venncommand.cpp

   1 /*
   2  *  venncommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by Sarah Westcott on 3/30/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10 #include "venncommand.h"
  11 #include "ace.h"
  12 #include "sobs.h"
  13 #include "chao1.h"
  14 //#include "jackknife.h"
  15 #include "sharedsobscollectsummary.h"
  16 #include "sharedchao1.h"
  17 #include "sharedace.h"
  18 #include "nseqs.h"
  19
  20
  21 //**********************************************************************************************************************
  22
  23 VennCommand::VennCommand(string option)  {
  24         try {
  25                 globaldata = GlobalData::getInstance();
  26                 abort = false;
  27                 allLines = 1;
  28                 labels.clear();
  29
  30                 //allow user to run help
  31                 if(option == "help") { help(); abort = true; }
  32
  33                 else {
  34                         //valid paramters for this command
  35                         string AlignArray[] =  {"groups","label","calc", "abund","outputdir","inputdir"};
  36                         vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
  37
  38                         OptionParser parser(option);
  39                         map<string,string> parameters = parser.getParameters();
  40
  41                         ValidParameters validParameter;
  42
  43                         //check to make sure all parameters are valid for command
  44                         for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
  45                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  46                         }
  47
  48                         //make sure the user has already run the read.otu command
  49                         if ((globaldata->getListFile() == "") && (globaldata->getSharedFile() == "")) {
  50                                 m->mothurOut("You must read a list, or a list and a group, or a shared before you can use the venn command."); m->mothurOutEndLine(); abort = true;
  51                         }
  52
  53                         //if the user changes the output directory command factory will send this info to us in the output parameter
  54                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
  55                                 outputDir = "";
  56                                 outputDir += m->hasPath(globaldata->inputFileName); //if user entered a file with a path then preserve it
  57                         }
  58
  59                         //check for optional parameter and set defaults
  60                         // ...at some point should added some additional type checking...
  61                         label = validParameter.validFile(parameters, "label", false);
  62                         if (label == "not found") { label = ""; }
  63                         else {
  64                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
  65                                 else { allLines = 1;  }
  66                         }
  67
  68                         //if the user has not specified any labels use the ones from read.otu
  69                         if (label == "") {
  70                                 allLines = globaldata->allLines;
  71                                 labels = globaldata->labels;
  72                         }
  73
  74                         groups = validParameter.validFile(parameters, "groups", false);
  75                         if (groups == "not found") { groups = ""; }
  76                         else {
  77                                 m->splitAtDash(groups, Groups);
  78                                 globaldata->Groups = Groups;
  79                         }
  80
  81                         format = globaldata->getFormat();
  82                         calc = validParameter.validFile(parameters, "calc", false);
  83                         if (calc == "not found") {
  84                                 if(format == "list") { calc = "sobs"; }
  85                                 else { calc = "sharedsobs"; }
  86                         }
  87                         else {
  88                                  if (calc == "default")  {
  89                                         if(format == "list") { calc = "sobs"; }
  90                                         else { calc = "sharedsobs"; }
  91                                 }
  92                         }
  93                         m->splitAtDash(calc, Estimators);
  94
  95                         string temp;
  96                         temp = validParameter.validFile(parameters, "abund", false);            if (temp == "not found") { temp = "10"; }
  97                         convert(temp, abund);
  98
  99                         if (abort == false) {
 100                                 validCalculator = new ValidCalculators();
 101
 102                                 int i;
 103
 104                                 if (format == "list") {
 105                                         for (i=0; i<Estimators.size(); i++) {
 106                                                 if (validCalculator->isValidCalculator("vennsingle", Estimators[i]) == true) {
 107                                                         if (Estimators[i] == "sobs") {
 108                                                                 vennCalculators.push_back(new Sobs());
 109                                                         }else if (Estimators[i] == "chao") {
 110                                                                 vennCalculators.push_back(new Chao1());
 111                                                         }else if (Estimators[i] == "ace") {
 112                                                                 if(abund < 5)
 113                                                                         abund = 10;
 114                                                                 vennCalculators.push_back(new Ace(abund));
 115                                                         }else if (Estimators[i] == "nseqs") {
 116                                                                 vennCalculators.push_back(new NSeqs());
 117                                                         }
 118                                                 }
 119                                         }
 120                                 }else {
 121                                         for (i=0; i<Estimators.size(); i++) {
 122                                                 if (validCalculator->isValidCalculator("vennshared", Estimators[i]) == true) {
 123                                                         if (Estimators[i] == "sharedsobs") {
 124                                                                 vennCalculators.push_back(new SharedSobsCS());
 125                                                         }else if (Estimators[i] == "sharedchao") {
 126                                                                 vennCalculators.push_back(new SharedChao1());
 127                                                         }else if (Estimators[i] == "sharedace") {
 128                                                                 vennCalculators.push_back(new SharedAce());
 129                                                         }else if (Estimators[i] == "nseqs") {
 130                                                                 vennCalculators.push_back(new NSeqs());
 131                                                         }
 132                                                 }
 133                                         }
 134                                 }
 135
 136                                 //if the users entered no valid calculators don't execute command
 137                                 if (vennCalculators.size() == 0) { m->mothurOut("No valid calculators given, please correct."); m->mothurOutEndLine(); abort = true;  }
 138                                 else {  venn = new Venn(outputDir);  }
 139                         }
 140
 141                 }
 142
 143
 144
 145         }
 146         catch(exception& e) {
 147                 m->errorOut(e, "VennCommand", "VennCommand");
 148                 exit(1);
 149         }
 150 }
 151
 152 //**********************************************************************************************************************
 153
 154 void VennCommand::help(){
 155         try {
 156                 m->mothurOut("The venn command can only be executed after a successful read.otu command.\n");
 157                 m->mothurOut("The venn command parameters are groups, calc, abund and label.  No parameters are required.\n");
 158                 m->mothurOut("The groups parameter allows you to specify which of the groups in your groupfile you would like included in your venn diagram, you may only use a maximum of 4 groups.\n");
 159                 m->mothurOut("The group names are separated by dashes. The label allows you to select what distance levels you would like a venn diagram created for, and are also separated by dashes.\n");
 160                 m->mothurOut("The venn command should be in the following format: venn(groups=yourGroups, calc=yourCalcs, label=yourLabels, abund=yourAbund).\n");
 161                 m->mothurOut("Example venn(groups=A-B-C, calc=sharedsobs-sharedchao, abund=20).\n");
 162                 m->mothurOut("The default value for groups is all the groups in your groupfile up to 4, and all labels in your inputfile will be used.\n");
 163                 m->mothurOut("The default value for calc is sobs if you have only read a list file or if you have selected only one group, and sharedsobs if you have multiple groups.\n");
 164                 m->mothurOut("The default available estimators for calc are sobs, chao and ace if you have only read a list file, and sharedsobs, sharedchao and sharedace if you have read a list and group file or a shared file.\n");
 165                 m->mothurOut("The only estmiator available four 4 groups is sharedsobs.\n");
 166                 m->mothurOut("The venn command outputs a .svg file for each calculator you specify at each distance you choose.\n");
 167                 m->mothurOut("Note: No spaces between parameter labels (i.e. groups), '=' and parameters (i.e.yourGroups).\n\n");
 168         }
 169         catch(exception& e) {
 170                 m->errorOut(e, "VennCommand", "help");
 171                 exit(1);
 172         }
 173 }
 174
 175
 176 //**********************************************************************************************************************
 177
 178 VennCommand::~VennCommand(){
 179         if (abort == false) {
 180                 delete input; globaldata->ginput = NULL;
 181                 delete read;
 182                 delete venn;
 183                 globaldata->sabund = NULL;
 184                 delete validCalculator;
 185         }
 186
 187 }
 188
 189 //**********************************************************************************************************************
 190
 191 int VennCommand::execute(){
 192         try {
 193
 194                 if (abort == true) { return 0; }
 195
 196                 string lastLabel;
 197                 vector<string> outputNames;
 198
 199                 if (format == "sharedfile") {
 200                         //you have groups
 201                         read = new ReadOTUFile(globaldata->inputFileName);
 202                         read->read(&*globaldata);
 203
 204                         input = globaldata->ginput;
 205                         lookup = input->getSharedRAbundVectors();
 206                         lastLabel = lookup[0]->getLabel();
 207                 }else if (format == "list") {
 208                         //you are using just a list file and have only one group
 209                         read = new ReadOTUFile(globaldata->inputFileName);
 210                         read->read(&*globaldata);
 211
 212                         sabund = globaldata->sabund;
 213                         lastLabel = sabund->getLabel();
 214                         input = globaldata->ginput;
 215                 }
 216
 217                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 218                 set<string> processedLabels;
 219                 set<string> userLabels = labels;
 220
 221                 if (format != "list") {
 222
 223                         //as long as you are not at the end of the file or done wih the lines you want
 224                         while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 225
 226                                 if (m->control_pressed) {
 227                                         for (int i = 0; i < vennCalculators.size(); i++) {      delete vennCalculators[i];      }
 228                                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 229                                         globaldata->Groups.clear();
 230                                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());  }
 231                                         return 0;
 232                                 }
 233
 234                                 if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){
 235                                         m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 236                                         processedLabels.insert(lookup[0]->getLabel());
 237                                         userLabels.erase(lookup[0]->getLabel());
 238
 239                                         if (lookup.size() > 4) {
 240                                                 m->mothurOut("Error: Too many groups chosen.  You may use up to 4 groups with the venn command.  I will use the first four groups in your groupfile."); m->mothurOutEndLine();
 241                                                 for (int i = lookup.size(); i > 4; i--) { lookup.pop_back(); } //no memmory leak because pop_back calls destructor
 242                                         }
 243
 244                                         vector<string> outfilenames = venn->getPic(lookup, vennCalculators);
 245                                         for(int i = 0; i < outfilenames.size(); i++) { if (outfilenames[i] != "control" ) { outputNames.push_back(outfilenames[i]); }  }
 246                                 }
 247
 248                                 if ((m->anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 249                                         string saveLabel = lookup[0]->getLabel();
 250
 251                                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 252                                         lookup = input->getSharedRAbundVectors(lastLabel);
 253
 254                                         m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 255                                         processedLabels.insert(lookup[0]->getLabel());
 256                                         userLabels.erase(lookup[0]->getLabel());
 257
 258                                         if (lookup.size() > 4) {
 259                                                 m->mothurOut("Error: Too many groups chosen.  You may use up to 4 groups with the venn command.  I will use the first four groups in your groupfile."); m->mothurOutEndLine();
 260                                                 for (int i = lookup.size(); i > 4; i--) { lookup.pop_back(); } //no memmory leak because pop_back calls destructor
 261                                         }
 262                                         vector<string> outfilenames = venn->getPic(lookup, vennCalculators);
 263                                         for(int i = 0; i < outfilenames.size(); i++) { if (outfilenames[i] != "control" ) { outputNames.push_back(outfilenames[i]); }  }
 264
 265                                         //restore real lastlabel to save below
 266                                         lookup[0]->setLabel(saveLabel);
 267                                 }
 268
 269
 270                                 lastLabel = lookup[0]->getLabel();
 271
 272                                 //get next line to process
 273                                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 274                                 lookup = input->getSharedRAbundVectors();
 275                         }
 276
 277                         if (m->control_pressed) {
 278                                         for (int i = 0; i < vennCalculators.size(); i++) {      delete vennCalculators[i];      }
 279                                         globaldata->Groups.clear();
 280                                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());  }
 281                                         return 0;
 282                         }
 283
 284
 285                         //output error messages about any remaining user labels
 286                         set<string>::iterator it;
 287                         bool needToRun = false;
 288                         for (it = userLabels.begin(); it != userLabels.end(); it++) {
 289                                 m->mothurOut("Your file does not include the label " + *it);
 290                                 if (processedLabels.count(lastLabel) != 1) {
 291                                         m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 292                                         needToRun = true;
 293                                 }else {
 294                                         m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 295                                 }
 296                         }
 297
 298                         //run last label if you need to
 299                         if (needToRun == true)  {
 300                                         for (int i = 0; i < lookup.size(); i++) {  if (lookup[i] != NULL) {     delete lookup[i]; }  }
 301                                         lookup = input->getSharedRAbundVectors(lastLabel);
 302
 303                                         m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 304                                         processedLabels.insert(lookup[0]->getLabel());
 305                                         userLabels.erase(lookup[0]->getLabel());
 306
 307                                         if (lookup.size() > 4) {
 308                                                 m->mothurOut("Error: Too many groups chosen.  You may use up to 4 groups with the venn command.  I will use the first four groups in your groupfile."); m->mothurOutEndLine();
 309                                                 for (int i = lookup.size(); i > 4; i--) { lookup.pop_back(); } //no memmory leak because pop_back calls destructor
 310                                         }
 311                                         vector<string> outfilenames = venn->getPic(lookup, vennCalculators);
 312                                         for(int i = 0; i < outfilenames.size(); i++) { if (outfilenames[i] != "control" ) { outputNames.push_back(outfilenames[i]); }  }
 313
 314                                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 315                         }
 316
 317
 318                         //reset groups parameter
 319                         globaldata->Groups.clear();
 320
 321                         if (m->control_pressed) {
 322                                         for (int i = 0; i < vennCalculators.size(); i++) {      delete vennCalculators[i];      }
 323                                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());  }
 324                                         return 0;
 325                         }
 326
 327
 328                 }else{
 329
 330                         while((sabund != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 331
 332                                 if (m->control_pressed) {
 333                                         for (int i = 0; i < vennCalculators.size(); i++) {      delete vennCalculators[i];      }
 334                                         delete sabund;
 335                                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());  }
 336                                         return 0;
 337                                 }
 338
 339                                 if(allLines == 1 || labels.count(sabund->getLabel()) == 1){
 340
 341                                         m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
 342                                         vector<string> outfilenames = venn->getPic(sabund, vennCalculators);
 343                                         for(int i = 0; i < outfilenames.size(); i++) { if (outfilenames[i] != "control" ) { outputNames.push_back(outfilenames[i]); }  }
 344
 345
 346                                         processedLabels.insert(sabund->getLabel());
 347                                         userLabels.erase(sabund->getLabel());
 348                                 }
 349
 350                                 if ((m->anyLabelsToProcess(sabund->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 351                                         string saveLabel = sabund->getLabel();
 352
 353                                         delete sabund;
 354                                         sabund = input->getSAbundVector(lastLabel);
 355
 356                                         m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
 357                                         vector<string> outfilenames = venn->getPic(sabund, vennCalculators);
 358                                         for(int i = 0; i < outfilenames.size(); i++) { if (outfilenames[i] != "control" ) { outputNames.push_back(outfilenames[i]); }  }
 359
 360
 361                                         processedLabels.insert(sabund->getLabel());
 362                                         userLabels.erase(sabund->getLabel());
 363
 364                                         //restore real lastlabel to save below
 365                                         sabund->setLabel(saveLabel);
 366                                 }
 367
 368                                 lastLabel = sabund->getLabel();
 369
 370                                 delete sabund;
 371                                 sabund = input->getSAbundVector();
 372                         }
 373
 374                         if (m->control_pressed) {
 375                                         for (int i = 0; i < vennCalculators.size(); i++) {      delete vennCalculators[i];      }
 376                                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());  }
 377                                         return 0;
 378                         }
 379
 380                         //output error messages about any remaining user labels
 381                         set<string>::iterator it;
 382                         bool needToRun = false;
 383                         for (it = userLabels.begin(); it != userLabels.end(); it++) {
 384                                 m->mothurOut("Your file does not include the label " + *it);
 385                                 if (processedLabels.count(lastLabel) != 1) {
 386                                         m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 387                                         needToRun = true;
 388                                 }else {
 389                                         m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 390                                 }
 391                         }
 392
 393                         //run last label if you need to
 394                         if (needToRun == true)  {
 395                                 if (sabund != NULL) {   delete sabund;  }
 396                                 sabund = input->getSAbundVector(lastLabel);
 397
 398                                 m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
 399                                 vector<string> outfilenames = venn->getPic(sabund, vennCalculators);
 400                                 for(int i = 0; i < outfilenames.size(); i++) { if (outfilenames[i] != "control" ) { outputNames.push_back(outfilenames[i]); }  }
 401
 402                                 delete sabund;
 403
 404                         }
 405
 406                         if (m->control_pressed) {
 407                                         for (int i = 0; i < vennCalculators.size(); i++) {      delete vennCalculators[i];      }
 408                                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());  }
 409                                         return 0;
 410                         }
 411                 }
 412
 413                 for (int i = 0; i < vennCalculators.size(); i++) {      delete vennCalculators[i];      }
 414
 415                 m->mothurOutEndLine();
 416                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 417                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 418                 m->mothurOutEndLine();
 419
 420
 421                 return 0;
 422         }
 423         catch(exception& e) {
 424                 m->errorOut(e, "VennCommand", "execute");
 425                 exit(1);
 426         }
 427 }
 428
 429 //**********************************************************************************************************************