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
  19
  20 //**********************************************************************************************************************
  21
  22 VennCommand::VennCommand(string option){
  23         try {
  24                 globaldata = GlobalData::getInstance();
  25                 abort = false;
  26                 allLines = 1;
  27                 lines.clear();
  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","line","label","calc", "abund"};
  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                                  cout << "You must read a list, or a list and a group, or a shared before you can use the venn command." << endl; abort = true;
  51                         }
  52
  53                         //check for optional parameter and set defaults
  54                         // ...at some point should added some additional type checking...
  55                         line = validParameter.validFile(parameters, "line", false);
  56                         if (line == "not found") { line = "";  }
  57                         else {
  58                                 if(line != "all") {  splitAtDash(line, lines);  allLines = 0;  }
  59                                 else { allLines = 1;  }
  60                         }
  61
  62                         label = validParameter.validFile(parameters, "label", false);
  63                         if (label == "not found") { label = ""; }
  64                         else {
  65                                 if(label != "all") {  splitAtDash(label, labels);  allLines = 0;  }
  66                                 else { allLines = 1;  }
  67                         }
  68
  69                         //make sure user did not use both the line and label parameters
  70                         if ((line != "") && (label != "")) { cout << "You cannot use both the line and label parameters at the same time. " << endl; abort = true; }
  71                         //if the user has not specified any line or labels use the ones from read.otu
  72                         else if ((line == "") && (label == "")) {
  73                                 allLines = globaldata->allLines;
  74                                 labels = globaldata->labels;
  75                                 lines = globaldata->lines;
  76                         }
  77
  78                         groups = validParameter.validFile(parameters, "groups", false);
  79                         if (groups == "not found") { groups = ""; }
  80                         else {
  81                                 splitAtDash(groups, Groups);
  82                                 globaldata->Groups = Groups;
  83                         }
  84
  85                         format = globaldata->getFormat();
  86                         calc = validParameter.validFile(parameters, "calc", false);
  87                         if (calc == "not found") {
  88                                 if(format == "list") { calc = "sobs"; }
  89                                 else { calc = "sharedsobs"; }
  90                         }
  91                         else {
  92                                  if (calc == "default")  {
  93                                         if(format == "list") { calc = "sobs"; }
  94                                         else { calc = "sharedsobs"; }
  95                                 }
  96                         }
  97                         splitAtDash(calc, Estimators);
  98
  99                         string temp;
 100                         temp = validParameter.validFile(parameters, "abund", false);            if (temp == "not found") { temp = "10"; }
 101                         convert(temp, abund);
 102
 103                         if (abort == false) {
 104                                 validCalculator = new ValidCalculators();
 105
 106                                 int i;
 107
 108                                 if (format == "list") {
 109                                         for (i=0; i<Estimators.size(); i++) {
 110                                                 if (validCalculator->isValidCalculator("vennsingle", Estimators[i]) == true) {
 111                                                         if (Estimators[i] == "sobs") {
 112                                                                 vennCalculators.push_back(new Sobs());
 113                                                         }else if (Estimators[i] == "chao") {
 114                                                                 vennCalculators.push_back(new Chao1());
 115                                                         }else if (Estimators[i] == "ace") {
 116                                                                 if(abund < 5)
 117                                                                         abund = 10;
 118                                                                 vennCalculators.push_back(new Ace(abund));
 119                                                         }
 120                                                 }
 121                                         }
 122                                 }else {
 123                                         for (i=0; i<Estimators.size(); i++) {
 124                                                 if (validCalculator->isValidCalculator("vennshared", Estimators[i]) == true) {
 125                                                         if (Estimators[i] == "sharedsobs") {
 126                                                                 vennCalculators.push_back(new SharedSobsCS());
 127                                                         }else if (Estimators[i] == "sharedchao") {
 128                                                                 vennCalculators.push_back(new SharedChao1());
 129                                                         }else if (Estimators[i] == "sharedace") {
 130                                                                 vennCalculators.push_back(new SharedAce());
 131                                                         }
 132                                                 }
 133                                         }
 134                                 }
 135
 136                                 venn = new Venn();
 137                         }
 138                 }
 139
 140
 141
 142         }
 143         catch(exception& e) {
 144                 cout << "Standard Error: " << e.what() << " has occurred in the VennCommand class Function VennCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 145                 exit(1);
 146         }
 147         catch(...) {
 148                 cout << "An unknown error has occurred in the VennCommand class function VennCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 149                 exit(1);
 150         }
 151 }
 152
 153 //**********************************************************************************************************************
 154
 155 void VennCommand::help(){
 156         try {
 157                 cout << "The venn command can only be executed after a successful read.otu command." << "\n";
 158                 cout << "The venn command parameters are groups, calc, abund, line and label.  No parameters are required, but you may not use line and label at the same time." << "\n";
 159                 cout << "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";
 160                 cout << "The group names are separated by dashes. The line and label allow you to select what distance levels you would like a venn diagram created for, and are also separated by dashes." << "\n";
 161                 cout << "The venn command should be in the following format: venn(groups=yourGroups, calc=yourCalcs, line=yourLines, label=yourLabels, abund=yourAbund)." << "\n";
 162                 cout << "Example venn(groups=A-B-C, line=1-3-5, calc=sharedsobs-sharedchao, abund=20)." << "\n";
 163                 cout << "The default value for groups is all the groups in your groupfile up to 4, and all lines in your inputfile will be used." << "\n";
 164                 cout << "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";
 165                 cout << "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";
 166                 cout << "The only estmiator available four 4 groups is sharedsobs." << "\n";
 167                 cout << "The venn command outputs a .svg file for each calculator you specify at each distance you choose." << "\n";
 168                 cout << "Note: No spaces between parameter labels (i.e. groups), '=' and parameters (i.e.yourGroups)." << "\n" << "\n";
 169         }
 170         catch(exception& e) {
 171                 cout << "Standard Error: " << e.what() << " has occurred in the VennCommand class Function help. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 172                 exit(1);
 173         }
 174         catch(...) {
 175                 cout << "An unknown error has occurred in the VennCommand class function help. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 176                 exit(1);
 177         }
 178 }
 179
 180
 181 //**********************************************************************************************************************
 182
 183 VennCommand::~VennCommand(){
 184         if (abort == false) {
 185                 delete input; globaldata->ginput = NULL;
 186                 delete read;
 187                 delete venn;
 188                 globaldata->sabund = NULL;
 189         }
 190
 191 }
 192
 193 //**********************************************************************************************************************
 194
 195 int VennCommand::execute(){
 196         try {
 197
 198                 if (abort == true) { return 0; }
 199
 200                 int count = 1;
 201                 string lastLabel;
 202
 203                 //if the users entered no valid calculators don't execute command
 204                 if (vennCalculators.size() == 0) { return 0; }
 205
 206                 if (format == "sharedfile") {
 207                         //you have groups
 208                         read = new ReadOTUFile(globaldata->inputFileName);
 209                         read->read(&*globaldata);
 210
 211                         input = globaldata->ginput;
 212                         lookup = input->getSharedRAbundVectors();
 213                         lastLabel = lookup[0]->getLabel();
 214                 }else if (format == "list") {
 215                         //you are using just a list file and have only one group
 216                         read = new ReadOTUFile(globaldata->inputFileName);
 217                         read->read(&*globaldata);
 218
 219                         sabund = globaldata->sabund;
 220                         lastLabel = sabund->getLabel();
 221                         input = globaldata->ginput;
 222                 }
 223
 224                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 225                 set<string> processedLabels;
 226                 set<string> userLabels = labels;
 227                 set<int> userLines = lines;
 228
 229                 if (format != "list") {
 230
 231                         //as long as you are not at the end of the file or done wih the lines you want
 232                         while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0) || (userLines.size() != 0))) {
 233
 234                                 if(allLines == 1 || lines.count(count) == 1 || labels.count(lookup[0]->getLabel()) == 1){
 235                                         cout << lookup[0]->getLabel() << '\t' << count << endl;
 236                                         processedLabels.insert(lookup[0]->getLabel());
 237                                         userLabels.erase(lookup[0]->getLabel());
 238                                         userLines.erase(count);
 239
 240                                         if (lookup.size() > 4) {
 241                                                 cout << "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." << endl;
 242                                                 for (int i = lookup.size(); i > 4; i--) { lookup.pop_back(); } //no memmory leak because pop_back calls destructor
 243                                         }
 244                                         venn->getPic(lookup, vennCalculators);
 245                                 }
 246
 247                                 if ((anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 248                                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 249                                         lookup = input->getSharedRAbundVectors(lastLabel);
 250
 251                                         cout << lookup[0]->getLabel() << '\t' << count << endl;
 252                                         processedLabels.insert(lookup[0]->getLabel());
 253                                         userLabels.erase(lookup[0]->getLabel());
 254
 255                                         if (lookup.size() > 4) {
 256                                                 cout << "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." << endl;
 257                                                 for (int i = lookup.size(); i > 4; i--) { lookup.pop_back(); } //no memmory leak because pop_back calls destructor
 258                                         }
 259                                         venn->getPic(lookup, vennCalculators);
 260                                 }
 261
 262
 263                                 lastLabel = lookup[0]->getLabel();
 264
 265                                 //get next line to process
 266                                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 267                                 lookup = input->getSharedRAbundVectors();
 268                                 count++;
 269                         }
 270
 271                         //output error messages about any remaining user labels
 272                         set<string>::iterator it;
 273                         bool needToRun = false;
 274                         for (it = userLabels.begin(); it != userLabels.end(); it++) {
 275                                 cout << "Your file does not include the label "<< *it;
 276                                 if (processedLabels.count(lastLabel) != 1) {
 277                                         cout << ". I will use " << lastLabel << "." << endl;
 278                                         needToRun = true;
 279                                 }else {
 280                                         cout << ". Please refer to " << lastLabel << "." << endl;
 281                                 }
 282                         }
 283
 284                         //run last line if you need to
 285                         if (needToRun == true)  {
 286                                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 287                                         lookup = input->getSharedRAbundVectors(lastLabel);
 288
 289                                         cout << lookup[0]->getLabel() << '\t' << count << endl;
 290                                         processedLabels.insert(lookup[0]->getLabel());
 291                                         userLabels.erase(lookup[0]->getLabel());
 292
 293                                         if (lookup.size() > 4) {
 294                                                 cout << "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." << endl;
 295                                                 for (int i = lookup.size(); i > 4; i--) { lookup.pop_back(); } //no memmory leak because pop_back calls destructor
 296                                         }
 297                                         venn->getPic(lookup, vennCalculators);
 298                                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 299                         }
 300
 301
 302                         //reset groups parameter
 303                         globaldata->Groups.clear();
 304
 305                 }else{
 306
 307                         while((sabund != NULL) && ((allLines == 1) || (userLabels.size() != 0) || (userLines.size() != 0))) {
 308
 309                                 if(allLines == 1 || lines.count(count) == 1 || labels.count(sabund->getLabel()) == 1){
 310
 311                                         cout << sabund->getLabel() << '\t' << count << endl;
 312                                         venn->getPic(sabund, vennCalculators);
 313
 314                                         processedLabels.insert(sabund->getLabel());
 315                                         userLabels.erase(sabund->getLabel());
 316                                         userLines.erase(count);
 317                                 }
 318
 319                                 if ((anyLabelsToProcess(sabund->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 320                                         delete sabund;
 321                                         sabund = input->getSAbundVector(lastLabel);
 322
 323                                         cout << sabund->getLabel() << '\t' << count << endl;
 324                                         venn->getPic(sabund, vennCalculators);
 325
 326                                         processedLabels.insert(sabund->getLabel());
 327                                         userLabels.erase(sabund->getLabel());
 328                                 }
 329
 330                                 lastLabel = sabund->getLabel();
 331
 332                                 delete sabund;
 333                                 sabund = input->getSAbundVector();
 334                                 count++;
 335                         }
 336
 337                         //output error messages about any remaining user labels
 338                         set<string>::iterator it;
 339                         bool needToRun = false;
 340                         for (it = userLabels.begin(); it != userLabels.end(); it++) {
 341                                 cout << "Your file does not include the label "<< *it;
 342                                 if (processedLabels.count(lastLabel) != 1) {
 343                                         cout << ". I will use " << lastLabel << "." << endl;
 344                                         needToRun = true;
 345                                 }else {
 346                                         cout << ". Please refer to " << lastLabel << "." << endl;
 347                                 }
 348                         }
 349
 350                         //run last line if you need to
 351                         if (needToRun == true)  {
 352                                 delete sabund;
 353                                 sabund = input->getSAbundVector(lastLabel);
 354
 355                                 cout << sabund->getLabel() << '\t' << count << endl;
 356                                 venn->getPic(sabund, vennCalculators);
 357                                 delete sabund;
 358
 359                         }
 360
 361                 }
 362
 363                 for (int i = 0; i < vennCalculators.size(); i++) {      delete vennCalculators[i];      }
 364                 return 0;
 365         }
 366         catch(exception& e) {
 367                 cout << "Standard Error: " << e.what() << " has occurred in the VennCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 368                 exit(1);
 369         }
 370         catch(...) {
 371                 cout << "An unknown error has occurred in the VennCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 372                 exit(1);
 373         }
 374 }
 375
 376 //**********************************************************************************************************************