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
  13 //**********************************************************************************************************************
  14 vector<string> SubSampleCommand::setParameters(){
  15         try {
  16                 CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "FLSSR", "none",false,false); parameters.push_back(pfasta);
  17                 CommandParameter pname("name", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pname);
  18                 CommandParameter pgroup("group", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pgroup);
  19                 CommandParameter plist("list", "InputTypes", "", "", "none", "FLSSR", "none",false,false); parameters.push_back(plist);
  20                 CommandParameter pshared("shared", "InputTypes", "", "", "none", "FLSSR", "none",false,false); parameters.push_back(pshared);
  21                 CommandParameter prabund("rabund", "InputTypes", "", "", "none", "FLSSR", "none",false,false); parameters.push_back(prabund);
  22                 CommandParameter psabund("sabund", "InputTypes", "", "", "none", "FLSSR", "none",false,false); parameters.push_back(psabund);
  23                 CommandParameter plabel("label", "String", "", "", "", "", "",false,false); parameters.push_back(plabel);
  24                 CommandParameter pgroups("groups", "String", "", "", "", "", "",false,false); parameters.push_back(pgroups);
  25                 CommandParameter psize("size", "Number", "", "0", "", "", "",false,false); parameters.push_back(psize);
  26                 CommandParameter ppersample("persample", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(ppersample);
  27                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  28                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  29
  30                 vector<string> myArray;
  31                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  32                 return myArray;
  33         }
  34         catch(exception& e) {
  35                 m->errorOut(e, "SubSampleCommand", "setParameters");
  36                 exit(1);
  37         }
  38 }
  39 //**********************************************************************************************************************
  40 string SubSampleCommand::getHelpString(){
  41         try {
  42                 string helpString = "";
  43                 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";
  44                 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";
  45                 helpString += "The namefile is only used with the fasta file, not with the listfile, because the list file should contain all sequences.\n";
  46                 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";
  47                 helpString += "The label parameter allows you to select what distance levels you would like, and are also separated by dashes.\n";
  48                 helpString += "The size parameter allows you indicate the size of your subsample.\n";
  49                 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";
  50                 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";
  51                 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";
  52                 helpString += "The sub.sample command should be in the following format: sub.sample(list=yourListFile, group=yourGroupFile, groups=yourGroups, label=yourLabels).\n";
  53                 helpString += "Example sub.sample(list=abrecovery.fn.list, group=abrecovery.groups, groups=B-C, size=20).\n";
  54                 helpString += "The default value for groups is all the groups in your groupfile, and all labels in your inputfile will be used.\n";
  55                 helpString += "The sub.sample command outputs a .subsample file.\n";
  56                 helpString += "Note: No spaces between parameter labels (i.e. groups), '=' and parameters (i.e.yourGroups).\n";
  57                 return helpString;
  58         }
  59         catch(exception& e) {
  60                 m->errorOut(e, "SubSampleCommand", "getHelpString");
  61                 exit(1);
  62         }
  63 }
  64 //**********************************************************************************************************************
  65 SubSampleCommand::SubSampleCommand(){
  66         try {
  67                 abort = true; calledHelp = true;
  68                 setParameters();
  69                 vector<string> tempOutNames;
  70                 outputTypes["shared"] = tempOutNames;
  71                 outputTypes["list"] = tempOutNames;
  72                 outputTypes["rabund"] = tempOutNames;
  73                 outputTypes["sabund"] = tempOutNames;
  74                 outputTypes["fasta"] = tempOutNames;
  75                 outputTypes["name"] = tempOutNames;
  76                 outputTypes["group"] = tempOutNames;
  77         }
  78         catch(exception& e) {
  79                 m->errorOut(e, "SubSampleCommand", "GetRelAbundCommand");
  80                 exit(1);
  81         }
  82 }
  83 //**********************************************************************************************************************
  84 SubSampleCommand::SubSampleCommand(string option) {
  85         try {
  86                 abort = false; calledHelp = false;
  87                 allLines = 1;
  88
  89                 //allow user to run help
  90                 if(option == "help") { help(); abort = true; calledHelp = true; }
  91                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  92
  93                 else {
  94                         vector<string> myArray = setParameters();
  95
  96                         OptionParser parser(option);
  97                         map<string,string> parameters = parser.getParameters();
  98
  99                         ValidParameters validParameter;
 100
 101                         //check to make sure all parameters are valid for command
 102                         map<string,string>::iterator it;
 103                         for (it = parameters.begin(); it != parameters.end(); it++) {
 104                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 105                         }
 106
 107                         //initialize outputTypes
 108                         vector<string> tempOutNames;
 109                         outputTypes["shared"] = tempOutNames;
 110                         outputTypes["list"] = tempOutNames;
 111                         outputTypes["rabund"] = tempOutNames;
 112                         outputTypes["sabund"] = tempOutNames;
 113                         outputTypes["fasta"] = tempOutNames;
 114                         outputTypes["name"] = tempOutNames;
 115                         outputTypes["group"] = tempOutNames;
 116
 117                         //if the user changes the output directory command factory will send this info to us in the output parameter
 118                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
 119
 120                         //if the user changes the input directory command factory will send this info to us in the output parameter
 121                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 122                         if (inputDir == "not found"){   inputDir = "";          }
 123                         else {
 124                                 string path;
 125                                 it = parameters.find("list");
 126                                 //user has given a template file
 127                                 if(it != parameters.end()){
 128                                         path = m->hasPath(it->second);
 129                                         //if the user has not given a path then, add inputdir. else leave path alone.
 130                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 131                                 }
 132
 133                                 it = parameters.find("fasta");
 134                                 //user has given a template file
 135                                 if(it != parameters.end()){
 136                                         path = m->hasPath(it->second);
 137                                         //if the user has not given a path then, add inputdir. else leave path alone.
 138                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 139                                 }
 140
 141                                 it = parameters.find("shared");
 142                                 //user has given a template file
 143                                 if(it != parameters.end()){
 144                                         path = m->hasPath(it->second);
 145                                         //if the user has not given a path then, add inputdir. else leave path alone.
 146                                         if (path == "") {       parameters["shared"] = inputDir + it->second;           }
 147                                 }
 148
 149                                 it = parameters.find("group");
 150                                 //user has given a template file
 151                                 if(it != parameters.end()){
 152                                         path = m->hasPath(it->second);
 153                                         //if the user has not given a path then, add inputdir. else leave path alone.
 154                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 155                                 }
 156
 157                                 it = parameters.find("sabund");
 158                                 //user has given a template file
 159                                 if(it != parameters.end()){
 160                                         path = m->hasPath(it->second);
 161                                         //if the user has not given a path then, add inputdir. else leave path alone.
 162                                         if (path == "") {       parameters["sabund"] = inputDir + it->second;           }
 163                                 }
 164
 165                                 it = parameters.find("rabund");
 166                                 //user has given a template file
 167                                 if(it != parameters.end()){
 168                                         path = m->hasPath(it->second);
 169                                         //if the user has not given a path then, add inputdir. else leave path alone.
 170                                         if (path == "") {       parameters["rabund"] = inputDir + it->second;           }
 171                                 }
 172
 173                                 it = parameters.find("name");
 174                                 //user has given a template file
 175                                 if(it != parameters.end()){
 176                                         path = m->hasPath(it->second);
 177                                         //if the user has not given a path then, add inputdir. else leave path alone.
 178                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 179                                 }
 180                         }
 181
 182                         //check for required parameters
 183                         listfile = validParameter.validFile(parameters, "list", true);
 184                         if (listfile == "not open") { listfile = ""; abort = true; }
 185                         else if (listfile == "not found") { listfile = ""; }
 186
 187                         sabundfile = validParameter.validFile(parameters, "sabund", true);
 188                         if (sabundfile == "not open") { sabundfile = ""; abort = true; }
 189                         else if (sabundfile == "not found") { sabundfile = ""; }
 190
 191                         rabundfile = validParameter.validFile(parameters, "rabund", true);
 192                         if (rabundfile == "not open") { rabundfile = ""; abort = true; }
 193                         else if (rabundfile == "not found") { rabundfile = ""; }
 194
 195                         fastafile = validParameter.validFile(parameters, "fasta", true);
 196                         if (fastafile == "not open") { fastafile = ""; abort = true; }
 197                         else if (fastafile == "not found") { fastafile = ""; }
 198
 199                         sharedfile = validParameter.validFile(parameters, "shared", true);
 200                         if (sharedfile == "not open") { sharedfile = ""; abort = true; }
 201                         else if (sharedfile == "not found") { sharedfile = ""; }
 202
 203                         namefile = validParameter.validFile(parameters, "name", true);
 204                         if (namefile == "not open") { namefile = ""; abort = true; }
 205                         else if (namefile == "not found") { namefile = ""; }
 206
 207                         groupfile = validParameter.validFile(parameters, "group", true);
 208                         if (groupfile == "not open") { groupfile = ""; abort = true; }
 209                         else if (groupfile == "not found") { groupfile = ""; }
 210
 211
 212                         //check for optional parameter and set defaults
 213                         // ...at some point should added some additional type checking...
 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                         groups = validParameter.validFile(parameters, "groups", false);
 222                         if (groups == "not found") { groups = ""; pickedGroups = false; }
 223                         else {
 224                                 pickedGroups = true;
 225                                 m->splitAtDash(groups, Groups);
 226                                 m->Groups = Groups;
 227                         }
 228
 229                         string temp = validParameter.validFile(parameters, "size", false);              if (temp == "not found"){       temp = "0";             }
 230                         convert(temp, size);
 231
 232                         temp = validParameter.validFile(parameters, "persample", false);                if (temp == "not found"){       temp = "f";             }
 233                         persample = m->isTrue(temp);
 234
 235                         if (groupfile == "") { persample = false; }
 236
 237                         if ((namefile != "") && (fastafile == "")) { m->mothurOut("You may only use a namefile with a fastafile."); m->mothurOutEndLine(); abort = true; }
 238
 239                         if ((fastafile == "") && (listfile == "") && (sabundfile == "") && (rabundfile == "") && (sharedfile == "")) {
 240                                 m->mothurOut("You must provide a fasta, list, sabund, rabund or shared file as an input file."); m->mothurOutEndLine(); abort = true; }
 241
 242                         if (pickedGroups && ((groupfile == "") && (sharedfile == ""))) {
 243                                 m->mothurOut("You cannot pick groups without a valid group file or shared file."); m->mothurOutEndLine(); abort = true; }
 244
 245                         if ((groupfile != "") && ((fastafile == "") && (listfile == ""))) {
 246                                 m->mothurOut("Group file only valid with listfile or fastafile."); m->mothurOutEndLine(); abort = true; }
 247
 248                         if ((groupfile != "") && ((fastafile != "") && (listfile != ""))) {
 249                                 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; }
 250
 251                 }
 252
 253         }
 254         catch(exception& e) {
 255                 m->errorOut(e, "SubSampleCommand", "SubSampleCommand");
 256                 exit(1);
 257         }
 258 }
 259 //**********************************************************************************************************************
 260
 261 int SubSampleCommand::execute(){
 262         try {
 263
 264                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 265
 266                 if (sharedfile != "")   {   getSubSampleShared();       }
 267                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       remove(outputNames[i].c_str()); return 0; } }
 268
 269                 if (listfile != "")             {   getSubSampleList();         }
 270                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       remove(outputNames[i].c_str()); return 0; } }
 271
 272                 if (rabundfile != "")   {   getSubSampleRabund();       }
 273                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       remove(outputNames[i].c_str()); return 0; } }
 274
 275                 if (sabundfile != "")   {   getSubSampleSabund();       }
 276                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       remove(outputNames[i].c_str()); return 0; } }
 277
 278                 if (fastafile != "")    {   getSubSampleFasta();        }
 279                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       remove(outputNames[i].c_str()); return 0; } }
 280
 281                 //set fasta file as new current fastafile
 282                 string current = "";
 283                 itTypes = outputTypes.find("fasta");
 284                 if (itTypes != outputTypes.end()) {
 285                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 286                 }
 287
 288                 itTypes = outputTypes.find("name");
 289                 if (itTypes != outputTypes.end()) {
 290                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
 291                 }
 292
 293                 itTypes = outputTypes.find("group");
 294                 if (itTypes != outputTypes.end()) {
 295                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setGroupFile(current); }
 296                 }
 297
 298                 itTypes = outputTypes.find("list");
 299                 if (itTypes != outputTypes.end()) {
 300                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 301                 }
 302
 303                 itTypes = outputTypes.find("shared");
 304                 if (itTypes != outputTypes.end()) {
 305                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setSharedFile(current); }
 306                 }
 307
 308                 itTypes = outputTypes.find("rabund");
 309                 if (itTypes != outputTypes.end()) {
 310                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setRabundFile(current); }
 311                 }
 312
 313                 itTypes = outputTypes.find("sabund");
 314                 if (itTypes != outputTypes.end()) {
 315                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setSabundFile(current); }
 316                 }
 317
 318
 319                 m->mothurOutEndLine();
 320                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 321                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 322                 m->mothurOutEndLine();
 323
 324                 return 0;
 325         }
 326         catch(exception& e) {
 327                 m->errorOut(e, "SubSampleCommand", "execute");
 328                 exit(1);
 329         }
 330 }
 331 //**********************************************************************************************************************
 332 int SubSampleCommand::getSubSampleFasta() {
 333         try {
 334
 335                 if (namefile != "") { readNames(); }    //fills names with all names in namefile.
 336                 else { getNames(); }//no name file, so get list of names to pick from
 337
 338                 GroupMap* groupMap;
 339                 if (groupfile != "") {
 340
 341                         groupMap = new GroupMap(groupfile);
 342                         groupMap->readMap();
 343
 344                         //takes care of user setting groupNames that are invalid or setting groups=all
 345                         SharedUtil* util = new SharedUtil();
 346                         util->setGroups(Groups, groupMap->namesOfGroups);
 347                         delete util;
 348
 349                         //file mismatch quit
 350                         if (names.size() != groupMap->getNumSeqs()) {
 351                                 m->mothurOut("[ERROR]: your fasta file contains " + toString(names.size()) + " sequences, and your groupfile contains " + toString(groupMap->getNumSeqs()) + ", please correct.");
 352                                 m->mothurOutEndLine();
 353                                 delete groupMap;
 354                                 return 0;
 355                         }
 356                 }
 357
 358                 if (m->control_pressed) { return 0; }
 359
 360
 361                 //make sure that if your picked groups size is not too big
 362                 int thisSize = names.size();
 363                 if (persample) {
 364                         if (size == 0) { //user has not set size, set size = smallest samples size
 365                                 size = groupMap->getNumSeqs(Groups[0]);
 366                                 for (int i = 1; i < Groups.size(); i++) {
 367                                         int thisSize = groupMap->getNumSeqs(Groups[i]);
 368
 369                                         if (thisSize < size) {  size = thisSize;        }
 370                                 }
 371                         }else { //make sure size is not too large
 372                                 vector<string> newGroups;
 373                                 for (int i = 0; i < Groups.size(); i++) {
 374                                         int thisSize = groupMap->getNumSeqs(Groups[i]);
 375
 376                                         if (thisSize >= size) { newGroups.push_back(Groups[i]); }
 377                                         else {  m->mothurOut("You have selected a size that is larger than " + Groups[i] + " number of sequences, removing " + Groups[i] + "."); m->mothurOutEndLine(); }
 378                                 }
 379                                 Groups = newGroups;
 380                         }
 381
 382                         m->mothurOut("Sampling " + toString(size) + " from each group."); m->mothurOutEndLine();
 383                 }else {
 384                         if (pickedGroups) {
 385                                 int total = 0;
 386                                 for(int i = 0; i < Groups.size(); i++) {
 387                                         total += groupMap->getNumSeqs(Groups[i]);
 388                                 }
 389
 390                                 if (size == 0) { //user has not set size, set size = 10% samples size
 391                                         size = int (total * 0.10);
 392                                 }
 393
 394                                 if (total < size) {
 395                                         if (size != 0) {
 396                                                 m->mothurOut("Your size is too large for the number of groups you selected. Adjusting to " + toString(int (total * 0.10)) + "."); m->mothurOutEndLine();
 397                                         }
 398                                         size = int (total * 0.10);
 399                                 }
 400
 401                                 m->mothurOut("Sampling " + toString(size) + " from " + toString(total) + "."); m->mothurOutEndLine();
 402                         }
 403
 404                         if (size == 0) { //user has not set size, set size = 10% samples size
 405                                 size = int (names.size() * 0.10);
 406                         }
 407
 408                         if (size > thisSize) { m->mothurOut("Your fasta file only contains " + toString(thisSize) + " sequences. Setting size to " + toString(thisSize) + "."); m->mothurOutEndLine();
 409                                 size = thisSize;
 410                         }
 411
 412                         if (!pickedGroups) { m->mothurOut("Sampling " + toString(size) + " from " + toString(thisSize) + "."); m->mothurOutEndLine(); }
 413
 414                 }
 415                 random_shuffle(names.begin(), names.end());
 416
 417                 set<string> subset; //dont want repeat sequence names added
 418                 if (persample) {
 419                         for (int i = 0; i < Groups.size(); i++) {
 420
 421                                 //randomly select a subset of those names from this group to include in the subsample
 422                                 for (int j = 0; j < size; j++) {
 423
 424                                         if (m->control_pressed) { return 0; }
 425
 426                                         //get random sequence to add, making sure we have not already added it
 427                                         bool done = false;
 428                                         int myrand;
 429                                         while (!done) {
 430                                                 myrand = int((float)(thisSize) * (float)(rand()) / ((float)RAND_MAX+1.0));
 431
 432                                                 if (subset.count(names[myrand]) == 0)  {
 433
 434                                                         string group = groupMap->getGroup(names[myrand]);
 435                                                         if (group == "not found") { m->mothurOut("[ERROR]: " + names[myrand] + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
 436
 437                                                         if (group == Groups[i]) {       subset.insert(names[myrand]); break;    }
 438                                                 }
 439                                         }
 440                                 }
 441                         }
 442                 }else {
 443
 444                         //randomly select a subset of those names to include in the subsample
 445                         for (int j = 0; j < size; j++) {
 446
 447                                 if (m->control_pressed) { return 0; }
 448
 449                                 //get random sequence to add, making sure we have not already added it
 450                                 bool done = false;
 451                                 int myrand;
 452                                 while (!done) {
 453                                         myrand = int((float)(thisSize) * (float)(rand()) / ((float)RAND_MAX+1.0));
 454
 455                                         if (subset.count(names[myrand]) == 0)  {
 456
 457                                                 if (groupfile != "") { //if there is a groupfile given fill in group info
 458                                                         string group = groupMap->getGroup(names[myrand]);
 459                                                         if (group == "not found") { m->mothurOut("[ERROR]: " + names[myrand] + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
 460
 461                                                         if (pickedGroups) { //if hte user picked groups, we only want to keep the names of sequences from those groups
 462                                                                 if (m->inUsersGroups(group, Groups)) {
 463                                                                         subset.insert(names[myrand]); break;
 464                                                                 }
 465                                                         }else{
 466                                                                 subset.insert(names[myrand]); break;
 467                                                         }
 468                                                 }else{ //save everyone, group
 469                                                         subset.insert(names[myrand]); break;
 470                                                 }
 471                                         }
 472                                 }
 473                         }
 474                 }
 475
 476                 if (subset.size() == 0) {  m->mothurOut("The size you selected is too large, skipping fasta file."); m->mothurOutEndLine();  return 0; }
 477
 478                 string thisOutputDir = outputDir;
 479                 if (outputDir == "") {  thisOutputDir += m->hasPath(fastafile);  }
 480                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(fastafile)) + "subsample" + m->getExtension(fastafile);
 481
 482                 ofstream out;
 483                 m->openOutputFile(outputFileName, out);
 484                 outputTypes["fasta"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 485
 486                 //read through fasta file outputting only the names on the subsample list
 487                 ifstream in;
 488                 m->openInputFile(fastafile, in);
 489
 490                 string thisname;
 491                 int count = 0;
 492                 map<string, vector<string> >::iterator itNameMap;
 493
 494                 while(!in.eof()){
 495
 496                         if (m->control_pressed) { in.close(); out.close();  return 0; }
 497
 498                         Sequence currSeq(in);
 499                         thisname = currSeq.getName();
 500
 501                         if (thisname != "") {
 502
 503                                 //does the subset contain a sequence that this sequence represents
 504                                 itNameMap = nameMap.find(thisname);
 505                                 if (itNameMap != nameMap.end()) {
 506                                         vector<string> nameRepresents = itNameMap->second;
 507
 508                                         for (int i = 0; i < nameRepresents.size(); i++){
 509                                                 if (subset.count(nameRepresents[i]) != 0) {
 510                                                         out << ">" << nameRepresents[i] << endl << currSeq.getAligned() << endl;
 511                                                         count++;
 512                                                 }
 513                                         }
 514                                 }else{
 515                                         m->mothurOut("[ERROR]: " + thisname + " is not in your namefile, please correct."); m->mothurOutEndLine();
 516                                 }
 517                         }
 518                         m->gobble(in);
 519                 }
 520                 in.close();
 521                 out.close();
 522
 523                 if (count != subset.size()) {
 524                         m->mothurOut("[ERROR]: The subset selected contained " + toString(subset.size()) + " sequences, but I only found " + toString(count) + " of those in the fastafile."); m->mothurOutEndLine();
 525                 }
 526
 527                 //if a groupfile is provided read through the group file only outputting the names on the subsample list
 528                 if (groupfile != "") {
 529
 530                         string groupOutputDir = outputDir;
 531                         if (outputDir == "") {  groupOutputDir += m->hasPath(groupfile);  }
 532                         string groupOutputFileName = groupOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "subsample" + m->getExtension(groupfile);
 533
 534                         ofstream outGroup;
 535                         m->openOutputFile(groupOutputFileName, outGroup);
 536                         outputTypes["group"].push_back(groupOutputFileName);  outputNames.push_back(groupOutputFileName);
 537
 538                         ifstream inGroup;
 539                         m->openInputFile(groupfile, inGroup);
 540                         string name, group;
 541
 542                         while(!inGroup.eof()){
 543
 544                                 if (m->control_pressed) { inGroup.close(); outGroup.close(); return 0; }
 545
 546                                 inGroup >> name;        m->gobble(inGroup);                     //read from first column
 547                                 inGroup >> group;                       //read from second column
 548
 549                                 //if this name is in the accnos file
 550                                 if (subset.count(name) != 0) {
 551                                         outGroup << name << '\t' << group << endl;
 552                                         subset.erase(name);
 553                                 }
 554
 555                                 m->gobble(inGroup);
 556                         }
 557                         inGroup.close();
 558                         outGroup.close();
 559
 560                         //sanity check
 561                         if (subset.size() != 0) {
 562                                 m->mothurOut("Your groupfile does not match your fasta file."); m->mothurOutEndLine();
 563                                 for (set<string>::iterator it = subset.begin(); it != subset.end(); it++) {
 564                                         m->mothurOut("[ERROR]: " + *it + " is missing from your groupfile."); m->mothurOutEndLine();
 565                                 }
 566                         }
 567                 }
 568
 569                 return 0;
 570
 571         }
 572         catch(exception& e) {
 573                 m->errorOut(e, "SubSampleCommand", "getSubSampleFasta");
 574                 exit(1);
 575         }
 576 }
 577 //**********************************************************************************************************************
 578 int SubSampleCommand::getNames() {
 579         try {
 580
 581                 ifstream in;
 582                 m->openInputFile(fastafile, in);
 583
 584                 string thisname;
 585                 while(!in.eof()){
 586
 587                         if (m->control_pressed) { in.close(); return 0; }
 588
 589                         Sequence currSeq(in);
 590                         thisname = currSeq.getName();
 591
 592                         if (thisname != "") {
 593                                 vector<string> temp; temp.push_back(thisname);
 594                                 nameMap[thisname] = temp;
 595                                 names.push_back(thisname);
 596                         }
 597                         m->gobble(in);
 598                 }
 599                 in.close();
 600
 601                 return 0;
 602
 603         }
 604         catch(exception& e) {
 605                 m->errorOut(e, "SubSampleCommand", "getNames");
 606                 exit(1);
 607         }
 608 }
 609 //**********************************************************************************************************************
 610 int SubSampleCommand::readNames() {
 611         try {
 612
 613                 ifstream in;
 614                 m->openInputFile(namefile, in);
 615
 616                 string thisname, repnames;
 617                 map<string, vector<string> >::iterator it;
 618
 619                 while(!in.eof()){
 620
 621                         if (m->control_pressed) { in.close(); return 0; }
 622
 623                         in >> thisname;         m->gobble(in);          //read from first column
 624                         in >> repnames;                 //read from second column
 625
 626                         it = nameMap.find(thisname);
 627                         if (it == nameMap.end()) {
 628
 629                                 vector<string> splitRepNames;
 630                                 m->splitAtComma(repnames, splitRepNames);
 631
 632                                 nameMap[thisname] = splitRepNames;
 633                                 for (int i = 0; i < splitRepNames.size(); i++) { names.push_back(splitRepNames[i]); }
 634
 635                         }else{  m->mothurOut(thisname + " is already in namesfile. I will use first definition."); m->mothurOutEndLine();  }
 636
 637                         m->gobble(in);
 638                 }
 639                 in.close();
 640
 641                 return 0;
 642
 643         }
 644         catch(exception& e) {
 645                 m->errorOut(e, "SubSampleCommand", "readNames");
 646                 exit(1);
 647         }
 648 }
 649 //**********************************************************************************************************************
 650 int SubSampleCommand::getSubSampleShared() {
 651         try {
 652
 653                 InputData* input = new InputData(sharedfile, "sharedfile");
 654                 vector<SharedRAbundVector*> lookup = input->getSharedRAbundVectors();
 655                 string lastLabel = lookup[0]->getLabel();
 656
 657                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 658                 set<string> processedLabels;
 659                 set<string> userLabels = labels;
 660
 661                 if (size == 0) { //user has not set size, set size = smallest samples size
 662                         size = lookup[0]->getNumSeqs();
 663                         for (int i = 1; i < lookup.size(); i++) {
 664                                 int thisSize = lookup[i]->getNumSeqs();
 665
 666                                 if (thisSize < size) {  size = thisSize;        }
 667                         }
 668                 }else {
 669                         m->Groups.clear();
 670                         vector<SharedRAbundVector*> temp;
 671                         for (int i = 0; i < lookup.size(); i++) {
 672                                 if (lookup[i]->getNumSeqs() < size) {
 673                                         m->mothurOut(lookup[i]->getGroup() + " contains " + toString(lookup[i]->getNumSeqs()) + ". Eliminating."); m->mothurOutEndLine();
 674                                         delete lookup[i];
 675                                 }else {
 676                                         m->Groups.push_back(lookup[i]->getGroup());
 677                                         temp.push_back(lookup[i]);
 678                                 }
 679                         }
 680                         lookup = temp;
 681                         Groups = m->Groups;
 682                 }
 683
 684                 if (lookup.size() == 0) {  m->mothurOut("The size you selected is too large, skipping shared file."); m->mothurOutEndLine(); delete input; return 0; }
 685
 686                 string thisOutputDir = outputDir;
 687                 if (outputDir == "") {  thisOutputDir += m->hasPath(sharedfile);  }
 688                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(sharedfile)) + "subsample" + m->getExtension(sharedfile);
 689
 690                 ofstream out;
 691                 m->openOutputFile(outputFileName, out);
 692                 outputTypes["shared"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 693
 694
 695                 m->mothurOut("Sampling " + toString(size) + " from each group."); m->mothurOutEndLine();
 696
 697                 //as long as you are not at the end of the file or done wih the lines you want
 698                 while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 699                         if (m->control_pressed) {  delete input; for (int i = 0; i < lookup.size(); i++) {  delete lookup[i]; lookup[i] = NULL; } out.close(); return 0;  }
 700
 701                         if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){
 702
 703                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 704
 705                                 processShared(lookup, out);
 706
 707                                 processedLabels.insert(lookup[0]->getLabel());
 708                                 userLabels.erase(lookup[0]->getLabel());
 709                         }
 710
 711                         if ((m->anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 712                                 string saveLabel = lookup[0]->getLabel();
 713
 714                                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 715
 716                                 lookup = input->getSharedRAbundVectors(lastLabel);
 717                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 718
 719                                 processShared(lookup, out);
 720
 721                                 processedLabels.insert(lookup[0]->getLabel());
 722                                 userLabels.erase(lookup[0]->getLabel());
 723
 724                                 //restore real lastlabel to save below
 725                                 lookup[0]->setLabel(saveLabel);
 726                         }
 727
 728                         lastLabel = lookup[0]->getLabel();
 729                         //prevent memory leak
 730                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i]; lookup[i] = NULL; }
 731
 732                         //get next line to process
 733                         lookup = input->getSharedRAbundVectors();
 734                 }
 735
 736
 737                 if (m->control_pressed) {  out.close(); return 0;  }
 738
 739                 //output error messages about any remaining user labels
 740                 set<string>::iterator it;
 741                 bool needToRun = false;
 742                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 743                         m->mothurOut("Your file does not include the label " + *it);
 744                         if (processedLabels.count(lastLabel) != 1) {
 745                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 746                                 needToRun = true;
 747                         }else {
 748                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 749                         }
 750                 }
 751
 752                 //run last label if you need to
 753                 if (needToRun == true)  {
 754                         for (int i = 0; i < lookup.size(); i++) { if (lookup[i] != NULL) { delete lookup[i]; } }
 755                         lookup = input->getSharedRAbundVectors(lastLabel);
 756
 757                         m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 758
 759                         processShared(lookup, out);
 760
 761                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 762                 }
 763
 764                 delete input;
 765                 out.close();
 766
 767                 return 0;
 768
 769         }
 770         catch(exception& e) {
 771                 m->errorOut(e, "SubSampleCommand", "getSubSampleShared");
 772                 exit(1);
 773         }
 774 }
 775 //**********************************************************************************************************************
 776 int SubSampleCommand::processShared(vector<SharedRAbundVector*>& thislookup, ofstream& out) {
 777         try {
 778
 779                 int numBins = thislookup[0]->getNumBins();
 780                 for (int i = 0; i < thislookup.size(); i++) {
 781                         int thisSize = thislookup[i]->getNumSeqs();
 782
 783                         if (thisSize != size) {
 784
 785                                 string thisgroup = thislookup[i]->getGroup();
 786
 787                                 OrderVector* order = new OrderVector();
 788                                 for(int p=0;p<numBins;p++){
 789                                         for(int j=0;j<thislookup[i]->getAbundance(p);j++){
 790                                                 order->push_back(p);
 791                                         }
 792                                 }
 793                                 random_shuffle(order->begin(), order->end());
 794
 795                                 SharedRAbundVector* temp = new SharedRAbundVector(numBins);
 796                                 temp->setLabel(thislookup[i]->getLabel());
 797                                 temp->setGroup(thislookup[i]->getGroup());
 798
 799                                 delete thislookup[i];
 800                                 thislookup[i] = temp;
 801
 802
 803                                 for (int j = 0; j < size; j++) {
 804
 805                                         if (m->control_pressed) { delete order; return 0; }
 806
 807                                         //get random number to sample from order between 0 and thisSize-1.
 808                                         int myrand = int((float)(thisSize) * (float)(rand()) / ((float)RAND_MAX+1.0));
 809
 810                                         int bin = order->get(myrand);
 811
 812                                         int abund = thislookup[i]->getAbundance(bin);
 813                                         thislookup[i]->set(bin, (abund+1), thisgroup);
 814                                 }
 815                                 delete order;
 816                         }
 817                 }
 818
 819                 //subsampling may have created some otus with no sequences in them
 820                 eliminateZeroOTUS(thislookup);
 821
 822                 if (m->control_pressed) { return 0; }
 823
 824                 for (int i = 0; i < thislookup.size(); i++) {
 825                         out << thislookup[i]->getLabel() << '\t' << thislookup[i]->getGroup() << '\t';
 826                         thislookup[i]->print(out);
 827                 }
 828
 829                 return 0;
 830
 831         }
 832         catch(exception& e) {
 833                 m->errorOut(e, "SubSampleCommand", "processShared");
 834                 exit(1);
 835         }
 836 }
 837 //**********************************************************************************************************************
 838 int SubSampleCommand::getSubSampleList() {
 839         try {
 840
 841                 string thisOutputDir = outputDir;
 842                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 843                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "subsample" + m->getExtension(listfile);
 844
 845                 ofstream out;
 846                 m->openOutputFile(outputFileName, out);
 847                 outputTypes["list"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 848
 849                 InputData* input = new InputData(listfile, "list");
 850                 ListVector* list = input->getListVector();
 851                 string lastLabel = list->getLabel();
 852
 853                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 854                 set<string> processedLabels;
 855                 set<string> userLabels = labels;
 856
 857                 ofstream outGroup;
 858                 GroupMap* groupMap;
 859                 if (groupfile != "") {
 860
 861                         groupMap = new GroupMap(groupfile);
 862                         groupMap->readMap();
 863
 864                         //takes care of user setting groupNames that are invalid or setting groups=all
 865                         SharedUtil* util = new SharedUtil();
 866                         util->setGroups(Groups, groupMap->namesOfGroups);
 867                         delete util;
 868
 869                         //create outputfiles
 870                         string groupOutputDir = outputDir;
 871                         if (outputDir == "") {  groupOutputDir += m->hasPath(groupfile);  }
 872                         string groupOutputFileName = groupOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "subsample" + m->getExtension(groupfile);
 873
 874                         m->openOutputFile(groupOutputFileName, outGroup);
 875                         outputTypes["group"].push_back(groupOutputFileName);  outputNames.push_back(groupOutputFileName);
 876
 877                         //file mismatch quit
 878                         if (list->getNumSeqs() != groupMap->getNumSeqs()) {
 879                                 m->mothurOut("[ERROR]: your list file contains " + toString(list->getNumSeqs()) + " sequences, and your groupfile contains " + toString(groupMap->getNumSeqs()) + ", please correct.");
 880                                 m->mothurOutEndLine();
 881                                 delete groupMap;
 882                                 delete list;
 883                                 delete input;
 884                                 out.close();
 885                                 outGroup.close();
 886                                 return 0;
 887                         }
 888                 }
 889
 890                 //make sure that if your picked groups size is not too big
 891                 if (persample) {
 892                         if (size == 0) { //user has not set size, set size = smallest samples size
 893                                 size = groupMap->getNumSeqs(Groups[0]);
 894                                 for (int i = 1; i < Groups.size(); i++) {
 895                                         int thisSize = groupMap->getNumSeqs(Groups[i]);
 896
 897                                         if (thisSize < size) {  size = thisSize;        }
 898                                 }
 899                         }else { //make sure size is not too large
 900                                 vector<string> newGroups;
 901                                 for (int i = 0; i < Groups.size(); i++) {
 902                                         int thisSize = groupMap->getNumSeqs(Groups[i]);
 903
 904                                         if (thisSize >= size) { newGroups.push_back(Groups[i]); }
 905                                         else {  m->mothurOut("You have selected a size that is larger than " + Groups[i] + " number of sequences, removing " + Groups[i] + "."); m->mothurOutEndLine(); }
 906                                 }
 907                                 Groups = newGroups;
 908                         }
 909
 910                         m->mothurOut("Sampling " + toString(size) + " from each group."); m->mothurOutEndLine();
 911                 }else{
 912                         if (pickedGroups) {
 913                                 int total = 0;
 914                                 for(int i = 0; i < Groups.size(); i++) {
 915                                         total += groupMap->getNumSeqs(Groups[i]);
 916                                 }
 917
 918                                 if (size == 0) { //user has not set size, set size = 10% samples size
 919                                         size = int (total * 0.10);
 920                                 }
 921
 922                                 if (total < size) {
 923                                         m->mothurOut("Your size is too large for the number of groups you selected. Adjusting to " + toString(int (total * 0.10)) + "."); m->mothurOutEndLine();
 924                                         size = int (total * 0.10);
 925                                 }
 926
 927                                 m->mothurOut("Sampling " + toString(size) + " from " + toString(total) + "."); m->mothurOutEndLine();
 928                         }else{
 929
 930                                 if (size == 0) { //user has not set size, set size = 10% samples size
 931                                         size = int (list->getNumSeqs() * 0.10);
 932                                 }
 933
 934                                 int thisSize = list->getNumSeqs();
 935                                 if (size > thisSize) { m->mothurOut("Your list file only contains " + toString(thisSize) + " sequences. Setting size to " + toString(thisSize) + "."); m->mothurOutEndLine();
 936                                         size = thisSize;
 937                                 }
 938
 939                                 m->mothurOut("Sampling " + toString(size) + " from " + toString(list->getNumSeqs()) + "."); m->mothurOutEndLine();
 940                         }
 941                 }
 942
 943
 944                 //fill names
 945                 for (int i = 0; i < list->getNumBins(); i++) {
 946                         string binnames = list->get(i);
 947
 948                         //parse names
 949                         string individual = "";
 950                         int length = binnames.length();
 951                         for(int j=0;j<length;j++){
 952                                 if(binnames[j] == ','){
 953
 954                                         if (groupfile != "") { //if there is a groupfile given fill in group info
 955                                                 string group = groupMap->getGroup(individual);
 956                                                 if (group == "not found") { m->mothurOut("[ERROR]: " + individual + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
 957
 958                                                 if (pickedGroups) { //if hte user picked groups, we only want to keep the names of sequences from those groups
 959                                                         if (m->inUsersGroups(group, Groups)) {
 960                                                                 names.push_back(individual);
 961                                                         }
 962                                                 }else{
 963                                                         names.push_back(individual);
 964                                                 }
 965                                         }else{ //save everyone, group
 966                                                 names.push_back(individual);
 967                                         }
 968                                         individual = "";
 969                                 }
 970                                 else{
 971                                         individual += binnames[j];
 972                                 }
 973                         }
 974                         //save last name
 975                         if (groupfile != "") { //if there is a groupfile given fill in group info
 976                                 string group = groupMap->getGroup(individual);
 977                                 if (group == "not found") { m->mothurOut("[ERROR]: " + individual + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
 978
 979                                 if (pickedGroups) { //if hte user picked groups, we only want to keep the names of sequences from those groups
 980                                         if (m->inUsersGroups(group, Groups)) {
 981                                                 names.push_back(individual);
 982                                         }
 983                                 }else{
 984                                         names.push_back(individual);
 985                                 }
 986                         }else{ //save everyone, group
 987                                 names.push_back(individual);
 988                         }
 989                 }
 990
 991                 random_shuffle(names.begin(), names.end());
 992
 993                 //randomly select a subset of those names to include in the subsample
 994                 set<string> subset; //dont want repeat sequence names added
 995                 if (persample) {
 996                         for (int i = 0; i < Groups.size(); i++) {
 997
 998                                 for (int j = 0; j < size; j++) {
 999
1000                                         if (m->control_pressed) { break; }
1001
1002                                         //get random sequence to add, making sure we have not already added it
1003                                         bool done = false;
1004                                         int myrand;
1005                                         while (!done) {
1006                                                 myrand = int((float)(names.size()) * (float)(rand()) / ((float)RAND_MAX+1.0));
1007
1008                                                 if (subset.count(names[myrand]) == 0) { //you are not already added
1009                                                         if (groupMap->getGroup(names[myrand]) == Groups[i])  { subset.insert(names[myrand]); break;     }
1010                                                 }
1011                                         }
1012                                 }
1013                         }
1014                 }else{
1015                         for (int j = 0; j < size; j++) {
1016
1017                                 if (m->control_pressed) { break; }
1018
1019                                 //get random sequence to add, making sure we have not already added it
1020                                 bool done = false;
1021                                 int myrand;
1022                                 while (!done) {
1023                                         myrand = int((float)(names.size()) * (float)(rand()) / ((float)RAND_MAX+1.0));
1024
1025                                         if (subset.count(names[myrand]) == 0)  { subset.insert(names[myrand]); break;   }
1026                                 }
1027                         }
1028                 }
1029
1030                 if (groupfile != "") {
1031                         //write out new groupfile
1032                         for (set<string>::iterator it = subset.begin(); it != subset.end(); it++) {
1033                                 string group = groupMap->getGroup(*it);
1034                                 if (group == "not found") { group = "NOTFOUND"; }
1035
1036                                 outGroup << *it << '\t' << group << endl;
1037                         }
1038                         outGroup.close(); delete groupMap;
1039                 }
1040
1041
1042                 //as long as you are not at the end of the file or done wih the lines you want
1043                 while((list != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
1044
1045                         if (m->control_pressed) {  delete list; delete input; out.close();  return 0;  }
1046
1047                         if(allLines == 1 || labels.count(list->getLabel()) == 1){
1048
1049                                 m->mothurOut(list->getLabel()); m->mothurOutEndLine();
1050
1051                                 processList(list, out, subset);
1052
1053                                 processedLabels.insert(list->getLabel());
1054                                 userLabels.erase(list->getLabel());
1055                         }
1056
1057                         if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
1058                                 string saveLabel = list->getLabel();
1059
1060                                 delete list;
1061
1062                                 list = input->getListVector(lastLabel);
1063                                 m->mothurOut(list->getLabel()); m->mothurOutEndLine();
1064
1065                                 processList(list, out, subset);
1066
1067                                 processedLabels.insert(list->getLabel());
1068                                 userLabels.erase(list->getLabel());
1069
1070                                 //restore real lastlabel to save below
1071                                 list->setLabel(saveLabel);
1072                         }
1073
1074                         lastLabel = list->getLabel();
1075
1076                         delete list; list = NULL;
1077
1078                         //get next line to process
1079                         list = input->getListVector();
1080                 }
1081
1082
1083                 if (m->control_pressed) {  if (list != NULL) { delete list; } delete input; out.close(); return 0;  }
1084
1085                 //output error messages about any remaining user labels
1086                 set<string>::iterator it;
1087                 bool needToRun = false;
1088                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
1089                         m->mothurOut("Your file does not include the label " + *it);
1090                         if (processedLabels.count(lastLabel) != 1) {
1091                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
1092                                 needToRun = true;
1093                         }else {
1094                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
1095                         }
1096                 }
1097
1098                 //run last label if you need to
1099                 if (needToRun == true)  {
1100                         if (list != NULL) { delete list; }
1101
1102                         list = input->getListVector(lastLabel);
1103
1104                         m->mothurOut(list->getLabel()); m->mothurOutEndLine();
1105
1106                         processList(list, out, subset);
1107
1108                         delete list; list = NULL;
1109                 }
1110
1111                 out.close();
1112                 if (list != NULL) { delete list; }
1113                 delete input;
1114
1115                 return 0;
1116
1117         }
1118         catch(exception& e) {
1119                 m->errorOut(e, "SubSampleCommand", "getSubSampleList");
1120                 exit(1);
1121         }
1122 }
1123 //**********************************************************************************************************************
1124 int SubSampleCommand::processList(ListVector*& list, ofstream& out, set<string>& subset) {
1125         try {
1126
1127                 int numBins = list->getNumBins();
1128
1129                 ListVector* temp = new ListVector();
1130                 temp->setLabel(list->getLabel());
1131
1132                 for (int i = 0; i < numBins; i++) {
1133
1134                         if (m->control_pressed) { break; }
1135
1136                         string binnames = list->get(i);
1137
1138                         //parse names
1139                         string individual = "";
1140                         string newNames = "";
1141                         int length = binnames.length();
1142                         for(int j=0;j<length;j++){
1143                                 if(binnames[j] == ','){
1144                                         if (subset.count(individual) != 0) {  newNames += individual + ",";  }
1145                                         individual = "";
1146                                 }else{
1147                                         individual += binnames[j];
1148                                 }
1149                         }
1150                         if (subset.count(individual) != 0) {  newNames += individual + ",";  }
1151
1152
1153                         //if there are names in this bin add to new list
1154                         if (newNames != "") {
1155                                 newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
1156                                 temp->push_back(newNames);
1157                         }
1158                 }
1159
1160                 delete list;
1161                 list = temp;
1162
1163                 if (m->control_pressed) { return 0; }
1164
1165                 list->print(out);
1166
1167                 return 0;
1168
1169         }
1170         catch(exception& e) {
1171                 m->errorOut(e, "SubSampleCommand", "processList");
1172                 exit(1);
1173         }
1174 }
1175 //**********************************************************************************************************************
1176 int SubSampleCommand::getSubSampleRabund() {
1177         try {
1178                 InputData* input = new InputData(rabundfile, "rabund");
1179                 RAbundVector* rabund = input->getRAbundVector();
1180                 string lastLabel = rabund->getLabel();
1181
1182                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
1183                 set<string> processedLabels;
1184                 set<string> userLabels = labels;
1185
1186                 if (size == 0) { //user has not set size, set size = 10%
1187                         size = int((rabund->getNumSeqs()) * 0.10);
1188                 }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; }
1189
1190                 m->mothurOut("Sampling " + toString(size) + " from " + toString(rabund->getNumSeqs()) + "."); m->mothurOutEndLine();
1191
1192                 string thisOutputDir = outputDir;
1193                 if (outputDir == "") {  thisOutputDir += m->hasPath(rabundfile);  }
1194                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(rabundfile)) + "subsample" + m->getExtension(rabundfile);
1195
1196                 ofstream out;
1197                 m->openOutputFile(outputFileName, out);
1198                 outputTypes["rabund"].push_back(outputFileName);  outputNames.push_back(outputFileName);
1199
1200                 //as long as you are not at the end of the file or done wih the lines you want
1201                 while((rabund != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
1202                         if (m->control_pressed) {  delete input; delete rabund; out.close(); return 0;  }
1203
1204                         if(allLines == 1 || labels.count(rabund->getLabel()) == 1){
1205
1206                                 m->mothurOut(rabund->getLabel()); m->mothurOutEndLine();
1207
1208                                 processRabund(rabund, out);
1209
1210                                 processedLabels.insert(rabund->getLabel());
1211                                 userLabels.erase(rabund->getLabel());
1212                         }
1213
1214                         if ((m->anyLabelsToProcess(rabund->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
1215                                 string saveLabel = rabund->getLabel();
1216
1217                                 delete rabund;
1218
1219                                 rabund = input->getRAbundVector(lastLabel);
1220                                 m->mothurOut(rabund->getLabel()); m->mothurOutEndLine();
1221
1222                                 processRabund(rabund, out);
1223
1224                                 processedLabels.insert(rabund->getLabel());
1225                                 userLabels.erase(rabund->getLabel());
1226
1227                                 //restore real lastlabel to save below
1228                                 rabund->setLabel(saveLabel);
1229                         }
1230
1231                         lastLabel = rabund->getLabel();
1232
1233                         //prevent memory leak
1234                         delete rabund; rabund = NULL;
1235
1236                         //get next line to process
1237                         rabund = input->getRAbundVector();
1238                 }
1239
1240
1241                 if (m->control_pressed) {  out.close(); return 0;  }
1242
1243                 //output error messages about any remaining user labels
1244                 set<string>::iterator it;
1245                 bool needToRun = false;
1246                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
1247                         m->mothurOut("Your file does not include the label " + *it);
1248                         if (processedLabels.count(lastLabel) != 1) {
1249                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
1250                                 needToRun = true;
1251                         }else {
1252                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
1253                         }
1254                 }
1255
1256                 //run last label if you need to
1257                 if (needToRun == true)  {
1258                         if (rabund != NULL) { delete rabund; }
1259
1260                         rabund = input->getRAbundVector(lastLabel);
1261
1262                         m->mothurOut(rabund->getLabel()); m->mothurOutEndLine();
1263
1264                         processRabund(rabund, out);
1265
1266                         delete rabund;
1267                 }
1268
1269                 delete input;
1270                 out.close();
1271
1272                 return 0;
1273
1274         }
1275         catch(exception& e) {
1276                 m->errorOut(e, "SubSampleCommand", "getSubSampleRabund");
1277                 exit(1);
1278         }
1279 }
1280 //**********************************************************************************************************************
1281 int SubSampleCommand::processRabund(RAbundVector*& rabund, ofstream& out) {
1282         try {
1283
1284                 int numBins = rabund->getNumBins();
1285                 int thisSize = rabund->getNumSeqs();
1286
1287                 if (thisSize != size) {
1288
1289                         OrderVector* order = new OrderVector();
1290                         for(int p=0;p<numBins;p++){
1291                                 for(int j=0;j<rabund->get(p);j++){
1292                                         order->push_back(p);
1293                                 }
1294                         }
1295                         random_shuffle(order->begin(), order->end());
1296
1297                         RAbundVector* temp = new RAbundVector(numBins);
1298                         temp->setLabel(rabund->getLabel());
1299
1300                         delete rabund;
1301                         rabund = temp;
1302
1303                         for (int j = 0; j < size; j++) {
1304
1305                                 if (m->control_pressed) { delete order; return 0; }
1306
1307                                 //get random number to sample from order between 0 and thisSize-1.
1308                                 int myrand = int((float)(thisSize) * (float)(rand()) / ((float)RAND_MAX+1.0));
1309
1310                                 int bin = order->get(myrand);
1311
1312                                 int abund = rabund->get(bin);
1313                                 rabund->set(bin, (abund+1));
1314                         }
1315
1316                         delete order;
1317                 }
1318
1319                 if (m->control_pressed) { return 0; }
1320
1321                 rabund->print(out);
1322
1323                 return 0;
1324
1325         }
1326         catch(exception& e) {
1327                 m->errorOut(e, "SubSampleCommand", "processRabund");
1328                 exit(1);
1329         }
1330 }
1331 //**********************************************************************************************************************
1332 int SubSampleCommand::getSubSampleSabund() {
1333         try {
1334
1335                 InputData* input = new InputData(sabundfile, "sabund");
1336                 SAbundVector* sabund = input->getSAbundVector();
1337                 string lastLabel = sabund->getLabel();
1338
1339                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
1340                 set<string> processedLabels;
1341                 set<string> userLabels = labels;
1342
1343                 if (size == 0) { //user has not set size, set size = 10%
1344                         size = int((sabund->getNumSeqs()) * 0.10);
1345                 }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; }
1346
1347
1348                 m->mothurOut("Sampling " + toString(size) + " from " + toString(sabund->getNumSeqs()) + "."); m->mothurOutEndLine();
1349
1350                 string thisOutputDir = outputDir;
1351                 if (outputDir == "") {  thisOutputDir += m->hasPath(sabundfile);  }
1352                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(sabundfile)) + "subsample" + m->getExtension(sabundfile);
1353
1354                 ofstream out;
1355                 m->openOutputFile(outputFileName, out);
1356                 outputTypes["sabund"].push_back(outputFileName);  outputNames.push_back(outputFileName);
1357
1358
1359                 //as long as you are not at the end of the file or done wih the lines you want
1360                 while((sabund != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
1361                         if (m->control_pressed) {  delete input; delete sabund; out.close(); return 0;  }
1362
1363                         if(allLines == 1 || labels.count(sabund->getLabel()) == 1){
1364
1365                                 m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
1366
1367                                 processSabund(sabund, out);
1368
1369                                 processedLabels.insert(sabund->getLabel());
1370                                 userLabels.erase(sabund->getLabel());
1371                         }
1372
1373                         if ((m->anyLabelsToProcess(sabund->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
1374                                 string saveLabel = sabund->getLabel();
1375
1376                                 delete sabund;
1377
1378                                 sabund = input->getSAbundVector(lastLabel);
1379                                 m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
1380
1381                                 processSabund(sabund, out);
1382
1383                                 processedLabels.insert(sabund->getLabel());
1384                                 userLabels.erase(sabund->getLabel());
1385
1386                                 //restore real lastlabel to save below
1387                                 sabund->setLabel(saveLabel);
1388                         }
1389
1390                         lastLabel = sabund->getLabel();
1391
1392                         //prevent memory leak
1393                         delete sabund; sabund = NULL;
1394
1395                         //get next line to process
1396                         sabund = input->getSAbundVector();
1397                 }
1398
1399
1400                 if (m->control_pressed) {  out.close(); return 0;  }
1401
1402                 //output error messages about any remaining user labels
1403                 set<string>::iterator it;
1404                 bool needToRun = false;
1405                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
1406                         m->mothurOut("Your file does not include the label " + *it);
1407                         if (processedLabels.count(lastLabel) != 1) {
1408                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
1409                                 needToRun = true;
1410                         }else {
1411                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
1412                         }
1413                 }
1414
1415                 //run last label if you need to
1416                 if (needToRun == true)  {
1417                         if (sabund != NULL) { delete sabund; }
1418
1419                         sabund = input->getSAbundVector(lastLabel);
1420
1421                         m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
1422
1423                         processSabund(sabund, out);
1424
1425                         delete sabund;
1426                 }
1427
1428                 delete input;
1429                 out.close();
1430
1431                 return 0;
1432
1433         }
1434         catch(exception& e) {
1435                 m->errorOut(e, "SubSampleCommand", "getSubSampleSabund");
1436                 exit(1);
1437         }
1438 }
1439 //**********************************************************************************************************************
1440 int SubSampleCommand::processSabund(SAbundVector*& sabund, ofstream& out) {
1441         try {
1442
1443                 RAbundVector* rabund = new RAbundVector();
1444                 *rabund = sabund->getRAbundVector();
1445
1446                 int numBins = rabund->getNumBins();
1447                 int thisSize = rabund->getNumSeqs();
1448
1449                 if (thisSize != size) {
1450
1451                         OrderVector* order = new OrderVector();
1452                         for(int p=0;p<numBins;p++){
1453                                 for(int j=0;j<rabund->get(p);j++){
1454                                         order->push_back(p);
1455                                 }
1456                         }
1457                         random_shuffle(order->begin(), order->end());
1458
1459                         RAbundVector* temp = new RAbundVector(numBins);
1460                         temp->setLabel(rabund->getLabel());
1461
1462                         delete rabund;
1463                         rabund = temp;
1464
1465                         for (int j = 0; j < size; j++) {
1466
1467                                 if (m->control_pressed) { delete order; return 0; }
1468
1469                                 //get random number to sample from order between 0 and thisSize-1.
1470                                 int myrand = int((float)(thisSize) * (float)(rand()) / ((float)RAND_MAX+1.0));
1471
1472                                 int bin = order->get(myrand);
1473
1474                                 int abund = rabund->get(bin);
1475                                 rabund->set(bin, (abund+1));
1476                         }
1477
1478                         delete order;
1479                 }
1480
1481                 if (m->control_pressed) { return 0; }
1482
1483                 delete sabund;
1484                 sabund = new SAbundVector();
1485                 *sabund = rabund->getSAbundVector();
1486                 delete rabund;
1487
1488                 sabund->print(out);
1489
1490                 return 0;
1491
1492         }
1493         catch(exception& e) {
1494                 m->errorOut(e, "SubSampleCommand", "processSabund");
1495                 exit(1);
1496         }
1497 }
1498 //**********************************************************************************************************************
1499 int SubSampleCommand::eliminateZeroOTUS(vector<SharedRAbundVector*>& thislookup) {
1500         try {
1501
1502                 vector<SharedRAbundVector*> newLookup;
1503                 for (int i = 0; i < thislookup.size(); i++) {
1504                         SharedRAbundVector* temp = new SharedRAbundVector();
1505                         temp->setLabel(thislookup[i]->getLabel());
1506                         temp->setGroup(thislookup[i]->getGroup());
1507                         newLookup.push_back(temp);
1508                 }
1509
1510                 //for each bin
1511                 for (int i = 0; i < thislookup[0]->getNumBins(); i++) {
1512                         if (m->control_pressed) { for (int j = 0; j < newLookup.size(); j++) {  delete newLookup[j];  } return 0; }
1513
1514                         //look at each sharedRabund and make sure they are not all zero
1515                         bool allZero = true;
1516                         for (int j = 0; j < thislookup.size(); j++) {
1517                                 if (thislookup[j]->getAbundance(i) != 0) { allZero = false;  break;  }
1518                         }
1519
1520                         //if they are not all zero add this bin
1521                         if (!allZero) {
1522                                 for (int j = 0; j < thislookup.size(); j++) {
1523                                         newLookup[j]->push_back(thislookup[j]->getAbundance(i), thislookup[j]->getGroup());
1524                                 }
1525                         }
1526                 }
1527
1528                 for (int j = 0; j < thislookup.size(); j++) {  delete thislookup[j];  }
1529                 thislookup.clear();
1530
1531                 thislookup = newLookup;
1532
1533                 return 0;
1534
1535         }
1536         catch(exception& e) {
1537                 m->errorOut(e, "SubSampleCommand", "eliminateZeroOTUS");
1538                 exit(1);
1539         }
1540 }
1541
1542 //**********************************************************************************************************************
1543
1544
1545