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 { m->mothurConvert(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++) {        m->mothurRemove(outputNames[i]); } 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                         vector<string> namesGroups = groupMap->getNamesOfGroups();
 357                         util.setGroups(Groups, namesGroups);
 358                         m->openOutputFile(outputGroupFileName, outGroup);
 359                 }
 360
 361
 362                 if (list != NULL) {
 363                         //make a new list vector
 364                         ListVector newList;
 365                         newList.setLabel(list->getLabel());
 366
 367                         //for each bin
 368                         for (int i = 0; i < list->getNumBins(); i++) {
 369                                 if (m->control_pressed) {  if (groupfile != "") { delete groupMap; outGroup.close(); m->mothurRemove(outputGroupFileName); } out.close();  m->mothurRemove(outputFileName);  return 0; }
 370
 371                                 //parse out names that are in accnos file
 372                                 string binnames = list->get(i);
 373                                 vector<string> names;
 374                                 string saveBinNames = binnames;
 375                                 m->splitAtComma(binnames, names);
 376
 377                                 vector<string> newGroupFile;
 378                                 if (groupfile != "") {
 379                                         vector<string> newNames;
 380                                         saveBinNames = "";
 381                                         for(int k = 0; k < names.size(); k++) {
 382                                                 string group = groupMap->getGroup(names[k]);
 383
 384                                                 if (m->inUsersGroups(group, Groups)) {
 385                                                         newGroupFile.push_back(names[k] + "\t" + group);
 386
 387                                                         newNames.push_back(names[k]);
 388                                                         saveBinNames += names[k] + ",";
 389                                                 }
 390                                         }
 391                                         names = newNames;
 392                                         saveBinNames = saveBinNames.substr(0, saveBinNames.length()-1);
 393                                 }
 394
 395                                 if (names.size() > nseqs) { //keep bin
 396                                         newList.push_back(saveBinNames);
 397                                         for(int k = 0; k < newGroupFile.size(); k++) { outGroup << newGroupFile[k] << endl; }
 398                                 }
 399                         }
 400
 401                         //print new listvector
 402                         if (newList.getNumBins() != 0) {
 403                                 wroteSomething = true;
 404                                 newList.print(out);
 405                         }
 406                 }
 407
 408                 out.close();
 409                 if (groupfile != "") { outGroup.close(); outputTypes["group"].push_back(outputGroupFileName); outputNames.push_back(outputGroupFileName); }
 410
 411                 if (wroteSomething == false) {  m->mothurOut("Your file contains only rare sequences."); m->mothurOutEndLine();  }
 412                 outputTypes["list"].push_back(outputFileName); outputNames.push_back(outputFileName);
 413
 414                 return 0;
 415         }
 416         catch(exception& e) {
 417                 m->errorOut(e, "RemoveRareCommand", "processList");
 418                 exit(1);
 419         }
 420 }
 421 //**********************************************************************************************************************
 422 int RemoveRareCommand::processSabund(){
 423         try {
 424                 string thisOutputDir = outputDir;
 425                 if (outputDir == "") {  thisOutputDir += m->hasPath(sabundfile);  }
 426                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(sabundfile)) + "pick" +  m->getExtension(sabundfile);
 427                 outputTypes["sabund"].push_back(outputFileName); outputNames.push_back(outputFileName);
 428
 429                 ofstream out;
 430                 m->openOutputFile(outputFileName, out);
 431
 432                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 433                 InputData input(sabundfile, "sabund");
 434                 SAbundVector* sabund = input.getSAbundVector();
 435                 string lastLabel = sabund->getLabel();
 436                 set<string> processedLabels;
 437                 set<string> userLabels = labels;
 438
 439                 while((sabund != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 440
 441                         if (m->control_pressed) { delete sabund; out.close(); return 0; }
 442
 443                         if(allLines == 1 || labels.count(sabund->getLabel()) == 1){
 444
 445                                 m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
 446                                 processedLabels.insert(sabund->getLabel());
 447                                 userLabels.erase(sabund->getLabel());
 448
 449                                 if (sabund->getMaxRank() > nseqs) {
 450                                         for(int i = 1; i <=nseqs; i++) {  sabund->set(i, 0); }
 451                                 }else { sabund->clear(); }
 452
 453                                 if (sabund->getNumBins() > 0) { sabund->print(out); }
 454                         }
 455
 456                         if ((m->anyLabelsToProcess(sabund->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 457                                 string saveLabel = sabund->getLabel();
 458
 459                                 delete sabund;
 460                                 sabund = input.getSAbundVector(lastLabel);
 461
 462                                 m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
 463                                 processedLabels.insert(sabund->getLabel());
 464                                 userLabels.erase(sabund->getLabel());
 465
 466                                 if (sabund->getMaxRank() > nseqs) {
 467                                         for(int i = 1; i <=nseqs; i++) {  sabund->set(i, 0); }
 468                                 }else { sabund->clear(); }
 469
 470                                 if (sabund->getNumBins() > 0) { sabund->print(out); }
 471
 472                                 //restore real lastlabel to save below
 473                                 sabund->setLabel(saveLabel);
 474                         }
 475
 476                         lastLabel = sabund->getLabel();
 477
 478                         delete sabund;
 479                         sabund = input.getSAbundVector();
 480                 }
 481
 482                 if (m->control_pressed) {  out.close(); return 0; }
 483
 484                 //output error messages about any remaining user labels
 485                 set<string>::iterator it;
 486                 bool needToRun = false;
 487                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 488                         m->mothurOut("Your file does not include the label " + *it);
 489                         if (processedLabels.count(lastLabel) != 1) {
 490                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 491                                 needToRun = true;
 492                         }else {
 493                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 494                         }
 495                 }
 496
 497                 //run last label if you need to
 498                 if (needToRun == true)  {
 499                         if (sabund != NULL) {   delete sabund;  }
 500                         sabund = input.getSAbundVector(lastLabel);
 501
 502                         m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
 503
 504                         if (sabund->getMaxRank() > nseqs) {
 505                                 for(int i = 1; i <=nseqs; i++) {  sabund->set(i, 0); }
 506                         }else { sabund->clear(); }
 507
 508                         if (sabund->getNumBins() > 0) { sabund->print(out); }
 509
 510                         delete sabund;
 511                 }
 512
 513                 return 0;
 514         }
 515         catch(exception& e) {
 516                 m->errorOut(e, "RemoveRareCommand", "processSabund");
 517                 exit(1);
 518         }
 519 }
 520 //**********************************************************************************************************************
 521 int RemoveRareCommand::processRabund(){
 522         try {
 523                 string thisOutputDir = outputDir;
 524                 if (outputDir == "") {  thisOutputDir += m->hasPath(rabundfile);  }
 525                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(rabundfile)) + "pick" +  m->getExtension(rabundfile);
 526                 outputTypes["rabund"].push_back(outputFileName); outputNames.push_back(outputFileName);
 527
 528                 ofstream out;
 529                 m->openOutputFile(outputFileName, out);
 530
 531                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 532                 InputData input(rabundfile, "rabund");
 533                 RAbundVector* rabund = input.getRAbundVector();
 534                 string lastLabel = rabund->getLabel();
 535                 set<string> processedLabels;
 536                 set<string> userLabels = labels;
 537
 538                 while((rabund != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 539
 540                         if (m->control_pressed) { delete rabund; out.close(); return 0; }
 541
 542                         if(allLines == 1 || labels.count(rabund->getLabel()) == 1){
 543
 544                                 m->mothurOut(rabund->getLabel()); m->mothurOutEndLine();
 545                                 processedLabels.insert(rabund->getLabel());
 546                                 userLabels.erase(rabund->getLabel());
 547
 548                                 RAbundVector newRabund; newRabund.setLabel(rabund->getLabel());
 549                                 for (int i = 0; i < rabund->getNumBins(); i++) {
 550                                         if (rabund->get(i) > nseqs) {
 551                                                 newRabund.push_back(rabund->get(i));
 552                                         }
 553                                 }
 554                                 if (newRabund.getNumBins() > 0) { newRabund.print(out); }
 555                         }
 556
 557                         if ((m->anyLabelsToProcess(rabund->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 558                                 string saveLabel = rabund->getLabel();
 559
 560                                 delete rabund;
 561                                 rabund = input.getRAbundVector(lastLabel);
 562
 563                                 m->mothurOut(rabund->getLabel()); m->mothurOutEndLine();
 564                                 processedLabels.insert(rabund->getLabel());
 565                                 userLabels.erase(rabund->getLabel());
 566
 567                                 RAbundVector newRabund; newRabund.setLabel(rabund->getLabel());
 568                                 for (int i = 0; i < rabund->getNumBins(); i++) {
 569                                         if (rabund->get(i) > nseqs) {
 570                                                 newRabund.push_back(rabund->get(i));
 571                                         }
 572                                 }
 573                                 if (newRabund.getNumBins() > 0) { newRabund.print(out); }
 574
 575                                 //restore real lastlabel to save below
 576                                 rabund->setLabel(saveLabel);
 577                         }
 578
 579                         lastLabel = rabund->getLabel();
 580
 581                         delete rabund;
 582                         rabund = input.getRAbundVector();
 583                 }
 584
 585                 if (m->control_pressed) {  out.close(); return 0; }
 586
 587                 //output error messages about any remaining user labels
 588                 set<string>::iterator it;
 589                 bool needToRun = false;
 590                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 591                         m->mothurOut("Your file does not include the label " + *it);
 592                         if (processedLabels.count(lastLabel) != 1) {
 593                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 594                                 needToRun = true;
 595                         }else {
 596                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 597                         }
 598                 }
 599
 600                 //run last label if you need to
 601                 if (needToRun == true)  {
 602                         if (rabund != NULL) {   delete rabund;  }
 603                         rabund = input.getRAbundVector(lastLabel);
 604
 605                         m->mothurOut(rabund->getLabel()); m->mothurOutEndLine();
 606
 607                         RAbundVector newRabund; newRabund.setLabel(rabund->getLabel());
 608                         for (int i = 0; i < rabund->getNumBins(); i++) {
 609                                 if (rabund->get(i) > nseqs) {
 610                                         newRabund.push_back(rabund->get(i));
 611                                 }
 612                         }
 613                         if (newRabund.getNumBins() > 0) { newRabund.print(out); }
 614
 615                         delete rabund;
 616                 }
 617
 618                 return 0;
 619         }
 620         catch(exception& e) {
 621                 m->errorOut(e, "RemoveRareCommand", "processRabund");
 622                 exit(1);
 623         }
 624 }
 625 //**********************************************************************************************************************
 626 int RemoveRareCommand::processShared(){
 627         try {
 628                 m->setGroups(Groups);
 629
 630                 string thisOutputDir = outputDir;
 631                 if (outputDir == "") {  thisOutputDir += m->hasPath(sharedfile);  }
 632                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(sharedfile)) + "pick" +  m->getExtension(sharedfile);
 633                 outputTypes["shared"].push_back(outputFileName); outputNames.push_back(outputFileName);
 634
 635                 ofstream out;
 636                 m->openOutputFile(outputFileName, out);
 637
 638                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 639                 InputData input(sharedfile, "sharedfile");
 640                 vector<SharedRAbundVector*> lookup = input.getSharedRAbundVectors();
 641                 string lastLabel = lookup[0]->getLabel();
 642                 set<string> processedLabels;
 643                 set<string> userLabels = labels;
 644
 645                 while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 646
 647                         if (m->control_pressed) { for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }  out.close(); return 0; }
 648
 649                         if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){
 650
 651                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 652                                 processedLabels.insert(lookup[0]->getLabel());
 653                                 userLabels.erase(lookup[0]->getLabel());
 654
 655                                 if (!m->printedHeaders) { lookup[0]->printHeaders(out); }
 656                                 processLookup(lookup, out);
 657                         }
 658
 659                         if ((m->anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 660                                 string saveLabel = lookup[0]->getLabel();
 661
 662                                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 663                                 lookup = input.getSharedRAbundVectors(lastLabel);
 664
 665                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 666                                 processedLabels.insert(lookup[0]->getLabel());
 667                                 userLabels.erase(lookup[0]->getLabel());
 668
 669                                 if (!m->printedHeaders) { lookup[0]->printHeaders(out); }
 670                                 processLookup(lookup, out);
 671
 672                                 //restore real lastlabel to save below
 673                                 lookup[0]->setLabel(saveLabel);
 674                         }
 675
 676                         lastLabel = lookup[0]->getLabel();
 677
 678                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 679                         lookup = input.getSharedRAbundVectors();
 680                 }
 681
 682                 if (m->control_pressed) {  out.close(); return 0; }
 683
 684                 //output error messages about any remaining user labels
 685                 set<string>::iterator it;
 686                 bool needToRun = false;
 687                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 688                         m->mothurOut("Your file does not include the label " + *it);
 689                         if (processedLabels.count(lastLabel) != 1) {
 690                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 691                                 needToRun = true;
 692                         }else {
 693                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 694                         }
 695                 }
 696
 697                 //run last label if you need to
 698                 if (needToRun == true)  {
 699                         for (int i = 0; i < lookup.size(); i++) {  if (lookup[i] != NULL) {     delete lookup[i];       }  }
 700                         lookup = input.getSharedRAbundVectors(lastLabel);
 701
 702                         m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 703
 704                         if (!m->printedHeaders) { lookup[0]->printHeaders(out); }
 705                         processLookup(lookup, out);
 706
 707                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 708                 }
 709
 710                 return 0;
 711         }
 712         catch(exception& e) {
 713                 m->errorOut(e, "RemoveRareCommand", "processSabund");
 714                 exit(1);
 715         }
 716 }
 717 //**********************************************************************************************************************
 718 int RemoveRareCommand::processLookup(vector<SharedRAbundVector*>& lookup, ofstream& out){
 719         try {
 720
 721                 vector<SharedRAbundVector> newRabunds;  newRabunds.resize(lookup.size());
 722                 for (int i = 0; i < lookup.size(); i++) {
 723                         newRabunds[i].setGroup(lookup[i]->getGroup());
 724                         newRabunds[i].setLabel(lookup[i]->getLabel());
 725                 }
 726
 727                 if (byGroup) {
 728
 729                         //for each otu
 730                         for (int i = 0; i < lookup[0]->getNumBins(); i++) {
 731                                 bool allZero = true;
 732
 733                                 if (m->control_pressed) { return 0; }
 734
 735                                 //for each group
 736                                 for (int j = 0; j < lookup.size(); j++) {
 737
 738                                         //are you rare?
 739                                         if (lookup[j]->getAbundance(i) > nseqs) {
 740                                                 newRabunds[j].push_back(lookup[j]->getAbundance(i), newRabunds[j].getGroup());
 741                                                 allZero = false;
 742                                         }else {
 743                                                 newRabunds[j].push_back(0, newRabunds[j].getGroup());
 744                                         }
 745                                 }
 746
 747                                 //eliminates zero otus
 748                                 if (allZero) { for (int j = 0; j < newRabunds.size(); j++) {  newRabunds[j].pop_back(); } }
 749                         }
 750                 }else {
 751                         //for each otu
 752                         for (int i = 0; i < lookup[0]->getNumBins(); i++) {
 753
 754                                 if (m->control_pressed) { return 0; }
 755
 756                                 int totalAbund = 0;
 757                                 //get total otu abundance
 758                                 for (int j = 0; j < lookup.size(); j++) {
 759                                         newRabunds[j].push_back(lookup[j]->getAbundance(i), newRabunds[j].getGroup());
 760                                         totalAbund += lookup[j]->getAbundance(i);
 761                                 }
 762
 763                                 //eliminates otus below rare cutoff
 764                                 if (totalAbund <= nseqs) { for (int j = 0; j < newRabunds.size(); j++) {  newRabunds[j].pop_back(); } }
 765                         }
 766                 }
 767
 768                 //do we have any otus above the rare cutoff
 769                 if (newRabunds[0].getNumBins() != 0) {
 770                         for (int j = 0; j < newRabunds.size(); j++) {
 771                                 out << newRabunds[j].getLabel() << '\t' << newRabunds[j].getGroup() << '\t';
 772                                 newRabunds[j].print(out);
 773                         }
 774                 }
 775
 776                 return 0;
 777         }
 778         catch(exception& e) {
 779                 m->errorOut(e, "RemoveRareCommand", "processLookup");
 780                 exit(1);
 781         }
 782 }
 783 //**********************************************************************************************************************
 784
 785
 786
 787