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                 lines.clear();
  44                 labels.clear();
  45                 Estimators.clear();
  46
  47                 //allow user to run help
  48                 if(option == "help") { validCalculator = new ValidCalculators(); help(); abort = true; }
  49
  50                 else {
  51                         //valid paramters for this command
  52                         string Array[] =  {"line","label","calc","groups"};
  53                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  54
  55                         OptionParser parser(option);
  56                         map<string, string> parameters = parser.getParameters();
  57
  58                         ValidParameters validParameter;
  59
  60                         //check to make sure all parameters are valid for command
  61                         for (map<string, string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
  62                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  63                         }
  64
  65                         //make sure the user has already run the read.otu command
  66                         if (globaldata->getSharedFile() == "") {
  67                                  cout << "You must read a list and a group, or a shared before you can use the summary.shared command." << endl; abort = true;
  68                         }
  69
  70                         //check for optional parameter and set defaults
  71                         // ...at some point should added some additional type checking...
  72                         line = validParameter.validFile(parameters, "line", false);
  73                         if (line == "not found") { line = "";  }
  74                         else {
  75                                 if(line != "all") {  splitAtDash(line, lines);  allLines = 0;  }
  76                                 else { allLines = 1;  }
  77                         }
  78
  79                         label = validParameter.validFile(parameters, "label", false);
  80                         if (label == "not found") { label = ""; }
  81                         else {
  82                                 if(label != "all") {  splitAtDash(label, labels);  allLines = 0;  }
  83                                 else { allLines = 1;  }
  84                         }
  85
  86                         //make sure user did not use both the line and label parameters
  87                         if ((line != "") && (label != "")) { cout << "You cannot use both the line and label parameters at the same time. " << endl; abort = true; }
  88                         //if the user has not specified any line or labels use the ones from read.otu
  89                         else if((line == "") && (label == "")) {
  90                                 allLines = globaldata->allLines;
  91                                 labels = globaldata->labels;
  92                                 lines = globaldata->lines;
  93                         }
  94
  95                         calc = validParameter.validFile(parameters, "calc", false);
  96                         if (calc == "not found") { calc = "sharedsobs-sharedchao-sharedace-jabund-sorabund-jclass-sorclass-jest-sorest-thetayc-thetan";  }
  97                         else {
  98                                  if (calc == "default")  {  calc = "sharedsobs-sharedchao-sharedace-jabund-sorabund-jclass-sorclass-jest-sorest-thetayc-thetan";  }
  99                         }
 100                         splitAtDash(calc, Estimators);
 101
 102                         groups = validParameter.validFile(parameters, "groups", false);
 103                         if (groups == "not found") { groups = ""; }
 104                         else {
 105                                 splitAtDash(groups, Groups);
 106                                 globaldata->Groups = Groups;
 107                         }
 108
 109                         if (abort == false) {
 110
 111                                 validCalculator = new ValidCalculators();
 112                                 int i;
 113
 114                                 for (i=0; i<Estimators.size(); i++) {
 115                                         if (validCalculator->isValidCalculator("sharedsummary", Estimators[i]) == true) {
 116                                                 if (Estimators[i] == "sharedsobs") {
 117                                                         sumCalculators.push_back(new SharedSobsCS());
 118                                                 }else if (Estimators[i] == "sharedchao") {
 119                                                         sumCalculators.push_back(new SharedChao1());
 120                                                 }else if (Estimators[i] == "sharedace") {
 121                                                         sumCalculators.push_back(new SharedAce());
 122                                                 }else if (Estimators[i] == "jabund") {
 123                                                         sumCalculators.push_back(new JAbund());
 124                                                 }else if (Estimators[i] == "sorabund") {
 125                                                         sumCalculators.push_back(new SorAbund());
 126                                                 }else if (Estimators[i] == "jclass") {
 127                                                         sumCalculators.push_back(new Jclass());
 128                                                 }else if (Estimators[i] == "sorclass") {
 129                                                         sumCalculators.push_back(new SorClass());
 130                                                 }else if (Estimators[i] == "jest") {
 131                                                         sumCalculators.push_back(new Jest());
 132                                                 }else if (Estimators[i] == "sorest") {
 133                                                         sumCalculators.push_back(new SorEst());
 134                                                 }else if (Estimators[i] == "thetayc") {
 135                                                         sumCalculators.push_back(new ThetaYC());
 136                                                 }else if (Estimators[i] == "thetan") {
 137                                                         sumCalculators.push_back(new ThetaN());
 138                                                 }else if (Estimators[i] == "kstest") {
 139                                                         sumCalculators.push_back(new KSTest());
 140                                                 }else if (Estimators[i] == "sharednseqs") {
 141                                                         sumCalculators.push_back(new SharedNSeqs());
 142                                                 }else if (Estimators[i] == "ochiai") {
 143                                                         sumCalculators.push_back(new Ochiai());
 144                                                 }else if (Estimators[i] == "anderberg") {
 145                                                         sumCalculators.push_back(new Anderberg());
 146                                                 }else if (Estimators[i] == "kulczynski") {
 147                                                         sumCalculators.push_back(new Kulczynski());
 148                                                 }else if (Estimators[i] == "kulczynskicody") {
 149                                                         sumCalculators.push_back(new KulczynskiCody());
 150                                                 }else if (Estimators[i] == "lennon") {
 151                                                         sumCalculators.push_back(new Lennon());
 152                                                 }else if (Estimators[i] == "morisitahorn") {
 153                                                         sumCalculators.push_back(new MorHorn());
 154                                                 }else if (Estimators[i] == "braycurtis") {
 155                                                         sumCalculators.push_back(new BrayCurtis());
 156                                                 }else if (Estimators[i] == "whittaker") {
 157                                                         sumCalculators.push_back(new Whittaker());
 158                                                 }
 159                                         }
 160                                 }
 161
 162                                 outputFileName = ((getRootName(globaldata->inputFileName)) + "shared.summary");
 163                                 openOutputFile(outputFileName, outputFileHandle);
 164                                 mult = false;
 165                         }
 166                 }
 167         }
 168         catch(exception& e) {
 169                 cout << "Standard Error: " << e.what() << " has occurred in the SummarySharedCommand class Function SummarySharedCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 170                 exit(1);
 171         }
 172         catch(...) {
 173                 cout << "An unknown error has occurred in the SummarySharedCommand class function SummarySharedCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 174                 exit(1);
 175         }
 176 }
 177
 178 //**********************************************************************************************************************
 179
 180 void SummarySharedCommand::help(){
 181         try {
 182                 cout << "The summary.shared command can only be executed after a successful read.otu command." << "\n";
 183                 cout << "The summary.shared command parameters are label, line and calc.  No parameters are required, but you may not use " << "\n";
 184                 cout << "both the line and label parameters at the same time. The summary.shared command should be in the following format: " << "\n";
 185                 cout << "summary.shared(label=yourLabel, line=yourLines, calc=yourEstimators, groups=yourGroups)." << "\n";
 186                 cout << "Example summary.shared(label=unique-.01-.03, line=0,5,10, groups=B-C, calc=sharedchao-sharedace-jabund-sorensonabund-jclass-sorclass-jest-sorest-thetayc-thetan)." << "\n";
 187                 validCalculator->printCalc("sharedsummary", cout);
 188                 cout << "The default value for calc is sharedsobs-sharedchao-sharedace-jabund-sorensonabund-jclass-sorclass-jest-sorest-thetayc-thetan" << "\n";
 189                 cout << "The default value for groups is all the groups in your groupfile." << "\n";
 190                 cout << "The label and line parameters are used to analyze specific lines in your input." << "\n";
 191                 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";
 192                 cout << "Note: No spaces between parameter labels (i.e. line), '=' and parameters (i.e.yourLines)." << "\n" << "\n";
 193         }
 194         catch(exception& e) {
 195                 cout << "Standard Error: " << e.what() << " has occurred in the SummarySharedCommand class Function help. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 196                 exit(1);
 197         }
 198         catch(...) {
 199                 cout << "An unknown error has occurred in the SummarySharedCommand class function help. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 200                 exit(1);
 201         }
 202 }
 203
 204 //**********************************************************************************************************************
 205
 206 SummarySharedCommand::~SummarySharedCommand(){
 207         delete read;
 208         delete validCalculator;
 209 }
 210
 211 //**********************************************************************************************************************
 212
 213 int SummarySharedCommand::execute(){
 214         try {
 215
 216                 if (abort == true) { return 0; }
 217
 218                 int count = 1;
 219
 220                 //if the users entered no valid calculators don't execute command
 221                 if (sumCalculators.size() == 0) { return 0; }
 222                 //check if any calcs can do multiples
 223                 else{
 224                         for (int i = 0; i < sumCalculators.size(); i++) {
 225                                 if (sumCalculators[i]->getMultiple() == true) { mult = true; }
 226                         }
 227                 }
 228
 229                 //read first line
 230                 read = new ReadOTUFile(globaldata->inputFileName);
 231                 read->read(&*globaldata);
 232
 233                 input = globaldata->ginput;
 234                 lookup = input->getSharedRAbundVectors();
 235                 vector<SharedRAbundVector*> lastLookup = lookup;
 236
 237                 //output estimator names as column headers
 238                 outputFileHandle << "label" <<'\t' << "comparison" << '\t';
 239                 for(int i=0;i<sumCalculators.size();i++){
 240                         outputFileHandle << '\t' << sumCalculators[i]->getName();
 241                 }
 242                 outputFileHandle << endl;
 243
 244                 //create file and put column headers for multiple groups file
 245                 if (mult == true) {
 246                         outAllFileName = ((getRootName(globaldata->inputFileName)) + "sharedmultiple.summary");
 247                         openOutputFile(outAllFileName, outAll);
 248
 249                         outAll << "label" <<'\t' << "comparison" << '\t';
 250                         for(int i=0;i<sumCalculators.size();i++){
 251                                 if (sumCalculators[i]->getMultiple() == true) {
 252                                         outAll << '\t' << sumCalculators[i]->getName();
 253                                 }
 254                         }
 255                         outAll << endl;
 256                 }
 257
 258                 if (lookup.size() < 2) {
 259                         cout << "I cannot run the command without at least 2 valid groups.";
 260                         for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; }
 261
 262                         //close files and clean up
 263                         outputFileHandle.close();  remove(outputFileName.c_str());
 264                         if (mult == true) {  outAll.close();  remove(outAllFileName.c_str());  }
 265                         return 0;
 266                 //if you only have 2 groups you don't need a .sharedmultiple file
 267                 }else if ((lookup.size() == 2) && (mult == true)) {
 268                         mult = false;
 269                         outAll.close();
 270                         remove(outAllFileName.c_str());
 271                 }
 272
 273                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 274                 set<string> processedLabels;
 275                 set<string> userLabels = labels;
 276                 set<int> userLines = lines;
 277
 278                 //as long as you are not at the end of the file or done wih the lines you want
 279                 while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0) || (userLines.size() != 0))) {
 280
 281                         if(allLines == 1 || lines.count(count) == 1 || labels.count(lookup[0]->getLabel()) == 1){
 282                                 cout << lookup[0]->getLabel() << '\t' << count << endl;
 283                                 process(lookup);
 284
 285                                 processedLabels.insert(lookup[0]->getLabel());
 286                                 userLabels.erase(lookup[0]->getLabel());
 287                                 userLines.erase(count);
 288                         }
 289
 290                         if ((anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLookup[0]->getLabel()) != 1)) {
 291                                         cout << lastLookup[0]->getLabel() << '\t' << count << endl;
 292                                         process(lastLookup);
 293
 294                                         processedLabels.insert(lastLookup[0]->getLabel());
 295                                         userLabels.erase(lastLookup[0]->getLabel());
 296                         }
 297
 298
 299                         //prevent memory leak
 300                         if (count != 1) { for (int i = 0; i < lastLookup.size(); i++) {  delete lastLookup[i];  } }
 301                         lastLookup = lookup;
 302
 303                         //get next line to process
 304                         lookup = input->getSharedRAbundVectors();
 305                         count++;
 306                 }
 307
 308                 //output error messages about any remaining user labels
 309                 set<string>::iterator it;
 310                 bool needToRun = false;
 311                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 312                         cout << "Your file does not include the label "<< *it;
 313                         if (processedLabels.count(lastLookup[0]->getLabel()) != 1) {
 314                                 cout << ". I will use " << lastLookup[0]->getLabel() << "." << endl;
 315                                 needToRun = true;
 316                         }else {
 317                                 cout << ". Please refer to " << lastLookup[0]->getLabel() << "." << endl;
 318                         }
 319                 }
 320
 321                 //run last line if you need to
 322                 if (needToRun == true)  {
 323                         cout << lastLookup[0]->getLabel() << '\t' << count << endl;
 324                         process(lastLookup);
 325                 }
 326
 327                 for (int i = 0; i < lastLookup.size(); i++) {  delete lastLookup[i];  }
 328
 329                 //reset groups parameter
 330                 globaldata->Groups.clear();
 331
 332                 //close files
 333                 outputFileHandle.close();
 334                 if (mult == true) {  outAll.close();  }
 335
 336                 for(int i=0;i<sumCalculators.size();i++){  delete sumCalculators[i]; }
 337
 338                 delete input;
 339
 340                 return 0;
 341         }
 342         catch(exception& e) {
 343                 cout << "Standard Error: " << e.what() << " has occurred in the SummarySharedCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 344                 exit(1);
 345         }
 346         catch(...) {
 347                 cout << "An unknown error has occurred in the SummarySharedCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 348                 exit(1);
 349         }
 350 }
 351
 352 /***********************************************************/
 353 void SummarySharedCommand::process(vector<SharedRAbundVector*> thisLookup) {
 354         try {
 355                                 //loop through calculators and add to file all for all calcs that can do mutiple groups
 356                                 if (mult == true) {
 357                                         //output label
 358                                         outAll << thisLookup[0]->getLabel() << '\t';
 359
 360                                         //output groups names
 361                                         string outNames = "";
 362                                         for (int j = 0; j < thisLookup.size(); j++) {
 363                                                 outNames += thisLookup[j]->getGroup() +  "-";
 364                                         }
 365                                         outNames = outNames.substr(0, outNames.length()-1); //rip off extra '-';
 366                                         outAll << outNames << '\t';
 367
 368                                         for(int i=0;i<sumCalculators.size();i++){
 369                                                 if (sumCalculators[i]->getMultiple() == true) {
 370                                                         sumCalculators[i]->getValues(thisLookup);
 371                                                         outAll << '\t';
 372                                                         sumCalculators[i]->print(outAll);
 373                                                 }
 374                                         }
 375                                         outAll << endl;
 376                                 }
 377
 378                                 int n = 1;
 379                                 vector<SharedRAbundVector*> subset;
 380                                 for (int k = 0; k < (thisLookup.size() - 1); k++) { // pass cdd each set of groups to commpare
 381                                         for (int l = n; l < thisLookup.size(); l++) {
 382
 383                                                 outputFileHandle << thisLookup[0]->getLabel() << '\t';
 384
 385                                                 subset.clear(); //clear out old pair of sharedrabunds
 386                                                 //add new pair of sharedrabunds
 387                                                 subset.push_back(thisLookup[k]); subset.push_back(thisLookup[l]);
 388
 389                                                 //sort groups to be alphanumeric
 390                                                 if (thisLookup[k]->getGroup() > thisLookup[l]->getGroup()) {
 391                                                         outputFileHandle << (thisLookup[l]->getGroup() +'\t' + thisLookup[k]->getGroup()) << '\t'; //print out groups
 392                                                 }else{
 393                                                         outputFileHandle << (thisLookup[k]->getGroup() +'\t' + thisLookup[l]->getGroup()) << '\t'; //print out groups
 394                                                 }
 395
 396                                                 for(int i=0;i<sumCalculators.size();i++) {
 397
 398                                                         sumCalculators[i]->getValues(subset); //saves the calculator outputs
 399                                                         outputFileHandle << '\t';
 400                                                         sumCalculators[i]->print(outputFileHandle);
 401                                                 }
 402                                                 outputFileHandle << endl;
 403                                         }
 404                                         n++;
 405                                 }
 406
 407         }
 408         catch(exception& e) {
 409                 cout << "Standard Error: " << e.what() << " has occurred in the SummarySharedCommand class Function process. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 410                 exit(1);
 411         }
 412         catch(...) {
 413                 cout << "An unknown error has occurred in the SummarySharedCommand class function process. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 414                 exit(1);
 415         }
 416 }
 417
 418 /***********************************************************/