFileName.c_str());  }
 246                         return 0;
 247                 //if you only have 2 groups you don't need a .sharedmultiple file
 248                 }else if ((lookup.size() == 2) && (mult == true)) {
 249                         mult = false;
 250                         outAll.close();
 251                         remove(outAllFileName.c_str());
 252                 }
 253
 254                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 255                 set<string> processedLabels;
 256                 set<string> userLabels = labels;
 257
 258                 //as long as you are not at the end of the file or done wih the lines you want
 259                 while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 260
 261                         if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){
 262                                 mothurOut(lookup[0]->getLabel()); mothurOutEndLine();
 263                                 process(lookup);
 264
 265                                 processedLabels.insert(lookup[0]->getLabel());
 266                                 userLabels.erase(lookup[0]->getLabel());
 267                         }
 268
 269                         if ((anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 270                                         string saveLabel = lookup[0]->getLabel();
 271
 272                                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 273                                         lookup = input->getSharedRAbundVectors(lastLabel);
 274
 275                                         mothurOut(lookup[0]->getLabel()); mothurOutEndLine();
 276                                         process(lookup);
 277
 278                                         processedLabels.insert(lookup[0]->getLabel());
 279                                         userLabels.erase(lookup[0]->getLabel());
 280
 281                                         //restore real lastlabel to save below
 282                                         lookup[0]->setLabel(saveLabel);
 283                         }
 284
 285                         lastLabel = lookup[0]->getLabel();
 286
 287                         //get next line to process
 288                         //prevent memory leak
 289                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 290                         lookup = input->getSharedRAbundVectors();
 291                 }
 292
 293                 //output error messages about any remaining user labels
 294                 set<string>::iterator it;
 295                 bool needToRun = false;
 296                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 297                         mothurOut("Your file does not include the label " + *it);
 298                         if (processedLabels.count(lastLabel) != 1) {
 299                                 mothurOut(". I will use " + lastLabel + "."); mothurOutEndLine();
 300                                 needToRun = true;
 301                         }else {
 302                                 mothurOut(". Please refer to " + lastLabel + "."); mothurOutEndLine();
 303                         }
 304                 }
 305
 306                 //run last label if you need to
 307                 if (needToRun == true)  {
 308                                 for (int i = 0; i < lookup.size(); i++) {  if (lookup[i] != NULL) {     delete lookup[i];       } }
 309                                 lookup = input->getSharedRAbundVectors(lastLabel);
 310
 311                                 mothurOut(lookup[0]->getLabel()); mothurOutEndLine();
 312                                 process(lookup);
 313                                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 314                 }
 315
 316
 317                 //reset groups parameter
 318                 globaldata->Groups.clear();
 319
 320                 //close files
 321                 outputFileHandle.close();
 322                 if (mult == true) {  outAll.close();  }
 323
 324                 for(int i=0;i<sumCalculators.size();i++){  delete sumCalculators[i]; }
 325
 326                 delete input;  globaldata->ginput = NULL;
 327
 328                 return 0;
 329         }
 330         catch(exception& e) {
 331                 errorOut(e, "SummarySharedCommand", "execute");
 332                 exit(1);
 333         }
 334 }
 335
 336 /***********************************************************/
 337 void SummarySharedCommand::process(vector<SharedRAbundVector*> thisLookup) {
 338         try {
 339                                 //loop through calculators and add to file all for all calcs that can do mutiple groups
 340                                 if (mult == true) {
 341                                         //output label
 342                                         outAll << thisLookup[0]->getLabel() << '\t';
 343
 344                                         //output groups names
 345                                         string outNames = "";
 346                                         for (int j = 0; j < thisLookup.size(); j++) {
 347                                                 outNames += thisLookup[j]->getGroup() +  "-";
 348                                         }
 349                                         outNames = outNames.substr(0, outNames.length()-1); //rip off extra '-';
 350                                         outAll << outNames << '\t';
 351
 352                                         for(int i=0;i<sumCalculators.size();i++){
 353                                                 if (sumCalculators[i]->getMultiple() == true) {
 354                                                         sumCalculators[i]->getValues(thisLookup);
 355                                                         outAll << '\t';
 356                                                         sumCalculators[i]->print(outAll);
 357                                                 }
 358                                         }
 359                                         outAll << endl;
 360                                 }
 361
 362                                 int n = 1;
 363                                 vector<SharedRAbundVector*> subset;
 364                                 for (int k = 0; k < (thisLookup.size() - 1); k++) { // pass cdd each set of groups to commpare
 365                                         for (int l = n; l < thisLookup.size(); l++) {
 366
 367                                                 outputFileHandle << thisLookup[0]->getLabel() << '\t';
 368
 369                                                 subset.clear(); //clear out old pair of sharedrabunds
 370                                                 //add new pair of sharedrabunds
 371                                                 subset.push_back(thisLookup[k]); subset.push_back(thisLookup[l]);
 372
 373                                                 //sort groups to be alphanumeric
 374                                                 if (thisLookup[k]->getGroup() > thisLookup[l]->getGroup()) {
 375                                                         outputFileHandle << (thisLookup[l]->getGroup() +'\t' + thisLookup[k]->getGroup()) << '\t'; //print out groups
 376                                                 }else{
 377                                                         outputFileHandle << (thisLookup[k]->getGroup() +'\t' + thisLookup[l]->getGroup()) << '\t'; //print out groups
 378                                                 }
 379
 380                                                 for(int i=0;i<sumCalculators.size();i++) {
 381
 382                                                         sumCalculators[i]->getValues(subset); //saves the calculator outputs
 383                                                         outputFileHandle << '\t';
 384                                                         sumCalculators[i]->print(outputFileHandle);
 385                                                 }
 386                                                 outputFileHandle << endl;
 387                                         }
 388                                         n++;
 389                                 }
 390
 391         }
 392         catch(exception& e) {
 393                 errorOut(e, "SummarySharedCommand", "process");
 394                 exit(1);
 395         }
 396 }
 397
 398 /***********************************************************/