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                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  87
  88                 else {
  89                         vector<string> myArray = setParameters();
  90
  91                         OptionParser parser(option);
  92                         map<string,string> parameters = parser.getParameters();
  93
  94                         ValidParameters validParameter;
  95                         map<string,string>::iterator it;
  96
  97                         //check to make sure all parameters are valid for command
  98                         for (it = parameters.begin(); it != parameters.end(); it++) {
  99                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 100                         }
 101
 102                         //initialize outputTypes
 103                         vector<string> tempOutNames;
 104                         outputTypes["rabund"] = tempOutNames;
 105                         outputTypes["sabund"] = tempOutNames;
 106                         outputTypes["list"] = tempOutNames;
 107                         outputTypes["group"] = tempOutNames;
 108                         outputTypes["shared"] = tempOutNames;
 109
 110                         //if the user changes the output directory command factory will send this info to us in the output parameter
 111                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 112
 113                         //if the user changes the input directory command factory will send this info to us in the output parameter
 114                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 115                         if (inputDir == "not found"){   inputDir = "";          }
 116                         else {
 117                                 string path;
 118                                 it = parameters.find("list");
 119                                 //user has given a template file
 120                                 if(it != parameters.end()){
 121                                         path = m->hasPath(it->second);
 122                                         //if the user has not given a path then, add inputdir. else leave path alone.
 123                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 124                                 }
 125
 126                                 it = parameters.find("group");
 127                                 //user has given a template file
 128                                 if(it != parameters.end()){
 129                                         path = m->hasPath(it->second);
 130                                         //if the user has not given a path then, add inputdir. else leave path alone.
 131                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 132                                 }
 133
 134                                 it = parameters.find("sabund");
 135                                 //user has given a template file
 136                                 if(it != parameters.end()){
 137                                         path = m->hasPath(it->second);
 138                                         //if the user has not given a path then, add inputdir. else leave path alone.
 139                                         if (path == "") {       parameters["sabund"] = inputDir + it->second;           }
 140                                 }
 141
 142                                 it = parameters.find("rabund");
 143                                 //user has given a template file
 144                                 if(it != parameters.end()){
 145                                         path = m->hasPath(it->second);
 146                                         //if the user has not given a path then, add inputdir. else leave path alone.
 147                                         if (path == "") {       parameters["rabund"] = inputDir + it->second;           }
 148                                 }
 149
 150                                 it = parameters.find("shared");
 151                                 //user has given a template file
 152                                 if(it != parameters.end()){
 153                                         path = m->hasPath(it->second);
 154                                         //if the user has not given a path then, add inputdir. else leave path alone.
 155                                         if (path == "") {       parameters["shared"] = inputDir + it->second;           }
 156                                 }
 157                         }
 158
 159
 160                         //check for file parameters
 161                         listfile = validParameter.validFile(parameters, "list", true);
 162                         if (listfile == "not open") { abort = true; }
 163                         else if (listfile == "not found") {  listfile = "";  }
 164                         else { m->setListFile(listfile); }
 165
 166                         sabundfile = validParameter.validFile(parameters, "sabund", true);
 167                         if (sabundfile == "not open") { abort = true; }
 168                         else if (sabundfile == "not found") {  sabundfile = "";  }
 169                         else { m->setSabundFile(sabundfile); }
 170
 171                         rabundfile = validParameter.validFile(parameters, "rabund", true);
 172                         if (rabundfile == "not open") { abort = true; }
 173                         else if (rabundfile == "not found") {  rabundfile = "";  }
 174                         else { m->setRabundFile(rabundfile); }
 175
 176                         groupfile = validParameter.validFile(parameters, "group", true);
 177                         if (groupfile == "not open") { groupfile = ""; abort = true; }
 178                         else if (groupfile == "not found") {  groupfile = "";  }
 179                         else { m->setGroupFile(groupfile); }
 180
 181                         sharedfile = validParameter.validFile(parameters, "shared", true);
 182                         if (sharedfile == "not open") { sharedfile = "";  abort = true; }
 183                         else if (sharedfile == "not found") {  sharedfile = "";  }
 184                         else { m->setSharedFile(sharedfile); }
 185
 186                         if ((sharedfile == "") && (listfile == "") && (rabundfile == "") && (sabundfile == "")) {
 187                                 //is there are current file available for any of these?
 188                                 //give priority to shared, then list, then rabund, then sabund
 189                                 //if there is a current shared file, use it
 190                                 sharedfile = m->getSharedFile();
 191                                 if (sharedfile != "") {  m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); }
 192                                 else {
 193                                         listfile = m->getListFile();
 194                                         if (listfile != "") {  m->mothurOut("Using " + listfile + " as input file for the list parameter."); m->mothurOutEndLine(); }
 195                                         else {
 196                                                 rabundfile = m->getRabundFile();
 197                                                 if (rabundfile != "") {  m->mothurOut("Using " + rabundfile + " as input file for the rabund parameter."); m->mothurOutEndLine(); }
 198                                                 else {
 199                                                         sabundfile = m->getSabundFile();
 200                                                         if (sabundfile != "") {  m->mothurOut("Using " + sabundfile + " as input file for the sabund parameter."); m->mothurOutEndLine(); }
 201                                                         else {
 202                                                                 m->mothurOut("No valid current files. You must provide a list, sabund, rabund or shared file."); m->mothurOutEndLine();
 203                                                                 abort = true;
 204                                                         }
 205                                                 }
 206                                         }
 207                                 }
 208                         }
 209
 210                         groups = validParameter.validFile(parameters, "groups", false);
 211                         if (groups == "not found") { groups = "all"; }
 212                         m->splitAtDash(groups, Groups);
 213
 214                         label = validParameter.validFile(parameters, "label", false);
 215                         if (label == "not found") { label = ""; }
 216                         else {
 217                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 218                                 else { allLines = 1;  }
 219                         }
 220
 221                         string temp = validParameter.validFile(parameters, "nseqs", false);
 222                         if (temp == "not found") { m->mothurOut("nseqs is a required parameter."); m->mothurOutEndLine(); abort = true; }
 223                         else { convert(temp, nseqs); }
 224
 225                         temp = validParameter.validFile(parameters, "bygroup", false);   if (temp == "not found") { temp = "f"; }
 226                         byGroup = m->isTrue(temp);
 227
 228                         if (byGroup && (sharedfile == "")) { m->mothurOut("The byGroup parameter is only valid with a shared file."); m->mothurOutEndLine(); }
 229
 230                         if ((groupfile != "") && (listfile == "")) { m->mothurOut("A groupfile is only valid with a list file."); m->mothurOutEndLine(); groupfile = ""; }
 231                 }
 232
 233         }
 234         catch(exception& e) {
 235                 m->errorOut(e, "RemoveRareCommand", "RemoveRareCommand");
 236                 exit(1);
 237         }
 238 }
 239 //**********************************************************************************************************************
 240
 241 int RemoveRareCommand::execute(){
 242         try {
 243
 244                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 245
 246                 if (m->control_pressed) { return 0; }
 247
 248                 //read through the correct file and output lines you want to keep
 249                 if (sabundfile != "")           {               processSabund();        }
 250                 if (rabundfile != "")           {               processRabund();        }
 251                 if (listfile != "")                     {               processList();          }
 252                 if (sharedfile != "")           {               processShared();        }
 253
 254                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str()); } return 0; }
 255
 256                 if (outputNames.size() != 0) {
 257                         m->mothurOutEndLine();
 258                         m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 259                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 260                         m->mothurOutEndLine();
 261
 262                         //set rabund file as new current rabundfile
 263                         string current = "";
 264                         itTypes = outputTypes.find("rabund");
 265                         if (itTypes != outputTypes.end()) {
 266                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setRabundFile(current); }
 267                         }
 268
 269                         itTypes = outputTypes.find("sabund");
 270                         if (itTypes != outputTypes.end()) {
 271                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setSabundFile(current); }
 272                         }
 273
 274                         itTypes = outputTypes.find("group");
 275                         if (itTypes != outputTypes.end()) {
 276                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setGroupFile(current); }
 277                         }
 278
 279                         itTypes = outputTypes.find("list");
 280                         if (itTypes != outputTypes.end()) {
 281                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 282                         }
 283
 284                         itTypes = outputTypes.find("shared");
 285                         if (itTypes != outputTypes.end()) {
 286                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setSharedFile(current); }
 287                         }
 288                 }
 289
 290                 return 0;
 291         }
 292
 293         catch(exception& e) {
 294                 m->errorOut(e, "RemoveRareCommand", "execute");
 295                 exit(1);
 296         }
 297 }
 298
 299 //**********************************************************************************************************************
 300 int RemoveRareCommand::processList(){
 301         try {
 302                 string thisOutputDir = outputDir;
 303                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 304                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "pick" +  m->getExtension(listfile);
 305                 string outputGroupFileName = thisOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "pick" +  m->getExtension(groupfile);
 306
 307                 ofstream out, outGroup;
 308                 m->openOutputFile(outputFileName, out);
 309
 310                 bool wroteSomething = false;
 311
 312                 //you must provide a label because the names in the listfile need to be consistent
 313                 string thisLabel = "";
 314                 if (allLines) { m->mothurOut("For the listfile you must select one label, using first label in your listfile."); m->mothurOutEndLine(); }
 315                 else if (labels.size() > 1) { m->mothurOut("For the listfile you must select one label, using " + (*labels.begin()) + "."); m->mothurOutEndLine(); thisLabel = *labels.begin(); }
 316                 else { thisLabel = *labels.begin(); }
 317
 318                 InputData input(listfile, "list");
 319                 ListVector* list = input.getListVector();
 320
 321                 //get first one or the one we want
 322                 if (thisLabel != "") {
 323                         //use smart distancing
 324                         set<string> userLabels; userLabels.insert(thisLabel);
 325                         set<string> processedLabels;
 326                         string lastLabel = list->getLabel();
 327                         while((list != NULL) && (userLabels.size() != 0)) {
 328                                 if(userLabels.count(list->getLabel()) == 1){
 329                                         processedLabels.insert(list->getLabel());
 330                                         userLabels.erase(list->getLabel());
 331                                         break;
 332                                 }
 333
 334                                 if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 335                                         processedLabels.insert(list->getLabel());
 336                                         userLabels.erase(list->getLabel());
 337                                         delete list;
 338                                         list = input.getListVector(lastLabel);
 339                                         break;
 340                                 }
 341                                 lastLabel = list->getLabel();
 342                                 delete list;
 343                                 list = input.getListVector();
 344                         }
 345                         if (userLabels.size() != 0) {
 346                                 m->mothurOut("Your file does not include the label " + thisLabel + ". I will use " + lastLabel + ".");  m->mothurOutEndLine();
 347                                 list = input.getListVector(lastLabel);
 348                         }
 349                 }
 350
 351                 //if groupfile is given then use it
 352                 GroupMap* groupMap;
 353                 if (groupfile != "") {
 354                         groupMap = new GroupMap(groupfile); groupMap->readMap();
 355                         SharedUtil util;
 356                         util.setGroups(Groups, groupMap->namesOfGroups);
 357                         m->openOutputFile(outputGroupFileName, outGroup);
 358                 }
 359
 360
 361                 if (list != NULL) {
 362                         //make a new list vector
 363                         ListVector newList;
 364                         newList.setLabel(list->getLabel());
 365
 366                         //for each bin
 367                         for (int i = 0; i < list->getNumBins(); i++) {
 368                                 if (m->control_pressed) {  if (groupfile != "") { delete groupMap; outGroup.close(); remove(outputGroupFileName.c_str()); } out.close();  remove(outputFileName.c_str());  return 0; }
 369
 370                                 //parse out names that are in accnos file
 371                                 string binnames = list->get(i);
 372                                 vector<string> names;
 373                                 string saveBinNames = binnames;
 374                                 m->splitAtComma(binnames, names);
 375
 376                                 vector<string> newGroupFile;
 377                                 if (groupfile != "") {
 378                                         vector<string> newNames;
 379                                         saveBinNames = "";
 380                                         for(int k = 0; k < names.size(); k++) {
 381                                                 string group = groupMap->getGroup(names[k]);
 382
 383                                                 if (m->inUsersGroups(group, Groups)) {
 384                                                         newGroupFile.push_back(names[k] + "\t" + group);
 385
 386                                                         newNames.push_back(names[k]);
 387                                                         saveBinNames += names[k] + ",";
 388                                                 }
 389                                         }
 390                                         names = newNames;
 391                                         saveBinNames = saveBinNames.substr(0, saveBinNames.length()-1);
 392                                 }
 393
 394                                 if (names.size() > nseqs) { //keep bin
 395                                         newList.push_back(saveBinNames);
 396                                         for(int k = 0; k < newGroupFile.size(); k++) { outGroup << newGroupFile[k] << endl; }
 397                                 }
 398                         }
 399
 400                         //print new listvector
 401                         if (newList.getNumBins() != 0) {
 402                                 wroteSomething = true;
 403                                 newList.print(out);
 404                         }
 405                 }
 406
 407                 out.close();
 408                 if (groupfile != "") { outGroup.close(); outputTypes["group"].push_back(outputGroupFileName); outputNames.push_back(outputGroupFileName); }
 409
 410                 if (wroteSomething == false) {  m->mothurOut("Your file contains only rare sequences."); m->mothurOutEndLine();  }
 411                 outputTypes["list"].push_back(outputFileName); outputNames.push_back(outputFileName);
 412
 413                 return 0;
 414         }
 415         catch(exception& e) {
 416                 m->errorOut(e, "RemoveRareCommand", "processList");
 417                 exit(1);
 418         }
 419 }
 420 //**********************************************************************************************************************
 421 int RemoveRareCommand::processSabund(){
 422         try {
 423                 string thisOutputDir = outputDir;
 424                 if (outputDir == "") {  thisOutputDir += m->hasPath(sabundfile);  }
 425                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(sabundfile)) + "pick" +  m->getExtension(sabundfile);
 426                 outputTypes["sabund"].push_back(outputFileName); outputNames.push_back(outputFileName);
 427
 428                 ofstream out;
 429                 m->openOutputFile(outputFileName, out);
 430
 431                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 432                 InputData input(sabundfile, "sabund");
 433                 SAbundVector* sabund = input.getSAbundVector();
 434                 string lastLabel = sabund->getLabel();
 435                 set<string> processedLabels;
 436                 set<string> userLabels = labels;
 437
 438                 while((sabund != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 439
 440                         if (m->control_pressed) { delete sabund; out.close(); return 0; }
 441
 442                         if(allLines == 1 || labels.count(sabund->getLabel()) == 1){
 443
 444                                 m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
 445                                 processedLabels.insert(sabund->getLabel());
 446                                 userLabels.erase(sabund->getLabel());
 447
 448                                 if (sabund->getMaxRank() > nseqs) {
 449                                         for(int i = 1; i <=nseqs; i++) {  sabund->set(i, 0); }
 450                                 }else { sabund->clear(); }
 451
 452                                 if (sabund->getNumBins() > 0) { sabund->print(out); }
 453                         }
 454
 455                         if ((m->anyLabelsToProcess(sabund->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 456                                 string saveLabel = sabund->getLabel();
 457
 458                                 delete sabund;
 459                                 sabund = input.getSAbundVector(lastLabel);
 460
 461                                 m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
 462                                 processedLabels.insert(sabund->getLabel());
 463                                 userLabels.erase(sabund->getLabel());
 464
 465                                 if (sabund->getMaxRank() > nseqs) {
 466                                         for(int i = 1; i <=nseqs; i++) {  sabund->set(i, 0); }
 467                                 }else { sabund->clear(); }
 468
 469                                 if (sabund->getNumBins() > 0) { sabund->print(out); }
 470
 471                                 //restore real lastlabel to save below
 472                                 sabund->setLabel(saveLabel);
 473                         }
 474
 475                         lastLabel = sabund->getLabel();
 476
 477                         delete sabund;
 478                         sabund = input.getSAbundVector();
 479                 }
 480
 481                 if (m->control_pressed) {  out.close(); return 0; }
 482
 483                 //output error messages about any remaining user labels
 484                 set<string>::iterator it;
 485                 bool needToRun = false;
 486                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 487                         m->mothurOut("Your file does not include the label " + *it);
 488                         if (processedLabels.count(lastLabel) != 1) {
 489                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 490                                 needToRun = true;
 491                         }else {
 492                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 493                         }
 494                 }
 495
 496                 //run last label if you need to
 497                 if (needToRun == true)  {
 498                         if (sabund != NULL) {   delete sabund;  }
 499                         sabund = input.getSAbundVector(lastLabel);
 500
 501                         m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
 502
 503                         if (sabund->getMaxRank() > nseqs) {
 504                                 for(int i = 1; i <=nseqs; i++) {  sabund->set(i, 0); }
 505                         }else { sabund->clear(); }
 506
 507                         if (sabund->getNumBins() > 0) { sabund->print(out); }
 508
 509                         delete sabund;
 510                 }
 511
 512                 return 0;
 513         }
 514         catch(exception& e) {
 515                 m->errorOut(e, "RemoveRareCommand", "processSabund");
 516                 exit(1);
 517         }
 518 }
 519 //**********************************************************************************************************************
 520 int RemoveRareCommand::processRabund(){
 521         try {
 522                 string thisOutputDir = outputDir;
 523                 if (outputDir == "") {  thisOutputDir += m->hasPath(rabundfile);  }
 524                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(rabundfile)) + "pick" +  m->getExtension(rabundfile);
 525                 outputTypes["rabund"].push_back(outputFileName); outputNames.push_back(outputFileName);
 526
 527                 ofstream out;
 528                 m->openOutputFile(outputFileName, out);
 529
 530                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 531                 InputData input(rabundfile, "rabund");
 532                 RAbundVector* rabund = input.getRAbundVector();
 533                 string lastLabel = rabund->getLabel();
 534                 set<string> processedLabels;
 535                 set<string> userLabels = labels;
 536
 537                 while((rabund != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 538
 539                         if (m->control_pressed) { delete rabund; out.close(); return 0; }
 540
 541                         if(allLines == 1 || labels.count(rabund->getLabel()) == 1){
 542
 543                                 m->mothurOut(rabund->getLabel()); m->mothurOutEndLine();
 544                                 processedLabels.insert(rabund->getLabel());
 545                                 userLabels.erase(rabund->getLabel());
 546
 547                                 RAbundVector newRabund; newRabund.setLabel(rabund->getLabel());
 548                                 for (int i = 0; i < rabund->getNumBins(); i++) {
 549                                         if (rabund->get(i) > nseqs) {
 550                                                 newRabund.push_back(rabund->get(i));
 551                                         }
 552                                 }
 553                                 if (newRabund.getNumBins() > 0) { newRabund.print(out); }
 554                         }
 555
 556                         if ((m->anyLabelsToProcess(rabund->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 557                                 string saveLabel = rabund->getLabel();
 558
 559                                 delete rabund;
 560                                 rabund = input.getRAbundVector(lastLabel);
 561
 562                                 m->mothurOut(rabund->getLabel()); m->mothurOutEndLine();
 563                                 processedLabels.insert(rabund->getLabel());
 564                                 userLabels.erase(rabund->getLabel());
 565
 566                                 RAbundVector newRabund; newRabund.setLabel(rabund->getLabel());
 567                                 for (int i = 0; i < rabund->getNumBins(); i++) {
 568                                         if (rabund->get(i) > nseqs) {
 569                                                 newRabund.push_back(rabund->get(i));
 570                                         }
 571                                 }
 572                                 if (newRabund.getNumBins() > 0) { newRabund.print(out); }
 573
 574                                 //restore real lastlabel to save below
 575                                 rabund->setLabel(saveLabel);
 576                         }
 577
 578                         lastLabel = rabund->getLabel();
 579
 580                         delete rabund;
 581                         rabund = input.getRAbundVector();
 582                 }
 583
 584                 if (m->control_pressed) {  out.close(); return 0; }
 585
 586                 //output error messages about any remaining user labels
 587                 set<string>::iterator it;
 588                 bool needToRun = false;
 589                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 590                         m->mothurOut("Your file does not include the label " + *it);
 591                         if (processedLabels.count(lastLabel) != 1) {
 592                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 593                                 needToRun = true;
 594                         }else {
 595                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 596                         }
 597                 }
 598
 599                 //run last label if you need to
 600                 if (needToRun == true)  {
 601                         if (rabund != NULL) {   delete rabund;  }
 602                         rabund = input.getRAbundVector(lastLabel);
 603
 604                         m->mothurOut(rabund->getLabel()); m->mothurOutEndLine();
 605
 606                         RAbundVector newRabund; newRabund.setLabel(rabund->getLabel());
 607                         for (int i = 0; i < rabund->getNumBins(); i++) {
 608                                 if (rabund->get(i) > nseqs) {
 609                                         newRabund.push_back(rabund->get(i));
 610                                 }
 611                         }
 612                         if (newRabund.getNumBins() > 0) { newRabund.print(out); }
 613
 614                         delete rabund;
 615                 }
 616
 617                 return 0;
 618         }
 619         catch(exception& e) {
 620                 m->errorOut(e, "RemoveRareCommand", "processRabund");
 621                 exit(1);
 622         }
 623 }
 624 //**********************************************************************************************************************
 625 int RemoveRareCommand::processShared(){
 626         try {
 627                 m->Groups = Groups;
 628
 629                 string thisOutputDir = outputDir;
 630                 if (outputDir == "") {  thisOutputDir += m->hasPath(sharedfile);  }
 631                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(sharedfile)) + "pick" +  m->getExtension(sharedfile);
 632                 outputTypes["shared"].push_back(outputFileName); outputNames.push_back(outputFileName);
 633
 634                 ofstream out;
 635                 m->openOutputFile(outputFileName, out);
 636
 637                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 638                 InputData input(sharedfile, "sharedfile");
 639                 vector<SharedRAbundVector*> lookup = input.getSharedRAbundVectors();
 640                 string lastLabel = lookup[0]->getLabel();
 641                 set<string> processedLabels;
 642                 set<string> userLabels = labels;
 643
 644                 while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 645
 646                         if (m->control_pressed) { for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }  out.close(); return 0; }
 647
 648                         if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){
 649
 650                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 651                                 processedLabels.insert(lookup[0]->getLabel());
 652                                 userLabels.erase(lookup[0]->getLabel());
 653
 654                                 if (!m->printedHeaders) { lookup[0]->printHeaders(out); }
 655                                 processLookup(lookup, out);
 656                         }
 657
 658                         if ((m->anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 659                                 string saveLabel = lookup[0]->getLabel();
 660
 661                                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 662                                 lookup = input.getSharedRAbundVectors(lastLabel);
 663
 664                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 665                                 processedLabels.insert(lookup[0]->getLabel());
 666                                 userLabels.erase(lookup[0]->getLabel());
 667
 668                                 if (!m->printedHeaders) { lookup[0]->printHeaders(out); }
 669                                 processLookup(lookup, out);
 670
 671                                 //restore real lastlabel to save below
 672                                 lookup[0]->setLabel(saveLabel);
 673                         }
 674
 675                         lastLabel = lookup[0]->getLabel();
 676
 677                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 678                         lookup = input.getSharedRAbundVectors();
 679                 }
 680
 681                 if (m->control_pressed) {  out.close(); return 0; }
 682
 683                 //output error messages about any remaining user labels
 684                 set<string>::iterator it;
 685                 bool needToRun = false;
 686                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 687                         m->mothurOut("Your file does not include the label " + *it);
 688                         if (processedLabels.count(lastLabel) != 1) {
 689                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 690                                 needToRun = true;
 691                         }else {
 692                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 693                         }
 694                 }
 695
 696                 //run last label if you need to
 697                 if (needToRun == true)  {
 698                         for (int i = 0; i < lookup.size(); i++) {  if (lookup[i] != NULL) {     delete lookup[i];       }  }
 699                         lookup = input.getSharedRAbundVectors(lastLabel);
 700
 701                         m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 702
 703                         if (!m->printedHeaders) { lookup[0]->printHeaders(out); }
 704                         processLookup(lookup, out);
 705
 706                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 707                 }
 708
 709                 return 0;
 710         }
 711         catch(exception& e) {
 712                 m->errorOut(e, "RemoveRareCommand", "processSabund");
 713                 exit(1);
 714         }
 715 }
 716 //**********************************************************************************************************************
 717 int RemoveRareCommand::processLookup(vector<SharedRAbundVector*>& lookup, ofstream& out){
 718         try {
 719
 720                 vector<SharedRAbundVector> newRabunds;  newRabunds.resize(lookup.size());
 721                 for (int i = 0; i < lookup.size(); i++) {
 722                         newRabunds[i].setGroup(lookup[i]->getGroup());
 723                         newRabunds[i].setLabel(lookup[i]->getLabel());
 724                 }
 725
 726                 if (byGroup) {
 727
 728                         //for each otu
 729                         for (int i = 0; i < lookup[0]->getNumBins(); i++) {
 730                                 bool allZero = true;
 731
 732                                 if (m->control_pressed) { return 0; }
 733
 734                                 //for each group
 735                                 for (int j = 0; j < lookup.size(); j++) {
 736
 737                                         //are you rare?
 738                                         if (lookup[j]->getAbundance(i) > nseqs) {
 739                                                 newRabunds[j].push_back(lookup[j]->getAbundance(i), newRabunds[j].getGroup());
 740                                                 allZero = false;
 741                                         }else {
 742                                                 newRabunds[j].push_back(0, newRabunds[j].getGroup());
 743                                         }
 744                                 }
 745
 746                                 //eliminates zero otus
 747                                 if (allZero) { for (int j = 0; j < newRabunds.size(); j++) {  newRabunds[j].pop_back(); } }
 748                         }
 749                 }else {
 750                         //for each otu
 751                         for (int i = 0; i < lookup[0]->getNumBins(); i++) {
 752
 753                                 if (m->control_pressed) { return 0; }
 754
 755                                 int totalAbund = 0;
 756                                 //get total otu abundance
 757                                 for (int j = 0; j < lookup.size(); j++) {
 758                                         newRabunds[j].push_back(lookup[j]->getAbundance(i), newRabunds[j].getGroup());
 759                                         totalAbund += lookup[j]->getAbundance(i);
 760                                 }
 761
 762                                 //eliminates otus below rare cutoff
 763                                 if (totalAbund <= nseqs) { for (int j = 0; j < newRabunds.size(); j++) {  newRabunds[j].pop_back(); } }
 764                         }
 765                 }
 766
 767                 //do we have any otus above the rare cutoff
 768                 if (newRabunds[0].getNumBins() != 0) {
 769                         for (int j = 0; j < newRabunds.size(); j++) {
 770                                 out << newRabunds[j].getLabel() << '\t' << newRabunds[j].getGroup() << '\t';
 771                                 newRabunds[j].print(out);
 772                         }
 773                 }
 774
 775                 return 0;
 776         }
 777         catch(exception& e) {
 778                 m->errorOut(e, "RemoveRareCommand", "processLookup");
 779                 exit(1);
 780         }
 781 }
 782 //**********************************************************************************************************************
 783
 784
 785
 786