removerarecommand.cpp

   1 /*
   2  *  removerarecommand.cpp
   3  *  mothur
   4  *
   5  *  Created by westcott on 1/21/11.
   6  *  Copyright 2011 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "removerarecommand.h"
  11 #include "sequence.hpp"
  12 #include "groupmap.h"
  13 #include "sharedutilities.h"
  14 #include "inputdata.h"
  15
  16 //**********************************************************************************************************************
  17 vector<string> RemoveRareCommand::setParameters(){
  18         try {
  19                 CommandParameter plist("list", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(plist);
  20                 CommandParameter prabund("rabund", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(prabund);
  21                 CommandParameter psabund("sabund", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(psabund);
  22                 CommandParameter pshared("shared", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pshared);
  23                 CommandParameter pgroup("group", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pgroup);
  24                 CommandParameter pgroups("groups", "String", "", "", "", "", "",false,false); parameters.push_back(pgroups);
  25                 CommandParameter plabel("label", "String", "", "", "", "", "",false,false); parameters.push_back(plabel);
  26                 CommandParameter pnseqs("nseqs", "Number", "", "0", "", "", "",false,true); parameters.push_back(pnseqs);
  27                 CommandParameter pbygroup("bygroup", "Boolean", "", "f", "", "", "",false,true); parameters.push_back(pbygroup);
  28                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  29                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  30
  31                 vector<string> myArray;
  32                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  33                 return myArray;
  34         }
  35         catch(exception& e) {
  36                 m->errorOut(e, "RemoveRareCommand", "setParameters");
  37                 exit(1);
  38         }
  39 }
  40 //**********************************************************************************************************************
  41 string RemoveRareCommand::getHelpString(){
  42         try {
  43                 string helpString = "";
  44                 helpString += "The remove.rare command parameters are list, rabund, sabund, shared, group, label, groups, bygroup and nseqs.\n";
  45                 helpString += "The remove.rare command reads one of the following file types: list, rabund, sabund or shared file. It outputs a new file after removing the rare otus.\n";
  46                 helpString += "The groups parameter allows you to specify which of the groups you would like analyzed.  Default=all. You may separate group names with dashes.\n";
  47                 helpString += "The label parameter is used to analyze specific labels in your input. default=all. You may separate label names with dashes.\n";
  48                 helpString += "The bygroup parameter is only valid with the shared file. default=f, meaning remove any OTU that has nseqs or fewer sequences across all groups.\n";
  49                 helpString += "bygroups=T means remove any OTU that has nseqs or fewer sequences in each group (if groupA has 1 sequence and group B has 100 sequences in OTUZ and nseqs=1, then set the groupA count for OTUZ to 0 and keep groupB's count at 100.) \n";
  50                 helpString += "The nseqs parameter allows you to set the cutoff for an otu to be deemed rare. It is required.\n";
  51                 helpString += "The remove.rare command should be in the following format: remove.rare(shared=yourSharedFile, nseqs=yourRareCutoff).\n";
  52                 helpString += "Example remove.rare(shared=amazon.fn.shared, nseqs=2).\n";
  53                 helpString += "Note: No spaces between parameter labels (i.e. shared), '=' and parameters (i.e.yourSharedFile).\n";
  54                 return helpString;
  55         }
  56         catch(exception& e) {
  57                 m->errorOut(e, "RemoveRareCommand", "getHelpString");
  58                 exit(1);
  59         }
  60 }
  61 //**********************************************************************************************************************
  62 RemoveRareCommand::RemoveRareCommand(){
  63         try {
  64                 abort = true; calledHelp = true;
  65                 setParameters();
  66                 vector<string> tempOutNames;
  67                 outputTypes["rabund"] = tempOutNames;
  68                 outputTypes["sabund"] = tempOutNames;
  69                 outputTypes["list"] = tempOutNames;
  70                 outputTypes["group"] = tempOutNames;
  71                 outputTypes["shared"] = tempOutNames;
  72         }
  73         catch(exception& e) {
  74                 m->errorOut(e, "RemoveRareCommand", "RemoveRareCommand");
  75                 exit(1);
  76         }
  77 }
  78 //**********************************************************************************************************************
  79 RemoveRareCommand::RemoveRareCommand(string option)  {
  80         try {
  81                 abort = false; calledHelp = false;
  82                 allLines = 1;
  83
  84                 //allow user to run help
  85                 if(option == "help") { help(); abort = true; calledHelp = true; }
  86
  87                 else {
  88                         vector<string> myArray = setParameters();
  89
  90                         OptionParser parser(option);
  91                         map<string,string> parameters = parser.getParameters();
  92
  93                         ValidParameters validParameter;
  94                         map<string,string>::iterator it;
  95
  96                         //check to make sure all parameters are valid for command
  97                         for (it = parameters.begin(); it != parameters.end(); it++) {
  98                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  99                         }
 100
 101                         //initialize outputTypes
 102                         vector<string> tempOutNames;
 103                         outputTypes["rabund"] = tempOutNames;
 104                         outputTypes["sabund"] = tempOutNames;
 105                         outputTypes["list"] = tempOutNames;
 106                         outputTypes["group"] = tempOutNames;
 107                         outputTypes["shared"] = tempOutNames;
 108
 109                         //if the user changes the output directory command factory will send this info to us in the output parameter
 110                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 111
 112                         //if the user changes the input directory command factory will send this info to us in the output parameter
 113                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 114                         if (inputDir == "not found"){   inputDir = "";          }
 115                         else {
 116                                 string path;
 117                                 it = parameters.find("list");
 118                                 //user has given a template file
 119                                 if(it != parameters.end()){
 120                                         path = m->hasPath(it->second);
 121                                         //if the user has not given a path then, add inputdir. else leave path alone.
 122                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 123                                 }
 124
 125                                 it = parameters.find("group");
 126                                 //user has given a template file
 127                                 if(it != parameters.end()){
 128                                         path = m->hasPath(it->second);
 129                                         //if the user has not given a path then, add inputdir. else leave path alone.
 130                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 131                                 }
 132
 133                                 it = parameters.find("sabund");
 134                                 //user has given a template file
 135                                 if(it != parameters.end()){
 136                                         path = m->hasPath(it->second);
 137                                         //if the user has not given a path then, add inputdir. else leave path alone.
 138                                         if (path == "") {       parameters["sabund"] = inputDir + it->second;           }
 139                                 }
 140
 141                                 it = parameters.find("rabund");
 142                                 //user has given a template file
 143                                 if(it != parameters.end()){
 144                                         path = m->hasPath(it->second);
 145                                         //if the user has not given a path then, add inputdir. else leave path alone.
 146                                         if (path == "") {       parameters["rabund"] = inputDir + it->second;           }
 147                                 }
 148
 149                                 it = parameters.find("shared");
 150                                 //user has given a template file
 151                                 if(it != parameters.end()){
 152                                         path = m->hasPath(it->second);
 153                                         //if the user has not given a path then, add inputdir. else leave path alone.
 154                                         if (path == "") {       parameters["shared"] = inputDir + it->second;           }
 155                                 }
 156                         }
 157
 158
 159                         //check for file parameters
 160                         listfile = validParameter.validFile(parameters, "list", true);
 161                         if (listfile == "not open") { abort = true; }
 162                         else if (listfile == "not found") {  listfile = "";  }
 163
 164                         sabundfile = validParameter.validFile(parameters, "sabund", true);
 165                         if (sabundfile == "not open") { abort = true; }
 166                         else if (sabundfile == "not found") {  sabundfile = "";  }
 167
 168                         rabundfile = validParameter.validFile(parameters, "rabund", true);
 169                         if (rabundfile == "not open") { abort = true; }
 170                         else if (rabundfile == "not found") {  rabundfile = "";  }
 171
 172                         groupfile = validParameter.validFile(parameters, "group", true);
 173                         if (groupfile == "not open") { groupfile = ""; abort = true; }
 174                         else if (groupfile == "not found") {  groupfile = "";  }
 175
 176                         sharedfile = validParameter.validFile(parameters, "shared", true);
 177                         if (sharedfile == "not open") { sharedfile = "";  abort = true; }
 178                         else if (sharedfile == "not found") {  sharedfile = "";  }
 179
 180                         if ((sharedfile == "") && (listfile == "") && (rabundfile == "") && (sabundfile == "")) {
 181                                 //is there are current file available for any of these?
 182                                 //give priority to shared, then list, then rabund, then sabund
 183                                 //if there is a current shared file, use it
 184                                 sharedfile = m->getSharedFile();
 185                                 if (sharedfile != "") {  m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); }
 186                                 else {
 187                                         listfile = m->getListFile();
 188                                         if (listfile != "") {  m->mothurOut("Using " + listfile + " as input file for the list parameter."); m->mothurOutEndLine(); }
 189                                         else {
 190                                                 rabundfile = m->getRabundFile();
 191                                                 if (rabundfile != "") {  m->mothurOut("Using " + rabundfile + " as input file for the rabund parameter."); m->mothurOutEndLine(); }
 192                                                 else {
 193                                                         sabundfile = m->getSabundFile();
 194                                                         if (sabundfile != "") {  m->mothurOut("Using " + sabundfile + " as input file for the sabund parameter."); m->mothurOutEndLine(); }
 195                                                         else {
 196                                                                 m->mothurOut("No valid current files. You must provide a list, sabund, rabund or shared file."); m->mothurOutEndLine();
 197                                                                 abort = true;
 198                                                         }
 199                                                 }
 200                                         }
 201                                 }
 202                         }
 203
 204                         groups = validParameter.validFile(parameters, "groups", false);
 205                         if (groups == "not found") { groups = "all"; }
 206                         m->splitAtDash(groups, Groups);
 207
 208                         label = validParameter.validFile(parameters, "label", false);
 209                         if (label == "not found") { label = ""; }
 210                         else {
 211                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 212                                 else { allLines = 1;  }
 213                         }
 214
 215                         string temp = validParameter.validFile(parameters, "nseqs", false);
 216                         if (temp == "not found") { m->mothurOut("nseqs is a required parameter."); m->mothurOutEndLine(); abort = true; }
 217                         else { convert(temp, nseqs); }
 218
 219                         temp = validParameter.validFile(parameters, "bygroup", false);   if (temp == "not found") { temp = "f"; }
 220                         byGroup = m->isTrue(temp);
 221
 222                         if (byGroup && (sharedfile == "")) { m->mothurOut("The byGroup parameter is only valid with a shared file."); m->mothurOutEndLine(); }
 223
 224                         if ((groupfile != "") && (listfile == "")) { m->mothurOut("A groupfile is only valid with a list file."); m->mothurOutEndLine(); groupfile = ""; }
 225                 }
 226
 227         }
 228         catch(exception& e) {
 229                 m->errorOut(e, "RemoveRareCommand", "RemoveRareCommand");
 230                 exit(1);
 231         }
 232 }
 233 //**********************************************************************************************************************
 234
 235 int RemoveRareCommand::execute(){
 236         try {
 237
 238                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 239
 240                 if (m->control_pressed) { return 0; }
 241
 242                 //read through the correct file and output lines you want to keep
 243                 if (sabundfile != "")           {               processSabund();        }
 244                 if (rabundfile != "")           {               processRabund();        }
 245                 if (listfile != "")                     {               processList();          }
 246                 if (sharedfile != "")           {               processShared();        }
 247
 248                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str()); } return 0; }
 249
 250                 if (outputNames.size() != 0) {
 251                         m->mothurOutEndLine();
 252                         m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 253                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 254                         m->mothurOutEndLine();
 255
 256                         //set rabund file as new current rabundfile
 257                         string current = "";
 258                         itTypes = outputTypes.find("rabund");
 259                         if (itTypes != outputTypes.end()) {
 260                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setRabundFile(current); }
 261                         }
 262
 263                         itTypes = outputTypes.find("sabund");
 264                         if (itTypes != outputTypes.end()) {
 265                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setSabundFile(current); }
 266                         }
 267
 268                         itTypes = outputTypes.find("group");
 269                         if (itTypes != outputTypes.end()) {
 270                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setGroupFile(current); }
 271                         }
 272
 273                         itTypes = outputTypes.find("list");
 274                         if (itTypes != outputTypes.end()) {
 275                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 276                         }
 277
 278                         itTypes = outputTypes.find("shared");
 279                         if (itTypes != outputTypes.end()) {
 280                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setSharedFile(current); }
 281                         }
 282                 }
 283
 284                 return 0;
 285         }
 286
 287         catch(exception& e) {
 288                 m->errorOut(e, "RemoveRareCommand", "execute");
 289                 exit(1);
 290         }
 291 }
 292
 293 //**********************************************************************************************************************
 294 int RemoveRareCommand::processList(){
 295         try {
 296                 string thisOutputDir = outputDir;
 297                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 298                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "pick" +  m->getExtension(listfile);
 299                 string outputGroupFileName = thisOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "pick" +  m->getExtension(groupfile);
 300
 301                 ofstream out, outGroup;
 302                 m->openOutputFile(outputFileName, out);
 303
 304                 bool wroteSomething = false;
 305
 306                 //you must provide a label because the names in the listfile need to be consistent
 307                 string thisLabel = "";
 308                 if (allLines) { m->mothurOut("For the listfile you must select one label, using first label in your listfile."); m->mothurOutEndLine(); }
 309                 else if (labels.size() > 1) { m->mothurOut("For the listfile you must select one label, using " + (*labels.begin()) + "."); m->mothurOutEndLine(); thisLabel = *labels.begin(); }
 310                 else { thisLabel = *labels.begin(); }
 311
 312                 InputData input(listfile, "list");
 313                 ListVector* list = input.getListVector();
 314
 315                 //get first one or the one we want
 316                 if (thisLabel != "") {
 317                         //use smart distancing
 318                         set<string> userLabels; userLabels.insert(thisLabel);
 319                         set<string> processedLabels;
 320                         string lastLabel = list->getLabel();
 321                         while((list != NULL) && (userLabels.size() != 0)) {
 322                                 if(userLabels.count(list->getLabel()) == 1){
 323                                         processedLabels.insert(list->getLabel());
 324                                         userLabels.erase(list->getLabel());
 325                                         break;
 326                                 }
 327
 328                                 if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 329                                         processedLabels.insert(list->getLabel());
 330                                         userLabels.erase(list->getLabel());
 331                                         delete list;
 332                                         list = input.getListVector(lastLabel);
 333                                         break;
 334                                 }
 335                                 lastLabel = list->getLabel();
 336                                 delete list;
 337                                 list = input.getListVector();
 338                         }
 339                         if (userLabels.size() != 0) {
 340                                 m->mothurOut("Your file does not include the label " + thisLabel + ". I will use " + lastLabel + ".");  m->mothurOutEndLine();
 341                                 list = input.getListVector(lastLabel);
 342                         }
 343                 }
 344
 345                 //if groupfile is given then use it
 346                 GroupMap* groupMap;
 347                 if (groupfile != "") {
 348                         groupMap = new GroupMap(groupfile); groupMap->readMap();
 349                         SharedUtil util;
 350                         util.setGroups(Groups, groupMap->namesOfGroups);
 351                         m->openOutputFile(outputGroupFileName, outGroup);
 352                 }
 353
 354
 355                 if (list != NULL) {
 356                         //make a new list vector
 357                         ListVector newList;
 358                         newList.setLabel(list->getLabel());
 359
 360                         //for each bin
 361                         for (int i = 0; i < list->getNumBins(); i++) {
 362                                 if (m->control_pressed) {  if (groupfile != "") { delete groupMap; outGroup.close(); remove(outputGroupFileName.c_str()); } out.close();  remove(outputFileName.c_str());  return 0; }
 363
 364                                 //parse out names that are in accnos file
 365                                 string binnames = list->get(i);
 366                                 vector<string> names;
 367                                 string saveBinNames = binnames;
 368                                 m->splitAtComma(binnames, names);
 369
 370                                 vector<string> newGroupFile;
 371                                 if (groupfile != "") {
 372                                         vector<string> newNames;
 373                                         saveBinNames = "";
 374                                         for(int k = 0; k < names.size(); k++) {
 375                                                 string group = groupMap->getGroup(names[k]);
 376
 377                                                 if (m->inUsersGroups(group, Groups)) {
 378                                                         newGroupFile.push_back(names[k] + "\t" + group);
 379
 380                                                         newNames.push_back(names[k]);
 381                                                         saveBinNames += names[k] + ",";
 382                                                 }
 383                                         }
 384                                         names = newNames;
 385                                         saveBinNames = saveBinNames.substr(0, saveBinNames.length()-1);
 386                                 }
 387
 388                                 if (names.size() > nseqs) { //keep bin
 389                                         newList.push_back(saveBinNames);
 390                                         for(int k = 0; k < newGroupFile.size(); k++) { outGroup << newGroupFile[k] << endl; }
 391                                 }
 392                         }
 393
 394                         //print new listvector
 395                         if (newList.getNumBins() != 0) {
 396                                 wroteSomething = true;
 397                                 newList.print(out);
 398                         }
 399                 }
 400
 401                 out.close();
 402                 if (groupfile != "") { outGroup.close(); outputTypes["group"].push_back(outputGroupFileName); outputNames.push_back(outputGroupFileName); }
 403
 404                 if (wroteSomething == false) {  m->mothurOut("Your file contains only rare sequences."); m->mothurOutEndLine();  }
 405                 outputTypes["list"].push_back(outputFileName); outputNames.push_back(outputFileName);
 406
 407                 return 0;
 408         }
 409         catch(exception& e) {
 410                 m->errorOut(e, "RemoveRareCommand", "processList");
 411                 exit(1);
 412         }
 413 }
 414 //**********************************************************************************************************************
 415 int RemoveRareCommand::processSabund(){
 416         try {
 417                 string thisOutputDir = outputDir;
 418                 if (outputDir == "") {  thisOutputDir += m->hasPath(sabundfile);  }
 419                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(sabundfile)) + "pick" +  m->getExtension(sabundfile);
 420                 outputTypes["sabund"].push_back(outputFileName); outputNames.push_back(outputFileName);
 421
 422                 ofstream out;
 423                 m->openOutputFile(outputFileName, out);
 424
 425                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 426                 InputData input(sabundfile, "sabund");
 427                 SAbundVector* sabund = input.getSAbundVector();
 428                 string lastLabel = sabund->getLabel();
 429                 set<string> processedLabels;
 430                 set<string> userLabels = labels;
 431
 432                 while((sabund != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 433
 434                         if (m->control_pressed) { delete sabund; out.close(); return 0; }
 435
 436                         if(allLines == 1 || labels.count(sabund->getLabel()) == 1){
 437
 438                                 m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
 439                                 processedLabels.insert(sabund->getLabel());
 440                                 userLabels.erase(sabund->getLabel());
 441
 442                                 if (sabund->getMaxRank() > nseqs) {
 443                                         for(int i = 1; i <=nseqs; i++) {  sabund->set(i, 0); }
 444                                 }else { sabund->clear(); }
 445
 446                                 if (sabund->getNumBins() > 0) { sabund->print(out); }
 447                         }
 448
 449                         if ((m->anyLabelsToProcess(sabund->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 450                                 string saveLabel = sabund->getLabel();
 451
 452                                 delete sabund;
 453                                 sabund = input.getSAbundVector(lastLabel);
 454
 455                                 m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
 456                                 processedLabels.insert(sabund->getLabel());
 457                                 userLabels.erase(sabund->getLabel());
 458
 459                                 if (sabund->getMaxRank() > nseqs) {
 460                                         for(int i = 1; i <=nseqs; i++) {  sabund->set(i, 0); }
 461                                 }else { sabund->clear(); }
 462
 463                                 if (sabund->getNumBins() > 0) { sabund->print(out); }
 464
 465                                 //restore real lastlabel to save below
 466                                 sabund->setLabel(saveLabel);
 467                         }
 468
 469                         lastLabel = sabund->getLabel();
 470
 471                         delete sabund;
 472                         sabund = input.getSAbundVector();
 473                 }
 474
 475                 if (m->control_pressed) {  out.close(); return 0; }
 476
 477                 //output error messages about any remaining user labels
 478                 set<string>::iterator it;
 479                 bool needToRun = false;
 480                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 481                         m->mothurOut("Your file does not include the label " + *it);
 482                         if (processedLabels.count(lastLabel) != 1) {
 483                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 484                                 needToRun = true;
 485                         }else {
 486                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 487                         }
 488                 }
 489
 490                 //run last label if you need to
 491                 if (needToRun == true)  {
 492                         if (sabund != NULL) {   delete sabund;  }
 493                         sabund = input.getSAbundVector(lastLabel);
 494
 495                         m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
 496
 497                         if (sabund->getMaxRank() > nseqs) {
 498                                 for(int i = 1; i <=nseqs; i++) {  sabund->set(i, 0); }
 499                         }else { sabund->clear(); }
 500
 501                         if (sabund->getNumBins() > 0) { sabund->print(out); }
 502
 503                         delete sabund;
 504                 }
 505
 506                 return 0;
 507         }
 508         catch(exception& e) {
 509                 m->errorOut(e, "RemoveRareCommand", "processSabund");
 510                 exit(1);
 511         }
 512 }
 513 //**********************************************************************************************************************
 514 int RemoveRareCommand::processRabund(){
 515         try {
 516                 string thisOutputDir = outputDir;
 517                 if (outputDir == "") {  thisOutputDir += m->hasPath(rabundfile);  }
 518                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(rabundfile)) + "pick" +  m->getExtension(rabundfile);
 519                 outputTypes["rabund"].push_back(outputFileName); outputNames.push_back(outputFileName);
 520
 521                 ofstream out;
 522                 m->openOutputFile(outputFileName, out);
 523
 524                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 525                 InputData input(rabundfile, "rabund");
 526                 RAbundVector* rabund = input.getRAbundVector();
 527                 string lastLabel = rabund->getLabel();
 528                 set<string> processedLabels;
 529                 set<string> userLabels = labels;
 530
 531                 while((rabund != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 532
 533                         if (m->control_pressed) { delete rabund; out.close(); return 0; }
 534
 535                         if(allLines == 1 || labels.count(rabund->getLabel()) == 1){
 536
 537                                 m->mothurOut(rabund->getLabel()); m->mothurOutEndLine();
 538                                 processedLabels.insert(rabund->getLabel());
 539                                 userLabels.erase(rabund->getLabel());
 540
 541                                 RAbundVector newRabund; newRabund.setLabel(rabund->getLabel());
 542                                 for (int i = 0; i < rabund->getNumBins(); i++) {
 543                                         if (rabund->get(i) > nseqs) {
 544                                                 newRabund.push_back(rabund->get(i));
 545                                         }
 546                                 }
 547                                 if (newRabund.getNumBins() > 0) { newRabund.print(out); }
 548                         }
 549
 550                         if ((m->anyLabelsToProcess(rabund->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 551                                 string saveLabel = rabund->getLabel();
 552
 553                                 delete rabund;
 554                                 rabund = input.getRAbundVector(lastLabel);
 555
 556                                 m->mothurOut(rabund->getLabel()); m->mothurOutEndLine();
 557                                 processedLabels.insert(rabund->getLabel());
 558                                 userLabels.erase(rabund->getLabel());
 559
 560                                 RAbundVector newRabund; newRabund.setLabel(rabund->getLabel());
 561                                 for (int i = 0; i < rabund->getNumBins(); i++) {
 562                                         if (rabund->get(i) > nseqs) {
 563                                                 newRabund.push_back(rabund->get(i));
 564                                         }
 565                                 }
 566                                 if (newRabund.getNumBins() > 0) { newRabund.print(out); }
 567
 568                                 //restore real lastlabel to save below
 569                                 rabund->setLabel(saveLabel);
 570                         }
 571
 572                         lastLabel = rabund->getLabel();
 573
 574                         delete rabund;
 575                         rabund = input.getRAbundVector();
 576                 }
 577
 578                 if (m->control_pressed) {  out.close(); return 0; }
 579
 580                 //output error messages about any remaining user labels
 581                 set<string>::iterator it;
 582                 bool needToRun = false;
 583                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 584                         m->mothurOut("Your file does not include the label " + *it);
 585                         if (processedLabels.count(lastLabel) != 1) {
 586                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 587                                 needToRun = true;
 588                         }else {
 589                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 590                         }
 591                 }
 592
 593                 //run last label if you need to
 594                 if (needToRun == true)  {
 595                         if (rabund != NULL) {   delete rabund;  }
 596                         rabund = input.getRAbundVector(lastLabel);
 597
 598                         m->mothurOut(rabund->getLabel()); m->mothurOutEndLine();
 599
 600                         RAbundVector newRabund; newRabund.setLabel(rabund->getLabel());
 601                         for (int i = 0; i < rabund->getNumBins(); i++) {
 602                                 if (rabund->get(i) > nseqs) {
 603                                         newRabund.push_back(rabund->get(i));
 604                                 }
 605                         }
 606                         if (newRabund.getNumBins() > 0) { newRabund.print(out); }
 607
 608                         delete rabund;
 609                 }
 610
 611                 return 0;
 612         }
 613         catch(exception& e) {
 614                 m->errorOut(e, "RemoveRareCommand", "processRabund");
 615                 exit(1);
 616         }
 617 }
 618 //**********************************************************************************************************************
 619 int RemoveRareCommand::processShared(){
 620         try {
 621                 m->Groups = Groups;
 622
 623                 string thisOutputDir = outputDir;
 624                 if (outputDir == "") {  thisOutputDir += m->hasPath(sharedfile);  }
 625                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(sharedfile)) + "pick" +  m->getExtension(sharedfile);
 626                 outputTypes["shared"].push_back(outputFileName); outputNames.push_back(outputFileName);
 627
 628                 ofstream out;
 629                 m->openOutputFile(outputFileName, out);
 630
 631                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 632                 InputData input(sharedfile, "sharedfile");
 633                 vector<SharedRAbundVector*> lookup = input.getSharedRAbundVectors();
 634                 string lastLabel = lookup[0]->getLabel();
 635                 set<string> processedLabels;
 636                 set<string> userLabels = labels;
 637
 638                 while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 639
 640                         if (m->control_pressed) { for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }  out.close(); return 0; }
 641
 642                         if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){
 643
 644                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 645                                 processedLabels.insert(lookup[0]->getLabel());
 646                                 userLabels.erase(lookup[0]->getLabel());
 647
 648                                 processLookup(lookup, out);
 649                         }
 650
 651                         if ((m->anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 652                                 string saveLabel = lookup[0]->getLabel();
 653
 654                                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 655                                 lookup = input.getSharedRAbundVectors(lastLabel);
 656
 657                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 658                                 processedLabels.insert(lookup[0]->getLabel());
 659                                 userLabels.erase(lookup[0]->getLabel());
 660
 661                                 processLookup(lookup, out);
 662
 663                                 //restore real lastlabel to save below
 664                                 lookup[0]->setLabel(saveLabel);
 665                         }
 666
 667                         lastLabel = lookup[0]->getLabel();
 668
 669                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 670                         lookup = input.getSharedRAbundVectors();
 671                 }
 672
 673                 if (m->control_pressed) {  out.close(); return 0; }
 674
 675                 //output error messages about any remaining user labels
 676                 set<string>::iterator it;
 677                 bool needToRun = false;
 678                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 679                         m->mothurOut("Your file does not include the label " + *it);
 680                         if (processedLabels.count(lastLabel) != 1) {
 681                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 682                                 needToRun = true;
 683                         }else {
 684                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 685                         }
 686                 }
 687
 688                 //run last label if you need to
 689                 if (needToRun == true)  {
 690                         for (int i = 0; i < lookup.size(); i++) {  if (lookup[i] != NULL) {     delete lookup[i];       }  }
 691                         lookup = input.getSharedRAbundVectors(lastLabel);
 692
 693                         m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 694
 695                         processLookup(lookup, out);
 696
 697                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 698                 }
 699
 700                 return 0;
 701         }
 702         catch(exception& e) {
 703                 m->errorOut(e, "RemoveRareCommand", "processSabund");
 704                 exit(1);
 705         }
 706 }
 707 //**********************************************************************************************************************
 708 int RemoveRareCommand::processLookup(vector<SharedRAbundVector*>& lookup, ofstream& out){
 709         try {
 710
 711                 vector<SharedRAbundVector> newRabunds;  newRabunds.resize(lookup.size());
 712                 for (int i = 0; i < lookup.size(); i++) {
 713                         newRabunds[i].setGroup(lookup[i]->getGroup());
 714                         newRabunds[i].setLabel(lookup[i]->getLabel());
 715                 }
 716
 717                 if (byGroup) {
 718
 719                         //for each otu
 720                         for (int i = 0; i < lookup[0]->getNumBins(); i++) {
 721                                 bool allZero = true;
 722
 723                                 if (m->control_pressed) { return 0; }
 724
 725                                 //for each group
 726                                 for (int j = 0; j < lookup.size(); j++) {
 727
 728                                         //are you rare?
 729                                         if (lookup[j]->getAbundance(i) > nseqs) {
 730                                                 newRabunds[j].push_back(lookup[j]->getAbundance(i), newRabunds[j].getGroup());
 731                                                 allZero = false;
 732                                         }else {
 733                                                 newRabunds[j].push_back(0, newRabunds[j].getGroup());
 734                                         }
 735                                 }
 736
 737                                 //eliminates zero otus
 738                                 if (allZero) { for (int j = 0; j < newRabunds.size(); j++) {  newRabunds[j].pop_back(); } }
 739                         }
 740                 }else {
 741                         //for each otu
 742                         for (int i = 0; i < lookup[0]->getNumBins(); i++) {
 743
 744                                 if (m->control_pressed) { return 0; }
 745
 746                                 int totalAbund = 0;
 747                                 //get total otu abundance
 748                                 for (int j = 0; j < lookup.size(); j++) {
 749                                         newRabunds[j].push_back(lookup[j]->getAbundance(i), newRabunds[j].getGroup());
 750                                         totalAbund += lookup[j]->getAbundance(i);
 751                                 }
 752
 753                                 //eliminates otus below rare cutoff
 754                                 if (totalAbund <= nseqs) { for (int j = 0; j < newRabunds.size(); j++) {  newRabunds[j].pop_back(); } }
 755                         }
 756                 }
 757
 758                 //do we have any otus above the rare cutoff
 759                 if (newRabunds[0].getNumBins() != 0) {
 760                         for (int j = 0; j < newRabunds.size(); j++) {
 761                                 out << newRabunds[j].getLabel() << '\t' << newRabunds[j].getGroup() << '\t';
 762                                 newRabunds[j].print(out);
 763                         }
 764                 }
 765
 766                 return 0;
 767         }
 768         catch(exception& e) {
 769                 m->errorOut(e, "RemoveRareCommand", "processLookup");
 770                 exit(1);
 771         }
 772 }
 773 //**********************************************************************************************************************
 774
 775
 776
 777