subsamplecommand.cpp

   1 /*
   2  *  subsamplecommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 10/27/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "subsamplecommand.h"
  11 #include "sharedutilities.h"
  12 #include "deconvolutecommand.h"
  13
  14 //**********************************************************************************************************************
  15 vector<string> SubSampleCommand::setParameters(){
  16         try {
  17                 CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "FLSSR", "none",false,false); parameters.push_back(pfasta);
  18                 CommandParameter pname("name", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pname);
  19                 CommandParameter pgroup("group", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pgroup);
  20                 CommandParameter plist("list", "InputTypes", "", "", "none", "FLSSR", "none",false,false); parameters.push_back(plist);
  21                 CommandParameter pshared("shared", "InputTypes", "", "", "none", "FLSSR", "none",false,false); parameters.push_back(pshared);
  22                 CommandParameter prabund("rabund", "InputTypes", "", "", "none", "FLSSR", "none",false,false); parameters.push_back(prabund);
  23                 CommandParameter psabund("sabund", "InputTypes", "", "", "none", "FLSSR", "none",false,false); parameters.push_back(psabund);
  24                 CommandParameter plabel("label", "String", "", "", "", "", "",false,false); parameters.push_back(plabel);
  25                 CommandParameter pgroups("groups", "String", "", "", "", "", "",false,false); parameters.push_back(pgroups);
  26                 CommandParameter psize("size", "Number", "", "0", "", "", "",false,false); parameters.push_back(psize);
  27                 CommandParameter ppersample("persample", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(ppersample);
  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, "SubSampleCommand", "setParameters");
  37                 exit(1);
  38         }
  39 }
  40 //**********************************************************************************************************************
  41 string SubSampleCommand::getHelpString(){
  42         try {
  43                 string helpString = "";
  44                 helpString += "The sub.sample command is designed to be used as a way to normalize your data, or create a smaller set from your original set.\n";
  45                 helpString += "The sub.sample command parameters are fasta, name, list, group, rabund, sabund, shared, groups, size, persample and label.  You must provide a fasta, list, sabund, rabund or shared file as an input file.\n";
  46                 helpString += "The namefile is only used with the fasta file, not with the listfile, because the list file should contain all sequences.\n";
  47                 helpString += "The groups parameter allows you to specify which of the groups in your groupfile you would like included. The group names are separated by dashes.\n";
  48                 helpString += "The label parameter allows you to select what distance levels you would like, and are also separated by dashes.\n";
  49                 helpString += "The size parameter allows you indicate the size of your subsample.\n";
  50                 helpString += "The persample parameter allows you indicate you want to select subsample of the same size from each of your groups, default=false. It is only used with the list and fasta files if a groupfile is given.\n";
  51                 helpString += "persample=false will select a random set of sequences of the size you select, but the number of seqs from each group may differ.\n";
  52                 helpString += "The size parameter is not set: with shared file size=number of seqs in smallest sample, with all other files if a groupfile is given and persample=true, then size=number of seqs in smallest sample, otherwise size=10% of number of seqs.\n";
  53                 helpString += "The sub.sample command should be in the following format: sub.sample(list=yourListFile, group=yourGroupFile, groups=yourGroups, label=yourLabels).\n";
  54                 helpString += "Example sub.sample(list=abrecovery.fn.list, group=abrecovery.groups, groups=B-C, size=20).\n";
  55                 helpString += "The default value for groups is all the groups in your groupfile, and all labels in your inputfile will be used.\n";
  56                 helpString += "The sub.sample command outputs a .subsample file.\n";
  57                 helpString += "Note: No spaces between parameter labels (i.e. groups), '=' and parameters (i.e.yourGroups).\n";
  58                 return helpString;
  59         }
  60         catch(exception& e) {
  61                 m->errorOut(e, "SubSampleCommand", "getHelpString");
  62                 exit(1);
  63         }
  64 }
  65 //**********************************************************************************************************************
  66 SubSampleCommand::SubSampleCommand(){
  67         try {
  68                 abort = true; calledHelp = true;
  69                 setParameters();
  70                 vector<string> tempOutNames;
  71                 outputTypes["shared"] = tempOutNames;
  72                 outputTypes["list"] = tempOutNames;
  73                 outputTypes["rabund"] = tempOutNames;
  74                 outputTypes["sabund"] = tempOutNames;
  75                 outputTypes["fasta"] = tempOutNames;
  76                 outputTypes["name"] = tempOutNames;
  77                 outputTypes["group"] = tempOutNames;
  78         }
  79         catch(exception& e) {
  80                 m->errorOut(e, "SubSampleCommand", "GetRelAbundCommand");
  81                 exit(1);
  82         }
  83 }
  84 //**********************************************************************************************************************
  85 SubSampleCommand::SubSampleCommand(string option) {
  86         try {
  87                 abort = false; calledHelp = false;
  88                 allLines = 1;
  89
  90                 //allow user to run help
  91                 if(option == "help") { help(); abort = true; calledHelp = true; }
  92                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  93
  94                 else {
  95                         vector<string> myArray = setParameters();
  96
  97                         OptionParser parser(option);
  98                         map<string,string> parameters = parser.getParameters();
  99
 100                         ValidParameters validParameter;
 101
 102                         //check to make sure all parameters are valid for command
 103                         map<string,string>::iterator it;
 104                         for (it = parameters.begin(); it != parameters.end(); it++) {
 105                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 106                         }
 107
 108                         //initialize outputTypes
 109                         vector<string> tempOutNames;
 110                         outputTypes["shared"] = tempOutNames;
 111                         outputTypes["list"] = tempOutNames;
 112                         outputTypes["rabund"] = tempOutNames;
 113                         outputTypes["sabund"] = tempOutNames;
 114                         outputTypes["fasta"] = tempOutNames;
 115                         outputTypes["name"] = tempOutNames;
 116                         outputTypes["group"] = tempOutNames;
 117
 118                         //if the user changes the output directory command factory will send this info to us in the output parameter
 119                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
 120
 121                         //if the user changes the input directory command factory will send this info to us in the output parameter
 122                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 123                         if (inputDir == "not found"){   inputDir = "";          }
 124                         else {
 125                                 string path;
 126                                 it = parameters.find("list");
 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["list"] = inputDir + it->second;             }
 132                                 }
 133
 134                                 it = parameters.find("fasta");
 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["fasta"] = inputDir + it->second;            }
 140                                 }
 141
 142                                 it = parameters.find("shared");
 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["shared"] = inputDir + it->second;           }
 148                                 }
 149
 150                                 it = parameters.find("group");
 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["group"] = inputDir + it->second;            }
 156                                 }
 157
 158                                 it = parameters.find("sabund");
 159                                 //user has given a template file
 160                                 if(it != parameters.end()){
 161                                         path = m->hasPath(it->second);
 162                                         //if the user has not given a path then, add inputdir. else leave path alone.
 163                                         if (path == "") {       parameters["sabund"] = inputDir + it->second;           }
 164                                 }
 165
 166                                 it = parameters.find("rabund");
 167                                 //user has given a template file
 168                                 if(it != parameters.end()){
 169                                         path = m->hasPath(it->second);
 170                                         //if the user has not given a path then, add inputdir. else leave path alone.
 171                                         if (path == "") {       parameters["rabund"] = inputDir + it->second;           }
 172                                 }
 173
 174                                 it = parameters.find("name");
 175                                 //user has given a template file
 176                                 if(it != parameters.end()){
 177                                         path = m->hasPath(it->second);
 178                                         //if the user has not given a path then, add inputdir. else leave path alone.
 179                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 180                                 }
 181                         }
 182
 183                         //check for required parameters
 184                         listfile = validParameter.validFile(parameters, "list", true);
 185                         if (listfile == "not open") { listfile = ""; abort = true; }
 186                         else if (listfile == "not found") { listfile = ""; }
 187                         else { m->setListFile(listfile); }
 188
 189                         sabundfile = validParameter.validFile(parameters, "sabund", true);
 190                         if (sabundfile == "not open") { sabundfile = ""; abort = true; }
 191                         else if (sabundfile == "not found") { sabundfile = ""; }
 192                         else { m->setSabundFile(sabundfile); }
 193
 194                         rabundfile = validParameter.validFile(parameters, "rabund", true);
 195                         if (rabundfile == "not open") { rabundfile = ""; abort = true; }
 196                         else if (rabundfile == "not found") { rabundfile = ""; }
 197                         else { m->setRabundFile(rabundfile); }
 198
 199                         fastafile = validParameter.validFile(parameters, "fasta", true);
 200                         if (fastafile == "not open") { fastafile = ""; abort = true; }
 201                         else if (fastafile == "not found") { fastafile = ""; }
 202                         else { m->setFastaFile(fastafile); }
 203
 204                         sharedfile = validParameter.validFile(parameters, "shared", true);
 205                         if (sharedfile == "not open") { sharedfile = ""; abort = true; }
 206                         else if (sharedfile == "not found") { sharedfile = ""; }
 207                         else { m->setSharedFile(sharedfile); }
 208
 209                         namefile = validParameter.validFile(parameters, "name", true);
 210                         if (namefile == "not open") { namefile = ""; abort = true; }
 211                         else if (namefile == "not found") { namefile = ""; }
 212                         else { m->setNameFile(namefile); }
 213
 214                         groupfile = validParameter.validFile(parameters, "group", true);
 215                         if (groupfile == "not open") { groupfile = ""; abort = true; }
 216                         else if (groupfile == "not found") { groupfile = ""; }
 217                         else { m->setGroupFile(groupfile); }
 218
 219                         //check for optional parameter and set defaults
 220                         // ...at some point should added some additional type checking...
 221                         label = validParameter.validFile(parameters, "label", false);
 222                         if (label == "not found") { label = ""; }
 223                         else {
 224                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 225                                 else { allLines = 1;  }
 226                         }
 227
 228                         groups = validParameter.validFile(parameters, "groups", false);
 229                         if (groups == "not found") { groups = ""; pickedGroups = false; }
 230                         else {
 231                                 pickedGroups = true;
 232                                 m->splitAtDash(groups, Groups);
 233                                 m->setGroups(Groups);
 234                         }
 235
 236                         string temp = validParameter.validFile(parameters, "size", false);              if (temp == "not found"){       temp = "0";             }
 237                         convert(temp, size);
 238
 239                         temp = validParameter.validFile(parameters, "persample", false);                if (temp == "not found"){       temp = "f";             }
 240                         persample = m->isTrue(temp);
 241
 242                         if (groupfile == "") { persample = false; }
 243
 244                         if ((namefile != "") && (fastafile == "")) { m->mothurOut("You may only use a namefile with a fastafile."); m->mothurOutEndLine(); abort = true; }
 245
 246                         if ((fastafile == "") && (listfile == "") && (sabundfile == "") && (rabundfile == "") && (sharedfile == "")) {
 247                                 m->mothurOut("You must provide a fasta, list, sabund, rabund or shared file as an input file."); m->mothurOutEndLine(); abort = true; }
 248
 249                         if (pickedGroups && ((groupfile == "") && (sharedfile == ""))) {
 250                                 m->mothurOut("You cannot pick groups without a valid group file or shared file."); m->mothurOutEndLine(); abort = true; }
 251
 252                         if ((groupfile != "") && ((fastafile == "") && (listfile == ""))) {
 253                                 m->mothurOut("Group file only valid with listfile or fastafile."); m->mothurOutEndLine(); abort = true; }
 254
 255                         if ((groupfile != "") && ((fastafile != "") && (listfile != ""))) {
 256                                 m->mothurOut("A new group file can only be made from the subsample of a listfile or fastafile, not both. Please correct."); m->mothurOutEndLine(); abort = true; }
 257
 258                 }
 259
 260         }
 261         catch(exception& e) {
 262                 m->errorOut(e, "SubSampleCommand", "SubSampleCommand");
 263                 exit(1);
 264         }
 265 }
 266 //**********************************************************************************************************************
 267
 268 int SubSampleCommand::execute(){
 269         try {
 270
 271                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 272
 273                 if (sharedfile != "")   {   getSubSampleShared();       }
 274                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]); return 0; } }
 275
 276                 if (listfile != "")             {   getSubSampleList();         }
 277                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]); return 0; } }
 278
 279                 if (rabundfile != "")   {   getSubSampleRabund();       }
 280                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]); return 0; } }
 281
 282                 if (sabundfile != "")   {   getSubSampleSabund();       }
 283                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]); return 0; } }
 284
 285                 if (fastafile != "")    {   getSubSampleFasta();        }
 286                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]); return 0; } }
 287
 288                 //set fasta file as new current fastafile
 289                 string current = "";
 290                 itTypes = outputTypes.find("fasta");
 291                 if (itTypes != outputTypes.end()) {
 292                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 293                 }
 294
 295                 itTypes = outputTypes.find("name");
 296                 if (itTypes != outputTypes.end()) {
 297                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
 298                 }
 299
 300                 itTypes = outputTypes.find("group");
 301                 if (itTypes != outputTypes.end()) {
 302                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setGroupFile(current); }
 303                 }
 304
 305                 itTypes = outputTypes.find("list");
 306                 if (itTypes != outputTypes.end()) {
 307                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 308                 }
 309
 310                 itTypes = outputTypes.find("shared");
 311                 if (itTypes != outputTypes.end()) {
 312                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setSharedFile(current); }
 313                 }
 314
 315                 itTypes = outputTypes.find("rabund");
 316                 if (itTypes != outputTypes.end()) {
 317                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setRabundFile(current); }
 318                 }
 319
 320                 itTypes = outputTypes.find("sabund");
 321                 if (itTypes != outputTypes.end()) {
 322                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setSabundFile(current); }
 323                 }
 324
 325
 326                 m->mothurOutEndLine();
 327                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 328                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 329                 m->mothurOutEndLine();
 330
 331                 return 0;
 332         }
 333         catch(exception& e) {
 334                 m->errorOut(e, "SubSampleCommand", "execute");
 335                 exit(1);
 336         }
 337 }
 338 //**********************************************************************************************************************
 339 int SubSampleCommand::getSubSampleFasta() {
 340         try {
 341
 342                 if (namefile != "") { readNames(); }    //fills names with all names in namefile.
 343                 else { getNames(); }//no name file, so get list of names to pick from
 344
 345                 GroupMap* groupMap;
 346                 if (groupfile != "") {
 347
 348                         groupMap = new GroupMap(groupfile);
 349                         groupMap->readMap();
 350
 351                         //takes care of user setting groupNames that are invalid or setting groups=all
 352                         SharedUtil* util = new SharedUtil();
 353                         vector<string> namesGroups = groupMap->getNamesOfGroups();
 354                         util->setGroups(Groups, namesGroups);
 355                         delete util;
 356
 357                         //file mismatch quit
 358                         if (names.size() != groupMap->getNumSeqs()) {
 359                                 m->mothurOut("[ERROR]: your fasta file contains " + toString(names.size()) + " sequences, and your groupfile contains " + toString(groupMap->getNumSeqs()) + ", please correct.");
 360                                 m->mothurOutEndLine();
 361                                 delete groupMap;
 362                                 return 0;
 363                         }
 364                 }
 365
 366                 if (m->control_pressed) { return 0; }
 367
 368
 369                 //make sure that if your picked groups size is not too big
 370                 int thisSize = names.size();
 371                 if (persample) {
 372                         if (size == 0) { //user has not set size, set size = smallest samples size
 373                                 size = groupMap->getNumSeqs(Groups[0]);
 374                                 for (int i = 1; i < Groups.size(); i++) {
 375                                         int thisSize = groupMap->getNumSeqs(Groups[i]);
 376
 377                                         if (thisSize < size) {  size = thisSize;        }
 378                                 }
 379                         }else { //make sure size is not too large
 380                                 vector<string> newGroups;
 381                                 for (int i = 0; i < Groups.size(); i++) {
 382                                         int thisSize = groupMap->getNumSeqs(Groups[i]);
 383
 384                                         if (thisSize >= size) { newGroups.push_back(Groups[i]); }
 385                                         else {  m->mothurOut("You have selected a size that is larger than " + Groups[i] + " number of sequences, removing " + Groups[i] + "."); m->mothurOutEndLine(); }
 386                                 }
 387                                 Groups = newGroups;
 388                         }
 389
 390                         m->mothurOut("Sampling " + toString(size) + " from each group."); m->mothurOutEndLine();
 391                 }else {
 392                         if (pickedGroups) {
 393                                 int total = 0;
 394                                 for(int i = 0; i < Groups.size(); i++) {
 395                                         total += groupMap->getNumSeqs(Groups[i]);
 396                                 }
 397
 398                                 if (size == 0) { //user has not set size, set size = 10% samples size
 399                                         size = int (total * 0.10);
 400                                 }
 401
 402                                 if (total < size) {
 403                                         if (size != 0) {
 404                                                 m->mothurOut("Your size is too large for the number of groups you selected. Adjusting to " + toString(int (total * 0.10)) + "."); m->mothurOutEndLine();
 405                                         }
 406                                         size = int (total * 0.10);
 407                                 }
 408
 409                                 m->mothurOut("Sampling " + toString(size) + " from " + toString(total) + "."); m->mothurOutEndLine();
 410                         }
 411
 412                         if (size == 0) { //user has not set size, set size = 10% samples size
 413                                 size = int (names.size() * 0.10);
 414                         }
 415
 416                         if (size > thisSize) { m->mothurOut("Your fasta file only contains " + toString(thisSize) + " sequences. Setting size to " + toString(thisSize) + "."); m->mothurOutEndLine();
 417                                 size = thisSize;
 418                         }
 419
 420                         if (!pickedGroups) { m->mothurOut("Sampling " + toString(size) + " from " + toString(thisSize) + "."); m->mothurOutEndLine(); }
 421
 422                 }
 423                 random_shuffle(names.begin(), names.end());
 424
 425                 set<string> subset; //dont want repeat sequence names added
 426                 if (persample) {
 427                         //initialize counts
 428                         map<string, int> groupCounts;
 429                         for (int i = 0; i < Groups.size(); i++) { groupCounts[Groups[i]] = 0; }
 430
 431                         for (int j = 0; j < names.size(); j++) {
 432
 433                                 if (m->control_pressed) { return 0; }
 434
 435                                 string group = groupMap->getGroup(names[j]);
 436                                 if (group == "not found") { m->mothurOut("[ERROR]: " + names[j] + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
 437                                 else{
 438                                         if (groupCounts[group] < size) {        subset.insert(names[j]);        }
 439                                 }
 440                         }
 441                 }else {
 442
 443                         //randomly select a subset of those names to include in the subsample
 444                         //since names was randomly shuffled just grab the next one
 445                         for (int j = 0; j < names.size(); j++) {
 446
 447                                 if (m->control_pressed) { return 0; }
 448
 449                                 if (groupfile != "") { //if there is a groupfile given fill in group info
 450                                         string group = groupMap->getGroup(names[j]);
 451                                         if (group == "not found") { m->mothurOut("[ERROR]: " + names[j] + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
 452
 453                                         if (pickedGroups) { //if hte user picked groups, we only want to keep the names of sequences from those groups
 454                                                 if (m->inUsersGroups(group, Groups)) {
 455                                                         subset.insert(names[j]);
 456                                                 }
 457                                         }else{
 458                                                 subset.insert(names[j]);
 459                                         }
 460                                 }else{ //save everyone, group
 461                                         subset.insert(names[j]);
 462                                 }
 463
 464                                 //do we have enough??
 465                                 if (subset.size() == size) { break; }
 466                         }
 467                 }
 468
 469                 if (subset.size() == 0) {  m->mothurOut("The size you selected is too large, skipping fasta file."); m->mothurOutEndLine();  return 0; }
 470
 471                 string thisOutputDir = outputDir;
 472                 if (outputDir == "") {  thisOutputDir += m->hasPath(fastafile);  }
 473                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(fastafile)) + "subsample" + m->getExtension(fastafile);
 474
 475                 ofstream out;
 476                 m->openOutputFile(outputFileName, out);
 477
 478                 //read through fasta file outputting only the names on the subsample list
 479                 ifstream in;
 480                 m->openInputFile(fastafile, in);
 481
 482                 string thisname;
 483                 int count = 0;
 484                 map<string, vector<string> >::iterator itNameMap;
 485
 486                 while(!in.eof()){
 487
 488                         if (m->control_pressed) { in.close(); out.close();  return 0; }
 489
 490                         Sequence currSeq(in);
 491                         thisname = currSeq.getName();
 492
 493                         if (thisname != "") {
 494
 495                                 //does the subset contain a sequence that this sequence represents
 496                                 itNameMap = nameMap.find(thisname);
 497                                 if (itNameMap != nameMap.end()) {
 498                                         vector<string> nameRepresents = itNameMap->second;
 499
 500                                         for (int i = 0; i < nameRepresents.size(); i++){
 501                                                 if (subset.count(nameRepresents[i]) != 0) {
 502                                                         out << ">" << nameRepresents[i] << endl << currSeq.getAligned() << endl;
 503                                                         count++;
 504                                                 }
 505                                         }
 506                                 }else{
 507                                         m->mothurOut("[ERROR]: " + thisname + " is not in your namefile, please correct."); m->mothurOutEndLine();
 508                                 }
 509                         }
 510                         m->gobble(in);
 511                 }
 512                 in.close();
 513                 out.close();
 514
 515                 if (count != subset.size()) {
 516                         m->mothurOut("[ERROR]: The subset selected contained " + toString(subset.size()) + " sequences, but I only found " + toString(count) + " of those in the fastafile."); m->mothurOutEndLine();
 517                 }
 518
 519                 if (namefile != "") {
 520                         m->mothurOut("Deconvoluting subsampled fasta file... "); m->mothurOutEndLine();
 521
 522                         //use unique.seqs to create new name and fastafile
 523                         string inputString = "fasta=" + outputFileName;
 524                         m->mothurOut("/******************************************/"); m->mothurOutEndLine();
 525                         m->mothurOut("Running command: unique.seqs(" + inputString + ")"); m->mothurOutEndLine();
 526
 527                         Command* uniqueCommand = new DeconvoluteCommand(inputString);
 528                         uniqueCommand->execute();
 529
 530                         map<string, vector<string> > filenames = uniqueCommand->getOutputFiles();
 531
 532                         delete uniqueCommand;
 533
 534                         outputTypes["name"].push_back(filenames["name"][0]);  outputNames.push_back(filenames["name"][0]);
 535                         m->mothurRemove(outputFileName);
 536                         outputFileName = filenames["fasta"][0];
 537
 538                         m->mothurOut("/******************************************/"); m->mothurOutEndLine();
 539
 540                         m->mothurOut("Done."); m->mothurOutEndLine();
 541                 }
 542
 543                 outputTypes["fasta"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 544
 545                 //if a groupfile is provided read through the group file only outputting the names on the subsample list
 546                 if (groupfile != "") {
 547
 548                         string groupOutputDir = outputDir;
 549                         if (outputDir == "") {  groupOutputDir += m->hasPath(groupfile);  }
 550                         string groupOutputFileName = groupOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "subsample" + m->getExtension(groupfile);
 551
 552                         ofstream outGroup;
 553                         m->openOutputFile(groupOutputFileName, outGroup);
 554                         outputTypes["group"].push_back(groupOutputFileName);  outputNames.push_back(groupOutputFileName);
 555
 556                         ifstream inGroup;
 557                         m->openInputFile(groupfile, inGroup);
 558                         string name, group;
 559
 560                         while(!inGroup.eof()){
 561
 562                                 if (m->control_pressed) { inGroup.close(); outGroup.close(); return 0; }
 563
 564                                 inGroup >> name;        m->gobble(inGroup);                     //read from first column
 565                                 inGroup >> group;                       //read from second column
 566
 567                                 //if this name is in the accnos file
 568                                 if (subset.count(name) != 0) {
 569                                         outGroup << name << '\t' << group << endl;
 570                                         subset.erase(name);
 571                                 }
 572
 573                                 m->gobble(inGroup);
 574                         }
 575                         inGroup.close();
 576                         outGroup.close();
 577
 578                         //sanity check
 579                         if (subset.size() != 0) {
 580                                 m->mothurOut("Your groupfile does not match your fasta file."); m->mothurOutEndLine();
 581                                 for (set<string>::iterator it = subset.begin(); it != subset.end(); it++) {
 582                                         m->mothurOut("[ERROR]: " + *it + " is missing from your groupfile."); m->mothurOutEndLine();
 583                                 }
 584                         }
 585                 }
 586
 587                 return 0;
 588
 589         }
 590         catch(exception& e) {
 591                 m->errorOut(e, "SubSampleCommand", "getSubSampleFasta");
 592                 exit(1);
 593         }
 594 }
 595 //**********************************************************************************************************************
 596 int SubSampleCommand::getNames() {
 597         try {
 598
 599                 ifstream in;
 600                 m->openInputFile(fastafile, in);
 601
 602                 string thisname;
 603                 while(!in.eof()){
 604
 605                         if (m->control_pressed) { in.close(); return 0; }
 606
 607                         Sequence currSeq(in);
 608                         thisname = currSeq.getName();
 609
 610                         if (thisname != "") {
 611                                 vector<string> temp; temp.push_back(thisname);
 612                                 nameMap[thisname] = temp;
 613                                 names.push_back(thisname);
 614                         }
 615                         m->gobble(in);
 616                 }
 617                 in.close();
 618
 619                 return 0;
 620
 621         }
 622         catch(exception& e) {
 623                 m->errorOut(e, "SubSampleCommand", "getNames");
 624                 exit(1);
 625         }
 626 }
 627 //**********************************************************************************************************************
 628 int SubSampleCommand::readNames() {
 629         try {
 630
 631                 ifstream in;
 632                 m->openInputFile(namefile, in);
 633
 634                 string thisname, repnames;
 635                 map<string, vector<string> >::iterator it;
 636
 637                 while(!in.eof()){
 638
 639                         if (m->control_pressed) { in.close(); return 0; }
 640
 641                         in >> thisname;         m->gobble(in);          //read from first column
 642                         in >> repnames;                 //read from second column
 643
 644                         it = nameMap.find(thisname);
 645                         if (it == nameMap.end()) {
 646
 647                                 vector<string> splitRepNames;
 648                                 m->splitAtComma(repnames, splitRepNames);
 649
 650                                 nameMap[thisname] = splitRepNames;
 651                                 for (int i = 0; i < splitRepNames.size(); i++) { names.push_back(splitRepNames[i]); }
 652
 653                         }else{  m->mothurOut(thisname + " is already in namesfile. I will use first definition."); m->mothurOutEndLine();  }
 654
 655                         m->gobble(in);
 656                 }
 657                 in.close();
 658
 659                 return 0;
 660
 661         }
 662         catch(exception& e) {
 663                 m->errorOut(e, "SubSampleCommand", "readNames");
 664                 exit(1);
 665         }
 666 }
 667 //**********************************************************************************************************************
 668 int SubSampleCommand::getSubSampleShared() {
 669         try {
 670
 671                 InputData* input = new InputData(sharedfile, "sharedfile");
 672                 vector<SharedRAbundVector*> lookup = input->getSharedRAbundVectors();
 673                 string lastLabel = lookup[0]->getLabel();
 674
 675                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 676                 set<string> processedLabels;
 677                 set<string> userLabels = labels;
 678
 679                 if (size == 0) { //user has not set size, set size = smallest samples size
 680                         size = lookup[0]->getNumSeqs();
 681                         for (int i = 1; i < lookup.size(); i++) {
 682                                 int thisSize = lookup[i]->getNumSeqs();
 683
 684                                 if (thisSize < size) {  size = thisSize;        }
 685                         }
 686                 }else {
 687                         m->clearGroups();
 688                         Groups.clear();
 689                         vector<SharedRAbundVector*> temp;
 690                         for (int i = 0; i < lookup.size(); i++) {
 691                                 if (lookup[i]->getNumSeqs() < size) {
 692                                         m->mothurOut(lookup[i]->getGroup() + " contains " + toString(lookup[i]->getNumSeqs()) + ". Eliminating."); m->mothurOutEndLine();
 693                                         delete lookup[i];
 694                                 }else {
 695                                         Groups.push_back(lookup[i]->getGroup());
 696                                         temp.push_back(lookup[i]);
 697                                 }
 698                         }
 699                         lookup = temp;
 700                         m->setGroups(Groups);
 701                 }
 702
 703                 if (lookup.size() == 0) {  m->mothurOut("The size you selected is too large, skipping shared file."); m->mothurOutEndLine(); delete input; return 0; }
 704
 705                 m->mothurOut("Sampling " + toString(size) + " from each group."); m->mothurOutEndLine();
 706
 707                 //as long as you are not at the end of the file or done wih the lines you want
 708                 while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 709                         if (m->control_pressed) {  delete input; for (int i = 0; i < lookup.size(); i++) {  delete lookup[i]; lookup[i] = NULL; }  return 0;  }
 710
 711                         if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){
 712
 713                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 714
 715                                 processShared(lookup);
 716
 717                                 processedLabels.insert(lookup[0]->getLabel());
 718                                 userLabels.erase(lookup[0]->getLabel());
 719                         }
 720
 721                         if ((m->anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 722                                 string saveLabel = lookup[0]->getLabel();
 723
 724                                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 725
 726                                 lookup = input->getSharedRAbundVectors(lastLabel);
 727                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 728
 729                                 processShared(lookup);
 730
 731                                 processedLabels.insert(lookup[0]->getLabel());
 732                                 userLabels.erase(lookup[0]->getLabel());
 733
 734                                 //restore real lastlabel to save below
 735                                 lookup[0]->setLabel(saveLabel);
 736                         }
 737
 738                         lastLabel = lookup[0]->getLabel();
 739                         //prevent memory leak
 740                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i]; lookup[i] = NULL; }
 741
 742                         //get next line to process
 743                         lookup = input->getSharedRAbundVectors();
 744                 }
 745
 746
 747                 if (m->control_pressed) {   return 0;  }
 748
 749                 //output error messages about any remaining user labels
 750                 set<string>::iterator it;
 751                 bool needToRun = false;
 752                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 753                         m->mothurOut("Your file does not include the label " + *it);
 754                         if (processedLabels.count(lastLabel) != 1) {
 755                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 756                                 needToRun = true;
 757                         }else {
 758                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 759                         }
 760                 }
 761
 762                 //run last label if you need to
 763                 if (needToRun == true)  {
 764                         for (int i = 0; i < lookup.size(); i++) { if (lookup[i] != NULL) { delete lookup[i]; } }
 765                         lookup = input->getSharedRAbundVectors(lastLabel);
 766
 767                         m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 768
 769                         processShared(lookup);
 770
 771                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 772                 }
 773
 774                 delete input;
 775
 776                 return 0;
 777
 778         }
 779         catch(exception& e) {
 780                 m->errorOut(e, "SubSampleCommand", "getSubSampleShared");
 781                 exit(1);
 782         }
 783 }
 784 //**********************************************************************************************************************
 785 int SubSampleCommand::processShared(vector<SharedRAbundVector*>& thislookup) {
 786         try {
 787
 788                 //save mothurOut's binLabels to restore for next label
 789                 vector<string> saveBinLabels = m->currentBinLabels;
 790
 791                 string thisOutputDir = outputDir;
 792                 if (outputDir == "") {  thisOutputDir += m->hasPath(sharedfile);  }
 793                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(sharedfile)) + thislookup[0]->getLabel() + ".subsample" + m->getExtension(sharedfile);
 794
 795
 796                 ofstream out;
 797                 m->openOutputFile(outputFileName, out);
 798                 outputTypes["shared"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 799
 800                 int numBins = thislookup[0]->getNumBins();
 801                 for (int i = 0; i < thislookup.size(); i++) {
 802                         int thisSize = thislookup[i]->getNumSeqs();
 803
 804                         if (thisSize != size) {
 805
 806                                 string thisgroup = thislookup[i]->getGroup();
 807
 808                                 OrderVector* order = new OrderVector();
 809                                 for(int p=0;p<numBins;p++){
 810                                         for(int j=0;j<thislookup[i]->getAbundance(p);j++){
 811                                                 order->push_back(p);
 812                                         }
 813                                 }
 814                                 random_shuffle(order->begin(), order->end());
 815
 816                                 SharedRAbundVector* temp = new SharedRAbundVector(numBins);
 817                                 temp->setLabel(thislookup[i]->getLabel());
 818                                 temp->setGroup(thislookup[i]->getGroup());
 819
 820                                 delete thislookup[i];
 821                                 thislookup[i] = temp;
 822
 823
 824                                 for (int j = 0; j < size; j++) {
 825
 826                                         if (m->control_pressed) { delete order; out.close(); return 0; }
 827
 828                                         //get random number to sample from order between 0 and thisSize-1.
 829                                         //don't need this because of the random shuffle above
 830                                         //int myrand = int((float)(thisSize) * (float)(rand()) / ((float)RAND_MAX+1.0));
 831
 832                                         int bin = order->get(j);
 833
 834                                         int abund = thislookup[i]->getAbundance(bin);
 835                                         thislookup[i]->set(bin, (abund+1), thisgroup);
 836                                 }
 837                                 delete order;
 838                         }
 839                 }
 840
 841                 //subsampling may have created some otus with no sequences in them
 842                 eliminateZeroOTUS(thislookup);
 843
 844                 if (m->control_pressed) { out.close(); return 0; }
 845
 846                 thislookup[0]->printHeaders(out);
 847
 848                 for (int i = 0; i < thislookup.size(); i++) {
 849                         out << thislookup[i]->getLabel() << '\t' << thislookup[i]->getGroup() << '\t';
 850                         thislookup[i]->print(out);
 851                 }
 852
 853                 out.close();
 854
 855                 //save mothurOut's binLabels to restore for next label
 856                 m->currentBinLabels = saveBinLabels;
 857
 858                 return 0;
 859
 860         }
 861         catch(exception& e) {
 862                 m->errorOut(e, "SubSampleCommand", "processShared");
 863                 exit(1);
 864         }
 865 }
 866 //**********************************************************************************************************************
 867 int SubSampleCommand::getSubSampleList() {
 868         try {
 869
 870                 string thisOutputDir = outputDir;
 871                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 872                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "subsample" + m->getExtension(listfile);
 873
 874                 ofstream out;
 875                 m->openOutputFile(outputFileName, out);
 876                 outputTypes["list"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 877
 878                 InputData* input = new InputData(listfile, "list");
 879                 ListVector* list = input->getListVector();
 880                 string lastLabel = list->getLabel();
 881
 882                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 883                 set<string> processedLabels;
 884                 set<string> userLabels = labels;
 885
 886                 ofstream outGroup;
 887                 GroupMap* groupMap;
 888                 if (groupfile != "") {
 889
 890                         groupMap = new GroupMap(groupfile);
 891                         groupMap->readMap();
 892
 893                         //takes care of user setting groupNames that are invalid or setting groups=all
 894                         SharedUtil* util = new SharedUtil();
 895                         vector<string> namesGroups = groupMap->getNamesOfGroups();
 896                         util->setGroups(Groups, namesGroups);
 897                         delete util;
 898
 899                         //create outputfiles
 900                         string groupOutputDir = outputDir;
 901                         if (outputDir == "") {  groupOutputDir += m->hasPath(groupfile);  }
 902                         string groupOutputFileName = groupOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "subsample" + m->getExtension(groupfile);
 903
 904                         m->openOutputFile(groupOutputFileName, outGroup);
 905                         outputTypes["group"].push_back(groupOutputFileName);  outputNames.push_back(groupOutputFileName);
 906
 907                         //file mismatch quit
 908                         if (list->getNumSeqs() != groupMap->getNumSeqs()) {
 909                                 m->mothurOut("[ERROR]: your list file contains " + toString(list->getNumSeqs()) + " sequences, and your groupfile contains " + toString(groupMap->getNumSeqs()) + ", please correct.");
 910                                 m->mothurOutEndLine();
 911                                 delete groupMap;
 912                                 delete list;
 913                                 delete input;
 914                                 out.close();
 915                                 outGroup.close();
 916                                 return 0;
 917                         }
 918                 }
 919
 920                 //make sure that if your picked groups size is not too big
 921                 if (persample) {
 922                         if (size == 0) { //user has not set size, set size = smallest samples size
 923                                 size = groupMap->getNumSeqs(Groups[0]);
 924                                 for (int i = 1; i < Groups.size(); i++) {
 925                                         int thisSize = groupMap->getNumSeqs(Groups[i]);
 926
 927                                         if (thisSize < size) {  size = thisSize;        }
 928                                 }
 929                         }else { //make sure size is not too large
 930                                 vector<string> newGroups;
 931                                 for (int i = 0; i < Groups.size(); i++) {
 932                                         int thisSize = groupMap->getNumSeqs(Groups[i]);
 933
 934                                         if (thisSize >= size) { newGroups.push_back(Groups[i]); }
 935                                         else {  m->mothurOut("You have selected a size that is larger than " + Groups[i] + " number of sequences, removing " + Groups[i] + "."); m->mothurOutEndLine(); }
 936                                 }
 937                                 Groups = newGroups;
 938                         }
 939
 940                         m->mothurOut("Sampling " + toString(size) + " from each group."); m->mothurOutEndLine();
 941                 }else{
 942                         if (pickedGroups) {
 943                                 int total = 0;
 944                                 for(int i = 0; i < Groups.size(); i++) {
 945                                         total += groupMap->getNumSeqs(Groups[i]);
 946                                 }
 947
 948                                 if (size == 0) { //user has not set size, set size = 10% samples size
 949                                         size = int (total * 0.10);
 950                                 }
 951
 952                                 if (total < size) {
 953                                         m->mothurOut("Your size is too large for the number of groups you selected. Adjusting to " + toString(int (total * 0.10)) + "."); m->mothurOutEndLine();
 954                                         size = int (total * 0.10);
 955                                 }
 956
 957                                 m->mothurOut("Sampling " + toString(size) + " from " + toString(total) + "."); m->mothurOutEndLine();
 958                         }else{
 959
 960                                 if (size == 0) { //user has not set size, set size = 10% samples size
 961                                         size = int (list->getNumSeqs() * 0.10);
 962                                 }
 963
 964                                 int thisSize = list->getNumSeqs();
 965                                 if (size > thisSize) { m->mothurOut("Your list file only contains " + toString(thisSize) + " sequences. Setting size to " + toString(thisSize) + "."); m->mothurOutEndLine();
 966                                         size = thisSize;
 967                                 }
 968
 969                                 m->mothurOut("Sampling " + toString(size) + " from " + toString(list->getNumSeqs()) + "."); m->mothurOutEndLine();
 970                         }
 971                 }
 972
 973
 974                 //fill names
 975                 for (int i = 0; i < list->getNumBins(); i++) {
 976                         string binnames = list->get(i);
 977
 978                         //parse names
 979                         string individual = "";
 980                         int length = binnames.length();
 981                         for(int j=0;j<length;j++){
 982                                 if(binnames[j] == ','){
 983
 984                                         if (groupfile != "") { //if there is a groupfile given fill in group info
 985                                                 string group = groupMap->getGroup(individual);
 986                                                 if (group == "not found") { m->mothurOut("[ERROR]: " + individual + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
 987
 988                                                 if (pickedGroups) { //if hte user picked groups, we only want to keep the names of sequences from those groups
 989                                                         if (m->inUsersGroups(group, Groups)) {
 990                                                                 names.push_back(individual);
 991                                                         }
 992                                                 }else{
 993                                                         names.push_back(individual);
 994                                                 }
 995                                         }else{ //save everyone, group
 996                                                 names.push_back(individual);
 997                                         }
 998                                         individual = "";
 999                                 }
1000                                 else{
1001                                         individual += binnames[j];
1002                                 }
1003                         }
1004                         //save last name
1005                         if (groupfile != "") { //if there is a groupfile given fill in group info
1006                                 string group = groupMap->getGroup(individual);
1007                                 if (group == "not found") { m->mothurOut("[ERROR]: " + individual + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
1008
1009                                 if (pickedGroups) { //if hte user picked groups, we only want to keep the names of sequences from those groups
1010                                         if (m->inUsersGroups(group, Groups)) {
1011                                                 names.push_back(individual);
1012                                         }
1013                                 }else{
1014                                         names.push_back(individual);
1015                                 }
1016                         }else{ //save everyone, group
1017                                 names.push_back(individual);
1018                         }
1019                 }
1020
1021                 random_shuffle(names.begin(), names.end());
1022
1023                 //randomly select a subset of those names to include in the subsample
1024                 set<string> subset; //dont want repeat sequence names added
1025                 if (persample) {
1026                         //initialize counts
1027                         map<string, int> groupCounts;
1028                         for (int i = 0; i < Groups.size(); i++) { groupCounts[Groups[i]] = 0; }
1029
1030                         for (int j = 0; j < names.size(); j++) {
1031
1032                                 if (m->control_pressed) { return 0; }
1033
1034                                 string group = groupMap->getGroup(names[j]);
1035                                 if (group == "not found") { m->mothurOut("[ERROR]: " + names[j] + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
1036                                 else{
1037                                         if (groupCounts[group] < size) {        subset.insert(names[j]);        }
1038                                 }
1039                         }
1040                 }else{
1041                         for (int j = 0; j < size; j++) {
1042
1043                                 if (m->control_pressed) { break; }
1044
1045                                 subset.insert(names[j]);
1046                         }
1047                 }
1048
1049                 if (groupfile != "") {
1050                         //write out new groupfile
1051                         for (set<string>::iterator it = subset.begin(); it != subset.end(); it++) {
1052                                 string group = groupMap->getGroup(*it);
1053                                 if (group == "not found") { group = "NOTFOUND"; }
1054
1055                                 outGroup << *it << '\t' << group << endl;
1056                         }
1057                         outGroup.close(); delete groupMap;
1058                 }
1059
1060
1061                 //as long as you are not at the end of the file or done wih the lines you want
1062                 while((list != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
1063
1064                         if (m->control_pressed) {  delete list; delete input; out.close();  return 0;  }
1065
1066                         if(allLines == 1 || labels.count(list->getLabel()) == 1){
1067
1068                                 m->mothurOut(list->getLabel()); m->mothurOutEndLine();
1069
1070                                 processList(list, out, subset);
1071
1072                                 processedLabels.insert(list->getLabel());
1073                                 userLabels.erase(list->getLabel());
1074                         }
1075
1076                         if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
1077                                 string saveLabel = list->getLabel();
1078
1079                                 delete list;
1080
1081                                 list = input->getListVector(lastLabel);
1082                                 m->mothurOut(list->getLabel()); m->mothurOutEndLine();
1083
1084                                 processList(list, out, subset);
1085
1086                                 processedLabels.insert(list->getLabel());
1087                                 userLabels.erase(list->getLabel());
1088
1089                                 //restore real lastlabel to save below
1090                                 list->setLabel(saveLabel);
1091                         }
1092
1093                         lastLabel = list->getLabel();
1094
1095                         delete list; list = NULL;
1096
1097                         //get next line to process
1098                         list = input->getListVector();
1099                 }
1100
1101
1102                 if (m->control_pressed) {  if (list != NULL) { delete list; } delete input; out.close(); return 0;  }
1103
1104                 //output error messages about any remaining user labels
1105                 set<string>::iterator it;
1106                 bool needToRun = false;
1107                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
1108                         m->mothurOut("Your file does not include the label " + *it);
1109                         if (processedLabels.count(lastLabel) != 1) {
1110                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
1111                                 needToRun = true;
1112                         }else {
1113                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
1114                         }
1115                 }
1116
1117                 //run last label if you need to
1118                 if (needToRun == true)  {
1119                         if (list != NULL) { delete list; }
1120
1121                         list = input->getListVector(lastLabel);
1122
1123                         m->mothurOut(list->getLabel()); m->mothurOutEndLine();
1124
1125                         processList(list, out, subset);
1126
1127                         delete list; list = NULL;
1128                 }
1129
1130                 out.close();
1131                 if (list != NULL) { delete list; }
1132                 delete input;
1133
1134                 return 0;
1135
1136         }
1137         catch(exception& e) {
1138                 m->errorOut(e, "SubSampleCommand", "getSubSampleList");
1139                 exit(1);
1140         }
1141 }
1142 //**********************************************************************************************************************
1143 int SubSampleCommand::processList(ListVector*& list, ofstream& out, set<string>& subset) {
1144         try {
1145
1146                 int numBins = list->getNumBins();
1147
1148                 ListVector* temp = new ListVector();
1149                 temp->setLabel(list->getLabel());
1150
1151                 for (int i = 0; i < numBins; i++) {
1152
1153                         if (m->control_pressed) { break; }
1154
1155                         string binnames = list->get(i);
1156
1157                         //parse names
1158                         string individual = "";
1159                         string newNames = "";
1160                         int length = binnames.length();
1161                         for(int j=0;j<length;j++){
1162                                 if(binnames[j] == ','){
1163                                         if (subset.count(individual) != 0) {  newNames += individual + ",";  }
1164                                         individual = "";
1165                                 }else{
1166                                         individual += binnames[j];
1167                                 }
1168                         }
1169                         if (subset.count(individual) != 0) {  newNames += individual + ",";  }
1170
1171
1172                         //if there are names in this bin add to new list
1173                         if (newNames != "") {
1174                                 newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
1175                                 temp->push_back(newNames);
1176                         }
1177                 }
1178
1179                 delete list;
1180                 list = temp;
1181
1182                 if (m->control_pressed) { return 0; }
1183
1184                 list->print(out);
1185
1186                 return 0;
1187
1188         }
1189         catch(exception& e) {
1190                 m->errorOut(e, "SubSampleCommand", "processList");
1191                 exit(1);
1192         }
1193 }
1194 //**********************************************************************************************************************
1195 int SubSampleCommand::getSubSampleRabund() {
1196         try {
1197                 InputData* input = new InputData(rabundfile, "rabund");
1198                 RAbundVector* rabund = input->getRAbundVector();
1199                 string lastLabel = rabund->getLabel();
1200
1201                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
1202                 set<string> processedLabels;
1203                 set<string> userLabels = labels;
1204
1205                 if (size == 0) { //user has not set size, set size = 10%
1206                         size = int((rabund->getNumSeqs()) * 0.10);
1207                 }else if (size > rabund->getNumSeqs()) { m->mothurOut("The size you selected is too large, skipping rabund file."); m->mothurOutEndLine(); delete input; delete rabund; return 0; }
1208
1209                 m->mothurOut("Sampling " + toString(size) + " from " + toString(rabund->getNumSeqs()) + "."); m->mothurOutEndLine();
1210
1211                 string thisOutputDir = outputDir;
1212                 if (outputDir == "") {  thisOutputDir += m->hasPath(rabundfile);  }
1213                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(rabundfile)) + "subsample" + m->getExtension(rabundfile);
1214
1215                 ofstream out;
1216                 m->openOutputFile(outputFileName, out);
1217                 outputTypes["rabund"].push_back(outputFileName);  outputNames.push_back(outputFileName);
1218
1219                 //as long as you are not at the end of the file or done wih the lines you want
1220                 while((rabund != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
1221                         if (m->control_pressed) {  delete input; delete rabund; out.close(); return 0;  }
1222
1223                         if(allLines == 1 || labels.count(rabund->getLabel()) == 1){
1224
1225                                 m->mothurOut(rabund->getLabel()); m->mothurOutEndLine();
1226
1227                                 processRabund(rabund, out);
1228
1229                                 processedLabels.insert(rabund->getLabel());
1230                                 userLabels.erase(rabund->getLabel());
1231                         }
1232
1233                         if ((m->anyLabelsToProcess(rabund->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
1234                                 string saveLabel = rabund->getLabel();
1235
1236                                 delete rabund;
1237
1238                                 rabund = input->getRAbundVector(lastLabel);
1239                                 m->mothurOut(rabund->getLabel()); m->mothurOutEndLine();
1240
1241                                 processRabund(rabund, out);
1242
1243                                 processedLabels.insert(rabund->getLabel());
1244                                 userLabels.erase(rabund->getLabel());
1245
1246                                 //restore real lastlabel to save below
1247                                 rabund->setLabel(saveLabel);
1248                         }
1249
1250                         lastLabel = rabund->getLabel();
1251
1252                         //prevent memory leak
1253                         delete rabund; rabund = NULL;
1254
1255                         //get next line to process
1256                         rabund = input->getRAbundVector();
1257                 }
1258
1259
1260                 if (m->control_pressed) {  out.close(); return 0;  }
1261
1262                 //output error messages about any remaining user labels
1263                 set<string>::iterator it;
1264                 bool needToRun = false;
1265                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
1266                         m->mothurOut("Your file does not include the label " + *it);
1267                         if (processedLabels.count(lastLabel) != 1) {
1268                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
1269                                 needToRun = true;
1270                         }else {
1271                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
1272                         }
1273                 }
1274
1275                 //run last label if you need to
1276                 if (needToRun == true)  {
1277                         if (rabund != NULL) { delete rabund; }
1278
1279                         rabund = input->getRAbundVector(lastLabel);
1280
1281                         m->mothurOut(rabund->getLabel()); m->mothurOutEndLine();
1282
1283                         processRabund(rabund, out);
1284
1285                         delete rabund;
1286                 }
1287
1288                 delete input;
1289                 out.close();
1290
1291                 return 0;
1292
1293         }
1294         catch(exception& e) {
1295                 m->errorOut(e, "SubSampleCommand", "getSubSampleRabund");
1296                 exit(1);
1297         }
1298 }
1299 //**********************************************************************************************************************
1300 int SubSampleCommand::processRabund(RAbundVector*& rabund, ofstream& out) {
1301         try {
1302
1303                 int numBins = rabund->getNumBins();
1304                 int thisSize = rabund->getNumSeqs();
1305
1306                 if (thisSize != size) {
1307
1308                         OrderVector* order = new OrderVector();
1309                         for(int p=0;p<numBins;p++){
1310                                 for(int j=0;j<rabund->get(p);j++){
1311                                         order->push_back(p);
1312                                 }
1313                         }
1314                         random_shuffle(order->begin(), order->end());
1315
1316                         RAbundVector* temp = new RAbundVector(numBins);
1317                         temp->setLabel(rabund->getLabel());
1318
1319                         delete rabund;
1320                         rabund = temp;
1321
1322                         for (int j = 0; j < size; j++) {
1323
1324                                 if (m->control_pressed) { delete order; return 0; }
1325
1326                                 int bin = order->get(j);
1327
1328                                 int abund = rabund->get(bin);
1329                                 rabund->set(bin, (abund+1));
1330                         }
1331
1332                         delete order;
1333                 }
1334
1335                 if (m->control_pressed) { return 0; }
1336
1337                 rabund->print(out);
1338
1339                 return 0;
1340
1341         }
1342         catch(exception& e) {
1343                 m->errorOut(e, "SubSampleCommand", "processRabund");
1344                 exit(1);
1345         }
1346 }
1347 //**********************************************************************************************************************
1348 int SubSampleCommand::getSubSampleSabund() {
1349         try {
1350
1351                 InputData* input = new InputData(sabundfile, "sabund");
1352                 SAbundVector* sabund = input->getSAbundVector();
1353                 string lastLabel = sabund->getLabel();
1354
1355                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
1356                 set<string> processedLabels;
1357                 set<string> userLabels = labels;
1358
1359                 if (size == 0) { //user has not set size, set size = 10%
1360                         size = int((sabund->getNumSeqs()) * 0.10);
1361                 }else if (size > sabund->getNumSeqs()) { m->mothurOut("The size you selected is too large, skipping sabund file."); m->mothurOutEndLine(); delete input; delete sabund; return 0; }
1362
1363
1364                 m->mothurOut("Sampling " + toString(size) + " from " + toString(sabund->getNumSeqs()) + "."); m->mothurOutEndLine();
1365
1366                 string thisOutputDir = outputDir;
1367                 if (outputDir == "") {  thisOutputDir += m->hasPath(sabundfile);  }
1368                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(sabundfile)) + "subsample" + m->getExtension(sabundfile);
1369
1370                 ofstream out;
1371                 m->openOutputFile(outputFileName, out);
1372                 outputTypes["sabund"].push_back(outputFileName);  outputNames.push_back(outputFileName);
1373
1374
1375                 //as long as you are not at the end of the file or done wih the lines you want
1376                 while((sabund != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
1377                         if (m->control_pressed) {  delete input; delete sabund; out.close(); return 0;  }
1378
1379                         if(allLines == 1 || labels.count(sabund->getLabel()) == 1){
1380
1381                                 m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
1382
1383                                 processSabund(sabund, out);
1384
1385                                 processedLabels.insert(sabund->getLabel());
1386                                 userLabels.erase(sabund->getLabel());
1387                         }
1388
1389                         if ((m->anyLabelsToProcess(sabund->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
1390                                 string saveLabel = sabund->getLabel();
1391
1392                                 delete sabund;
1393
1394                                 sabund = input->getSAbundVector(lastLabel);
1395                                 m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
1396
1397                                 processSabund(sabund, out);
1398
1399                                 processedLabels.insert(sabund->getLabel());
1400                                 userLabels.erase(sabund->getLabel());
1401
1402                                 //restore real lastlabel to save below
1403                                 sabund->setLabel(saveLabel);
1404                         }
1405
1406                         lastLabel = sabund->getLabel();
1407
1408                         //prevent memory leak
1409                         delete sabund; sabund = NULL;
1410
1411                         //get next line to process
1412                         sabund = input->getSAbundVector();
1413                 }
1414
1415
1416                 if (m->control_pressed) {  out.close(); return 0;  }
1417
1418                 //output error messages about any remaining user labels
1419                 set<string>::iterator it;
1420                 bool needToRun = false;
1421                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
1422                         m->mothurOut("Your file does not include the label " + *it);
1423                         if (processedLabels.count(lastLabel) != 1) {
1424                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
1425                                 needToRun = true;
1426                         }else {
1427                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
1428                         }
1429                 }
1430
1431                 //run last label if you need to
1432                 if (needToRun == true)  {
1433                         if (sabund != NULL) { delete sabund; }
1434
1435                         sabund = input->getSAbundVector(lastLabel);
1436
1437                         m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
1438
1439                         processSabund(sabund, out);
1440
1441                         delete sabund;
1442                 }
1443
1444                 delete input;
1445                 out.close();
1446
1447                 return 0;
1448
1449         }
1450         catch(exception& e) {
1451                 m->errorOut(e, "SubSampleCommand", "getSubSampleSabund");
1452                 exit(1);
1453         }
1454 }
1455 //**********************************************************************************************************************
1456 int SubSampleCommand::processSabund(SAbundVector*& sabund, ofstream& out) {
1457         try {
1458
1459                 RAbundVector* rabund = new RAbundVector();
1460                 *rabund = sabund->getRAbundVector();
1461
1462                 int numBins = rabund->getNumBins();
1463                 int thisSize = rabund->getNumSeqs();
1464
1465                 if (thisSize != size) {
1466
1467                         OrderVector* order = new OrderVector();
1468                         for(int p=0;p<numBins;p++){
1469                                 for(int j=0;j<rabund->get(p);j++){
1470                                         order->push_back(p);
1471                                 }
1472                         }
1473                         random_shuffle(order->begin(), order->end());
1474
1475                         RAbundVector* temp = new RAbundVector(numBins);
1476                         temp->setLabel(rabund->getLabel());
1477
1478                         delete rabund;
1479                         rabund = temp;
1480
1481                         for (int j = 0; j < size; j++) {
1482
1483                                 if (m->control_pressed) { delete order; return 0; }
1484
1485                                 int bin = order->get(j);
1486
1487                                 int abund = rabund->get(bin);
1488                                 rabund->set(bin, (abund+1));
1489                         }
1490
1491                         delete order;
1492                 }
1493
1494                 if (m->control_pressed) { return 0; }
1495
1496                 delete sabund;
1497                 sabund = new SAbundVector();
1498                 *sabund = rabund->getSAbundVector();
1499                 delete rabund;
1500
1501                 sabund->print(out);
1502
1503                 return 0;
1504
1505         }
1506         catch(exception& e) {
1507                 m->errorOut(e, "SubSampleCommand", "processSabund");
1508                 exit(1);
1509         }
1510 }
1511 //**********************************************************************************************************************
1512 int SubSampleCommand::eliminateZeroOTUS(vector<SharedRAbundVector*>& thislookup) {
1513         try {
1514
1515                 vector<SharedRAbundVector*> newLookup;
1516                 for (int i = 0; i < thislookup.size(); i++) {
1517                         SharedRAbundVector* temp = new SharedRAbundVector();
1518                         temp->setLabel(thislookup[i]->getLabel());
1519                         temp->setGroup(thislookup[i]->getGroup());
1520                         newLookup.push_back(temp);
1521                 }
1522
1523                 //for each bin
1524                 vector<string> newBinLabels;
1525                 string snumBins = toString(thislookup[0]->getNumBins());
1526                 for (int i = 0; i < thislookup[0]->getNumBins(); i++) {
1527                         if (m->control_pressed) { for (int j = 0; j < newLookup.size(); j++) {  delete newLookup[j];  } return 0; }
1528
1529                         //look at each sharedRabund and make sure they are not all zero
1530                         bool allZero = true;
1531                         for (int j = 0; j < thislookup.size(); j++) {
1532                                 if (thislookup[j]->getAbundance(i) != 0) { allZero = false;  break;  }
1533                         }
1534
1535                         //if they are not all zero add this bin
1536                         if (!allZero) {
1537                                 for (int j = 0; j < thislookup.size(); j++) {
1538                                         newLookup[j]->push_back(thislookup[j]->getAbundance(i), thislookup[j]->getGroup());
1539                                 }
1540                                 //if there is a bin label use it otherwise make one
1541                                 string binLabel = "Otu";
1542                                 string sbinNumber = toString(i+1);
1543                                 if (sbinNumber.length() < snumBins.length()) {
1544                                         int diff = snumBins.length() - sbinNumber.length();
1545                                         for (int h = 0; h < diff; h++) { binLabel += "0"; }
1546                                 }
1547                                 binLabel += sbinNumber;
1548                                 if (i < m->currentBinLabels.size()) {  binLabel = m->currentBinLabels[i]; }
1549
1550                                 newBinLabels.push_back(binLabel);
1551                         }
1552                 }
1553
1554                 for (int j = 0; j < thislookup.size(); j++) {  delete thislookup[j];  }
1555                 thislookup.clear();
1556
1557                 thislookup = newLookup;
1558                 m->currentBinLabels = newBinLabels;
1559
1560                 return 0;
1561
1562         }
1563         catch(exception& e) {
1564                 m->errorOut(e, "SubSampleCommand", "eliminateZeroOTUS");
1565                 exit(1);
1566         }
1567 }
1568
1569 //**********************************************************************************************************************
1570
1571
1572