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
  92                 else {
  93                         vector<string> myArray = setParameters();
  94
  95                         OptionParser parser(option);
  96                         map<string,string> parameters = parser.getParameters();
  97
  98                         ValidParameters validParameter;
  99
 100                         //check to make sure all parameters are valid for command
 101                         map<string,string>::iterator it;
 102                         for (it = parameters.begin(); it != parameters.end(); it++) {
 103                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 104                         }
 105
 106                         //initialize outputTypes
 107                         vector<string> tempOutNames;
 108                         outputTypes["shared"] = tempOutNames;
 109                         outputTypes["list"] = tempOutNames;
 110                         outputTypes["rabund"] = tempOutNames;
 111                         outputTypes["sabund"] = tempOutNames;
 112                         outputTypes["fasta"] = tempOutNames;
 113                         outputTypes["name"] = tempOutNames;
 114                         outputTypes["group"] = tempOutNames;
 115
 116                         //if the user changes the output directory command factory will send this info to us in the output parameter
 117                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
 118
 119                         //if the user changes the input directory command factory will send this info to us in the output parameter
 120                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 121                         if (inputDir == "not found"){   inputDir = "";          }
 122                         else {
 123                                 string path;
 124                                 it = parameters.find("list");
 125                                 //user has given a template file
 126                                 if(it != parameters.end()){
 127                                         path = m->hasPath(it->second);
 128                                         //if the user has not given a path then, add inputdir. else leave path alone.
 129                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 130                                 }
 131
 132                                 it = parameters.find("fasta");
 133                                 //user has given a template file
 134                                 if(it != parameters.end()){
 135                                         path = m->hasPath(it->second);
 136                                         //if the user has not given a path then, add inputdir. else leave path alone.
 137                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 138                                 }
 139
 140                                 it = parameters.find("shared");
 141                                 //user has given a template file
 142                                 if(it != parameters.end()){
 143                                         path = m->hasPath(it->second);
 144                                         //if the user has not given a path then, add inputdir. else leave path alone.
 145                                         if (path == "") {       parameters["shared"] = inputDir + it->second;           }
 146                                 }
 147
 148                                 it = parameters.find("group");
 149                                 //user has given a template file
 150                                 if(it != parameters.end()){
 151                                         path = m->hasPath(it->second);
 152                                         //if the user has not given a path then, add inputdir. else leave path alone.
 153                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 154                                 }
 155
 156                                 it = parameters.find("sabund");
 157                                 //user has given a template file
 158                                 if(it != parameters.end()){
 159                                         path = m->hasPath(it->second);
 160                                         //if the user has not given a path then, add inputdir. else leave path alone.
 161                                         if (path == "") {       parameters["sabund"] = inputDir + it->second;           }
 162                                 }
 163
 164                                 it = parameters.find("rabund");
 165                                 //user has given a template file
 166                                 if(it != parameters.end()){
 167                                         path = m->hasPath(it->second);
 168                                         //if the user has not given a path then, add inputdir. else leave path alone.
 169                                         if (path == "") {       parameters["rabund"] = inputDir + it->second;           }
 170                                 }
 171
 172                                 it = parameters.find("name");
 173                                 //user has given a template file
 174                                 if(it != parameters.end()){
 175                                         path = m->hasPath(it->second);
 176                                         //if the user has not given a path then, add inputdir. else leave path alone.
 177                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 178                                 }
 179                         }
 180
 181                         //check for required parameters
 182                         listfile = validParameter.validFile(parameters, "list", true);
 183                         if (listfile == "not open") { listfile = ""; abort = true; }
 184                         else if (listfile == "not found") { listfile = ""; }
 185
 186                         sabundfile = validParameter.validFile(parameters, "sabund", true);
 187                         if (sabundfile == "not open") { sabundfile = ""; abort = true; }
 188                         else if (sabundfile == "not found") { sabundfile = ""; }
 189
 190                         rabundfile = validParameter.validFile(parameters, "rabund", true);
 191                         if (rabundfile == "not open") { rabundfile = ""; abort = true; }
 192                         else if (rabundfile == "not found") { rabundfile = ""; }
 193
 194                         fastafile = validParameter.validFile(parameters, "fasta", true);
 195                         if (fastafile == "not open") { fastafile = ""; abort = true; }
 196                         else if (fastafile == "not found") { fastafile = ""; }
 197
 198                         sharedfile = validParameter.validFile(parameters, "shared", true);
 199                         if (sharedfile == "not open") { sharedfile = ""; abort = true; }
 200                         else if (sharedfile == "not found") { sharedfile = ""; }
 201
 202                         namefile = validParameter.validFile(parameters, "name", true);
 203                         if (namefile == "not open") { namefile = ""; abort = true; }
 204                         else if (namefile == "not found") { namefile = ""; }
 205
 206                         groupfile = validParameter.validFile(parameters, "group", true);
 207                         if (groupfile == "not open") { groupfile = ""; abort = true; }
 208                         else if (groupfile == "not found") { groupfile = ""; }
 209
 210
 211                         //check for optional parameter and set defaults
 212                         // ...at some point should added some additional type checking...
 213                         label = validParameter.validFile(parameters, "label", false);
 214                         if (label == "not found") { label = ""; }
 215                         else {
 216                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 217                                 else { allLines = 1;  }
 218                         }
 219
 220                         groups = validParameter.validFile(parameters, "groups", false);
 221                         if (groups == "not found") { groups = ""; pickedGroups = false; }
 222                         else {
 223                                 pickedGroups = true;
 224                                 m->splitAtDash(groups, Groups);
 225                                 m->Groups = Groups;
 226                         }
 227
 228                         string temp = validParameter.validFile(parameters, "size", false);              if (temp == "not found"){       temp = "0";             }
 229                         convert(temp, size);
 230
 231                         temp = validParameter.validFile(parameters, "persample", false);                if (temp == "not found"){       temp = "f";             }
 232                         persample = m->isTrue(temp);
 233
 234                         if (groupfile == "") { persample = false; }
 235
 236                         if ((namefile != "") && (fastafile == "")) { m->mothurOut("You may only use a namefile with a fastafile."); m->mothurOutEndLine(); abort = true; }
 237
 238                         if ((fastafile == "") && (listfile == "") && (sabundfile == "") && (rabundfile == "") && (sharedfile == "")) {
 239                                 m->mothurOut("You must provide a fasta, list, sabund, rabund or shared file as an input file."); m->mothurOutEndLine(); abort = true; }
 240
 241                         if (pickedGroups && ((groupfile == "") && (sharedfile == ""))) {
 242                                 m->mothurOut("You cannot pick groups without a valid group file or shared file."); m->mothurOutEndLine(); abort = true; }
 243
 244                         if ((groupfile != "") && ((fastafile == "") && (listfile == ""))) {
 245                                 m->mothurOut("Group file only valid with listfile or fastafile."); m->mothurOutEndLine(); abort = true; }
 246
 247                         if ((groupfile != "") && ((fastafile != "") && (listfile != ""))) {
 248                                 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; }
 249
 250                 }
 251
 252         }
 253         catch(exception& e) {
 254                 m->errorOut(e, "SubSampleCommand", "SubSampleCommand");
 255                 exit(1);
 256         }
 257 }
 258 //**********************************************************************************************************************
 259
 260 int SubSampleCommand::execute(){
 261         try {
 262
 263                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 264
 265                 if (sharedfile != "")   {   getSubSampleShared();       }
 266                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       remove(outputNames[i].c_str()); return 0; } }
 267
 268                 if (listfile != "")             {   getSubSampleList();         }
 269                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       remove(outputNames[i].c_str()); return 0; } }
 270
 271                 if (rabundfile != "")   {   getSubSampleRabund();       }
 272                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       remove(outputNames[i].c_str()); return 0; } }
 273
 274                 if (sabundfile != "")   {   getSubSampleSabund();       }
 275                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       remove(outputNames[i].c_str()); return 0; } }
 276
 277                 if (fastafile != "")    {   getSubSampleFasta();        }
 278                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       remove(outputNames[i].c_str()); return 0; } }
 279
 280                 //set fasta file as new current fastafile
 281                 string current = "";
 282                 itTypes = outputTypes.find("fasta");
 283                 if (itTypes != outputTypes.end()) {
 284                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 285                 }
 286
 287                 itTypes = outputTypes.find("name");
 288                 if (itTypes != outputTypes.end()) {
 289                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
 290                 }
 291
 292                 itTypes = outputTypes.find("group");
 293                 if (itTypes != outputTypes.end()) {
 294                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setGroupFile(current); }
 295                 }
 296
 297                 itTypes = outputTypes.find("list");
 298                 if (itTypes != outputTypes.end()) {
 299                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 300                 }
 301
 302                 itTypes = outputTypes.find("shared");
 303                 if (itTypes != outputTypes.end()) {
 304                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setSharedFile(current); }
 305                 }
 306
 307                 itTypes = outputTypes.find("rabund");
 308                 if (itTypes != outputTypes.end()) {
 309                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setRabundFile(current); }
 310                 }
 311
 312                 itTypes = outputTypes.find("sabund");
 313                 if (itTypes != outputTypes.end()) {
 314                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setSabundFile(current); }
 315                 }
 316
 317
 318                 m->mothurOutEndLine();
 319                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 320                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 321                 m->mothurOutEndLine();
 322
 323                 return 0;
 324         }
 325         catch(exception& e) {
 326                 m->errorOut(e, "SubSampleCommand", "execute");
 327                 exit(1);
 328         }
 329 }
 330 //**********************************************************************************************************************
 331 int SubSampleCommand::getSubSampleFasta() {
 332         try {
 333
 334                 if (namefile != "") { readNames(); }    //fills names with all names in namefile.
 335                 else { getNames(); }//no name file, so get list of names to pick from
 336
 337                 GroupMap* groupMap;
 338                 if (groupfile != "") {
 339
 340                         groupMap = new GroupMap(groupfile);
 341                         groupMap->readMap();
 342
 343                         //takes care of user setting groupNames that are invalid or setting groups=all
 344                         SharedUtil* util = new SharedUtil();
 345                         util->setGroups(Groups, groupMap->namesOfGroups);
 346                         delete util;
 347
 348                         //file mismatch quit
 349                         if (names.size() != groupMap->getNumSeqs()) {
 350                                 m->mothurOut("[ERROR]: your fasta file contains " + toString(names.size()) + " sequences, and your groupfile contains " + toString(groupMap->getNumSeqs()) + ", please correct.");
 351                                 m->mothurOutEndLine();
 352                                 delete groupMap;
 353                                 return 0;
 354                         }
 355                 }
 356
 357                 if (m->control_pressed) { return 0; }
 358
 359
 360                 //make sure that if your picked groups size is not too big
 361                 int thisSize = names.size();
 362                 if (persample) {
 363                         if (size == 0) { //user has not set size, set size = smallest samples size
 364                                 size = groupMap->getNumSeqs(Groups[0]);
 365                                 for (int i = 1; i < Groups.size(); i++) {
 366                                         int thisSize = groupMap->getNumSeqs(Groups[i]);
 367
 368                                         if (thisSize < size) {  size = thisSize;        }
 369                                 }
 370                         }else { //make sure size is not too large
 371                                 vector<string> newGroups;
 372                                 for (int i = 0; i < Groups.size(); i++) {
 373                                         int thisSize = groupMap->getNumSeqs(Groups[i]);
 374
 375                                         if (thisSize >= size) { newGroups.push_back(Groups[i]); }
 376                                         else {  m->mothurOut("You have selected a size that is larger than " + Groups[i] + " number of sequences, removing " + Groups[i] + "."); m->mothurOutEndLine(); }
 377                                 }
 378                                 Groups = newGroups;
 379                         }
 380
 381                         m->mothurOut("Sampling " + toString(size) + " from each group."); m->mothurOutEndLine();
 382                 }else {
 383                         if (pickedGroups) {
 384                                 int total = 0;
 385                                 for(int i = 0; i < Groups.size(); i++) {
 386                                         total += groupMap->getNumSeqs(Groups[i]);
 387                                 }
 388
 389                                 if (size == 0) { //user has not set size, set size = 10% samples size
 390                                         size = int (total * 0.10);
 391                                 }
 392
 393                                 if (total < size) {
 394                                         if (size != 0) {
 395                                                 m->mothurOut("Your size is too large for the number of groups you selected. Adjusting to " + toString(int (total * 0.10)) + "."); m->mothurOutEndLine();
 396                                         }
 397                                         size = int (total * 0.10);
 398                                 }
 399
 400                                 m->mothurOut("Sampling " + toString(size) + " from " + toString(total) + "."); m->mothurOutEndLine();
 401                         }
 402
 403                         if (size == 0) { //user has not set size, set size = 10% samples size
 404                                 size = int (names.size() * 0.10);
 405                         }
 406
 407                         if (size > thisSize) { m->mothurOut("Your fasta file only contains " + toString(thisSize) + " sequences. Setting size to " + toString(thisSize) + "."); m->mothurOutEndLine();
 408                                 size = thisSize;
 409                         }
 410
 411                         if (!pickedGroups) { m->mothurOut("Sampling " + toString(size) + " from " + toString(thisSize) + "."); m->mothurOutEndLine(); }
 412
 413                 }
 414                 random_shuffle(names.begin(), names.end());
 415
 416                 set<string> subset; //dont want repeat sequence names added
 417                 if (persample) {
 418                         for (int i = 0; i < Groups.size(); i++) {
 419
 420                                 //randomly select a subset of those names from this group to include in the subsample
 421                                 for (int j = 0; j < size; j++) {
 422
 423                                         if (m->control_pressed) { return 0; }
 424
 425                                         //get random sequence to add, making sure we have not already added it
 426                                         bool done = false;
 427                                         int myrand;
 428                                         while (!done) {
 429                                                 myrand = int((float)(thisSize) * (float)(rand()) / ((float)RAND_MAX+1.0));
 430
 431                                                 if (subset.count(names[myrand]) == 0)  {
 432
 433                                                         string group = groupMap->getGroup(names[myrand]);
 434                                                         if (group == "not found") { m->mothurOut("[ERROR]: " + names[myrand] + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
 435
 436                                                         if (group == Groups[i]) {       subset.insert(names[myrand]); break;    }
 437                                                 }
 438                                         }
 439                                 }
 440                         }
 441                 }else {
 442
 443                         //randomly select a subset of those names to include in the subsample
 444                         for (int j = 0; j < size; j++) {
 445
 446                                 if (m->control_pressed) { return 0; }
 447
 448                                 //get random sequence to add, making sure we have not already added it
 449                                 bool done = false;
 450                                 int myrand;
 451                                 while (!done) {
 452                                         myrand = int((float)(thisSize) * (float)(rand()) / ((float)RAND_MAX+1.0));
 453
 454                                         if (subset.count(names[myrand]) == 0)  {
 455
 456                                                 if (groupfile != "") { //if there is a groupfile given fill in group info
 457                                                         string group = groupMap->getGroup(names[myrand]);
 458                                                         if (group == "not found") { m->mothurOut("[ERROR]: " + names[myrand] + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
 459
 460                                                         if (pickedGroups) { //if hte user picked groups, we only want to keep the names of sequences from those groups
 461                                                                 if (m->inUsersGroups(group, Groups)) {
 462                                                                         subset.insert(names[myrand]); break;
 463                                                                 }
 464                                                         }else{
 465                                                                 subset.insert(names[myrand]); break;
 466                                                         }
 467                                                 }else{ //save everyone, group
 468                                                         subset.insert(names[myrand]); break;
 469                                                 }
 470                                         }
 471                                 }
 472                         }
 473                 }
 474
 475                 if (subset.size() == 0) {  m->mothurOut("The size you selected is too large, skipping fasta file."); m->mothurOutEndLine();  return 0; }
 476
 477                 string thisOutputDir = outputDir;
 478                 if (outputDir == "") {  thisOutputDir += m->hasPath(fastafile);  }
 479                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(fastafile)) + "subsample" + m->getExtension(fastafile);
 480
 481                 ofstream out;
 482                 m->openOutputFile(outputFileName, out);
 483                 outputTypes["fasta"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 484
 485                 //read through fasta file outputting only the names on the subsample list
 486                 ifstream in;
 487                 m->openInputFile(fastafile, in);
 488
 489                 string thisname;
 490                 int count = 0;
 491                 map<string, vector<string> >::iterator itNameMap;
 492
 493                 while(!in.eof()){
 494
 495                         if (m->control_pressed) { in.close(); out.close();  return 0; }
 496
 497                         Sequence currSeq(in);
 498                         thisname = currSeq.getName();
 499
 500                         if (thisname != "") {
 501
 502                                 //does the subset contain a sequence that this sequence represents
 503                                 itNameMap = nameMap.find(thisname);
 504                                 if (itNameMap != nameMap.end()) {
 505                                         vector<string> nameRepresents = itNameMap->second;
 506
 507                                         for (int i = 0; i < nameRepresents.size(); i++){
 508                                                 if (subset.count(nameRepresents[i]) != 0) {
 509                                                         out << ">" << nameRepresents[i] << endl << currSeq.getAligned() << endl;
 510                                                         count++;
 511                                                 }
 512                                         }
 513                                 }else{
 514                                         m->mothurOut("[ERROR]: " + thisname + " is not in your namefile, please correct."); m->mothurOutEndLine();
 515                                 }
 516                         }
 517                         m->gobble(in);
 518                 }
 519                 in.close();
 520                 out.close();
 521
 522                 if (count != subset.size()) {
 523                         m->mothurOut("[ERROR]: The subset selected contained " + toString(subset.size()) + " sequences, but I only found " + toString(count) + " of those in the fastafile."); m->mothurOutEndLine();
 524                 }
 525
 526                 //if a groupfile is provided read through the group file only outputting the names on the subsample list
 527                 if (groupfile != "") {
 528
 529                         string groupOutputDir = outputDir;
 530                         if (outputDir == "") {  groupOutputDir += m->hasPath(groupfile);  }
 531                         string groupOutputFileName = groupOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "subsample" + m->getExtension(groupfile);
 532
 533                         ofstream outGroup;
 534                         m->openOutputFile(groupOutputFileName, outGroup);
 535                         outputTypes["group"].push_back(groupOutputFileName);  outputNames.push_back(groupOutputFileName);
 536
 537                         ifstream inGroup;
 538                         m->openInputFile(groupfile, inGroup);
 539                         string name, group;
 540
 541                         while(!inGroup.eof()){
 542
 543                                 if (m->control_pressed) { inGroup.close(); outGroup.close(); return 0; }
 544
 545                                 inGroup >> name;        m->gobble(inGroup);                     //read from first column
 546                                 inGroup >> group;                       //read from second column
 547
 548                                 //if this name is in the accnos file
 549                                 if (subset.count(name) != 0) {
 550                                         outGroup << name << '\t' << group << endl;
 551                                         subset.erase(name);
 552                                 }
 553
 554                                 m->gobble(inGroup);
 555                         }
 556                         inGroup.close();
 557                         outGroup.close();
 558
 559                         //sanity check
 560                         if (subset.size() != 0) {
 561                                 m->mothurOut("Your groupfile does not match your fasta file."); m->mothurOutEndLine();
 562                                 for (set<string>::iterator it = subset.begin(); it != subset.end(); it++) {
 563                                         m->mothurOut("[ERROR]: " + *it + " is missing from your groupfile."); m->mothurOutEndLine();
 564                                 }
 565                         }
 566                 }
 567
 568                 return 0;
 569
 570         }
 571         catch(exception& e) {
 572                 m->errorOut(e, "SubSampleCommand", "getSubSampleFasta");
 573                 exit(1);
 574         }
 575 }
 576 //**********************************************************************************************************************
 577 int SubSampleCommand::getNames() {
 578         try {
 579
 580                 ifstream in;
 581                 m->openInputFile(fastafile, in);
 582
 583                 string thisname;
 584                 while(!in.eof()){
 585
 586                         if (m->control_pressed) { in.close(); return 0; }
 587
 588                         Sequence currSeq(in);
 589                         thisname = currSeq.getName();
 590
 591                         if (thisname != "") {
 592                                 vector<string> temp; temp.push_back(thisname);
 593                                 nameMap[thisname] = temp;
 594                                 names.push_back(thisname);
 595                         }
 596                         m->gobble(in);
 597                 }
 598                 in.close();
 599
 600                 return 0;
 601
 602         }
 603         catch(exception& e) {
 604                 m->errorOut(e, "SubSampleCommand", "getNames");
 605                 exit(1);
 606         }
 607 }
 608 //**********************************************************************************************************************
 609 int SubSampleCommand::readNames() {
 610         try {
 611
 612                 ifstream in;
 613                 m->openInputFile(namefile, in);
 614
 615                 string thisname, repnames;
 616                 map<string, vector<string> >::iterator it;
 617
 618                 while(!in.eof()){
 619
 620                         if (m->control_pressed) { in.close(); return 0; }
 621
 622                         in >> thisname;         m->gobble(in);          //read from first column
 623                         in >> repnames;                 //read from second column
 624
 625                         it = nameMap.find(thisname);
 626                         if (it == nameMap.end()) {
 627
 628                                 vector<string> splitRepNames;
 629                                 m->splitAtComma(repnames, splitRepNames);
 630
 631                                 nameMap[thisname] = splitRepNames;
 632                                 for (int i = 0; i < splitRepNames.size(); i++) { names.push_back(splitRepNames[i]); }
 633
 634                         }else{  m->mothurOut(thisname + " is already in namesfile. I will use first definition."); m->mothurOutEndLine();  }
 635
 636                         m->gobble(in);
 637                 }
 638                 in.close();
 639
 640                 return 0;
 641
 642         }
 643         catch(exception& e) {
 644                 m->errorOut(e, "SubSampleCommand", "readNames");
 645                 exit(1);
 646         }
 647 }
 648 //**********************************************************************************************************************
 649 int SubSampleCommand::getSubSampleShared() {
 650         try {
 651
 652                 InputData* input = new InputData(sharedfile, "sharedfile");
 653                 vector<SharedRAbundVector*> lookup = input->getSharedRAbundVectors();
 654                 string lastLabel = lookup[0]->getLabel();
 655
 656                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 657                 set<string> processedLabels;
 658                 set<string> userLabels = labels;
 659
 660                 if (size == 0) { //user has not set size, set size = smallest samples size
 661                         size = lookup[0]->getNumSeqs();
 662                         for (int i = 1; i < lookup.size(); i++) {
 663                                 int thisSize = lookup[i]->getNumSeqs();
 664
 665                                 if (thisSize < size) {  size = thisSize;        }
 666                         }
 667                 }else {
 668                         m->Groups.clear();
 669                         vector<SharedRAbundVector*> temp;
 670                         for (int i = 0; i < lookup.size(); i++) {
 671                                 if (lookup[i]->getNumSeqs() < size) {
 672                                         m->mothurOut(lookup[i]->getGroup() + " contains " + toString(lookup[i]->getNumSeqs()) + ". Eliminating."); m->mothurOutEndLine();
 673                                         delete lookup[i];
 674                                 }else {
 675                                         m->Groups.push_back(lookup[i]->getGroup());
 676                                         temp.push_back(lookup[i]);
 677                                 }
 678                         }
 679                         lookup = temp;
 680                         Groups = m->Groups;
 681                 }
 682
 683                 if (lookup.size() == 0) {  m->mothurOut("The size you selected is too large, skipping shared file."); m->mothurOutEndLine(); delete input; return 0; }
 684
 685                 string thisOutputDir = outputDir;
 686                 if (outputDir == "") {  thisOutputDir += m->hasPath(sharedfile);  }
 687                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(sharedfile)) + "subsample" + m->getExtension(sharedfile);
 688
 689                 ofstream out;
 690                 m->openOutputFile(outputFileName, out);
 691                 outputTypes["shared"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 692
 693
 694                 m->mothurOut("Sampling " + toString(size) + " from each group."); m->mothurOutEndLine();
 695
 696                 //as long as you are not at the end of the file or done wih the lines you want
 697                 while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 698                         if (m->control_pressed) {  delete input; for (int i = 0; i < lookup.size(); i++) {  delete lookup[i]; lookup[i] = NULL; } out.close(); return 0;  }
 699
 700                         if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){
 701
 702                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 703
 704                                 processShared(lookup, out);
 705
 706                                 processedLabels.insert(lookup[0]->getLabel());
 707                                 userLabels.erase(lookup[0]->getLabel());
 708                         }
 709
 710                         if ((m->anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 711                                 string saveLabel = lookup[0]->getLabel();
 712
 713                                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 714
 715                                 lookup = input->getSharedRAbundVectors(lastLabel);
 716                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 717
 718                                 processShared(lookup, out);
 719
 720                                 processedLabels.insert(lookup[0]->getLabel());
 721                                 userLabels.erase(lookup[0]->getLabel());
 722
 723                                 //restore real lastlabel to save below
 724                                 lookup[0]->setLabel(saveLabel);
 725                         }
 726
 727                         lastLabel = lookup[0]->getLabel();
 728                         //prevent memory leak
 729                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i]; lookup[i] = NULL; }
 730
 731                         //get next line to process
 732                         lookup = input->getSharedRAbundVectors();
 733                 }
 734
 735
 736                 if (m->control_pressed) {  out.close(); return 0;  }
 737
 738                 //output error messages about any remaining user labels
 739                 set<string>::iterator it;
 740                 bool needToRun = false;
 741                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 742                         m->mothurOut("Your file does not include the label " + *it);
 743                         if (processedLabels.count(lastLabel) != 1) {
 744                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 745                                 needToRun = true;
 746                         }else {
 747                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 748                         }
 749                 }
 750
 751                 //run last label if you need to
 752                 if (needToRun == true)  {
 753                         for (int i = 0; i < lookup.size(); i++) { if (lookup[i] != NULL) { delete lookup[i]; } }
 754                         lookup = input->getSharedRAbundVectors(lastLabel);
 755
 756                         m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 757
 758                         processShared(lookup, out);
 759
 760                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 761                 }
 762
 763                 delete input;
 764                 out.close();
 765
 766                 return 0;
 767
 768         }
 769         catch(exception& e) {
 770                 m->errorOut(e, "SubSampleCommand", "getSubSampleShared");
 771                 exit(1);
 772         }
 773 }
 774 //**********************************************************************************************************************
 775 int SubSampleCommand::processShared(vector<SharedRAbundVector*>& thislookup, ofstream& out) {
 776         try {
 777
 778                 int numBins = thislookup[0]->getNumBins();
 779                 for (int i = 0; i < thislookup.size(); i++) {
 780                         int thisSize = thislookup[i]->getNumSeqs();
 781
 782                         if (thisSize != size) {
 783
 784                                 string thisgroup = thislookup[i]->getGroup();
 785
 786                                 OrderVector* order = new OrderVector();
 787                                 for(int p=0;p<numBins;p++){
 788                                         for(int j=0;j<thislookup[i]->getAbundance(p);j++){
 789                                                 order->push_back(p);
 790                                         }
 791                                 }
 792                                 random_shuffle(order->begin(), order->end());
 793
 794                                 SharedRAbundVector* temp = new SharedRAbundVector(numBins);
 795                                 temp->setLabel(thislookup[i]->getLabel());
 796                                 temp->setGroup(thislookup[i]->getGroup());
 797
 798                                 delete thislookup[i];
 799                                 thislookup[i] = temp;
 800
 801
 802                                 for (int j = 0; j < size; j++) {
 803
 804                                         if (m->control_pressed) { delete order; return 0; }
 805
 806                                         //get random number to sample from order between 0 and thisSize-1.
 807                                         int myrand = int((float)(thisSize) * (float)(rand()) / ((float)RAND_MAX+1.0));
 808
 809                                         int bin = order->get(myrand);
 810
 811                                         int abund = thislookup[i]->getAbundance(bin);
 812                                         thislookup[i]->set(bin, (abund+1), thisgroup);
 813                                 }
 814                                 delete order;
 815                         }
 816                 }
 817
 818                 //subsampling may have created some otus with no sequences in them
 819                 eliminateZeroOTUS(thislookup);
 820
 821                 if (m->control_pressed) { return 0; }
 822
 823                 for (int i = 0; i < thislookup.size(); i++) {
 824                         out << thislookup[i]->getLabel() << '\t' << thislookup[i]->getGroup() << '\t';
 825                         thislookup[i]->print(out);
 826                 }
 827
 828                 return 0;
 829
 830         }
 831         catch(exception& e) {
 832                 m->errorOut(e, "SubSampleCommand", "processShared");
 833                 exit(1);
 834         }
 835 }
 836 //**********************************************************************************************************************
 837 int SubSampleCommand::getSubSampleList() {
 838         try {
 839
 840                 string thisOutputDir = outputDir;
 841                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 842                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "subsample" + m->getExtension(listfile);
 843
 844                 ofstream out;
 845                 m->openOutputFile(outputFileName, out);
 846                 outputTypes["list"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 847
 848                 InputData* input = new InputData(listfile, "list");
 849                 ListVector* list = input->getListVector();
 850                 string lastLabel = list->getLabel();
 851
 852                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 853                 set<string> processedLabels;
 854                 set<string> userLabels = labels;
 855
 856                 ofstream outGroup;
 857                 GroupMap* groupMap;
 858                 if (groupfile != "") {
 859
 860                         groupMap = new GroupMap(groupfile);
 861                         groupMap->readMap();
 862
 863                         //takes care of user setting groupNames that are invalid or setting groups=all
 864                         SharedUtil* util = new SharedUtil();
 865                         util->setGroups(Groups, groupMap->namesOfGroups);
 866                         delete util;
 867
 868                         //create outputfiles
 869                         string groupOutputDir = outputDir;
 870                         if (outputDir == "") {  groupOutputDir += m->hasPath(groupfile);  }
 871                         string groupOutputFileName = groupOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "subsample" + m->getExtension(groupfile);
 872
 873                         m->openOutputFile(groupOutputFileName, outGroup);
 874                         outputTypes["group"].push_back(groupOutputFileName);  outputNames.push_back(groupOutputFileName);
 875
 876                         //file mismatch quit
 877                         if (list->getNumSeqs() != groupMap->getNumSeqs()) {
 878                                 m->mothurOut("[ERROR]: your list file contains " + toString(list->getNumSeqs()) + " sequences, and your groupfile contains " + toString(groupMap->getNumSeqs()) + ", please correct.");
 879                                 m->mothurOutEndLine();
 880                                 delete groupMap;
 881                                 delete list;
 882                                 delete input;
 883                                 out.close();
 884                                 outGroup.close();
 885                                 return 0;
 886                         }
 887                 }
 888
 889                 //make sure that if your picked groups size is not too big
 890                 if (persample) {
 891                         if (size == 0) { //user has not set size, set size = smallest samples size
 892                                 size = groupMap->getNumSeqs(Groups[0]);
 893                                 for (int i = 1; i < Groups.size(); i++) {
 894                                         int thisSize = groupMap->getNumSeqs(Groups[i]);
 895
 896                                         if (thisSize < size) {  size = thisSize;        }
 897                                 }
 898                         }else { //make sure size is not too large
 899                                 vector<string> newGroups;
 900                                 for (int i = 0; i < Groups.size(); i++) {
 901                                         int thisSize = groupMap->getNumSeqs(Groups[i]);
 902
 903                                         if (thisSize >= size) { newGroups.push_back(Groups[i]); }
 904                                         else {  m->mothurOut("You have selected a size that is larger than " + Groups[i] + " number of sequences, removing " + Groups[i] + "."); m->mothurOutEndLine(); }
 905                                 }
 906                                 Groups = newGroups;
 907                         }
 908
 909                         m->mothurOut("Sampling " + toString(size) + " from each group."); m->mothurOutEndLine();
 910                 }else{
 911                         if (pickedGroups) {
 912                                 int total = 0;
 913                                 for(int i = 0; i < Groups.size(); i++) {
 914                                         total += groupMap->getNumSeqs(Groups[i]);
 915                                 }
 916
 917                                 if (size == 0) { //user has not set size, set size = 10% samples size
 918                                         size = int (total * 0.10);
 919                                 }
 920
 921                                 if (total < size) {
 922                                         m->mothurOut("Your size is too large for the number of groups you selected. Adjusting to " + toString(int (total * 0.10)) + "."); m->mothurOutEndLine();
 923                                         size = int (total * 0.10);
 924                                 }
 925
 926                                 m->mothurOut("Sampling " + toString(size) + " from " + toString(total) + "."); m->mothurOutEndLine();
 927                         }else{
 928
 929                                 if (size == 0) { //user has not set size, set size = 10% samples size
 930                                         size = int (list->getNumSeqs() * 0.10);
 931                                 }
 932
 933                                 int thisSize = list->getNumSeqs();
 934                                 if (size > thisSize) { m->mothurOut("Your list file only contains " + toString(thisSize) + " sequences. Setting size to " + toString(thisSize) + "."); m->mothurOutEndLine();
 935                                         size = thisSize;
 936                                 }
 937
 938                                 m->mothurOut("Sampling " + toString(size) + " from " + toString(list->getNumSeqs()) + "."); m->mothurOutEndLine();
 939                         }
 940                 }
 941
 942
 943                 //fill names
 944                 for (int i = 0; i < list->getNumBins(); i++) {
 945                         string binnames = list->get(i);
 946
 947                         //parse names
 948                         string individual = "";
 949                         int length = binnames.length();
 950                         for(int j=0;j<length;j++){
 951                                 if(binnames[j] == ','){
 952
 953                                         if (groupfile != "") { //if there is a groupfile given fill in group info
 954                                                 string group = groupMap->getGroup(individual);
 955                                                 if (group == "not found") { m->mothurOut("[ERROR]: " + individual + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
 956
 957                                                 if (pickedGroups) { //if hte user picked groups, we only want to keep the names of sequences from those groups
 958                                                         if (m->inUsersGroups(group, Groups)) {
 959                                                                 names.push_back(individual);
 960                                                         }
 961                                                 }else{
 962                                                         names.push_back(individual);
 963                                                 }
 964                                         }else{ //save everyone, group
 965                                                 names.push_back(individual);
 966                                         }
 967                                         individual = "";
 968                                 }
 969                                 else{
 970                                         individual += binnames[j];
 971                                 }
 972                         }
 973                         //save last name
 974                         if (groupfile != "") { //if there is a groupfile given fill in group info
 975                                 string group = groupMap->getGroup(individual);
 976                                 if (group == "not found") { m->mothurOut("[ERROR]: " + individual + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
 977
 978                                 if (pickedGroups) { //if hte user picked groups, we only want to keep the names of sequences from those groups
 979                                         if (m->inUsersGroups(group, Groups)) {
 980                                                 names.push_back(individual);
 981                                         }
 982                                 }else{
 983                                         names.push_back(individual);
 984                                 }
 985                         }else{ //save everyone, group
 986                                 names.push_back(individual);
 987                         }
 988                 }
 989
 990                 random_shuffle(names.begin(), names.end());
 991
 992                 //randomly select a subset of those names to include in the subsample
 993                 set<string> subset; //dont want repeat sequence names added
 994                 if (persample) {
 995                         for (int i = 0; i < Groups.size(); i++) {
 996
 997                                 for (int j = 0; j < size; j++) {
 998
 999                                         if (m->control_pressed) { break; }
1000
1001                                         //get random sequence to add, making sure we have not already added it
1002                                         bool done = false;
1003                                         int myrand;
1004                                         while (!done) {
1005                                                 myrand = int((float)(names.size()) * (float)(rand()) / ((float)RAND_MAX+1.0));
1006
1007                                                 if (subset.count(names[myrand]) == 0) { //you are not already added
1008                                                         if (groupMap->getGroup(names[myrand]) == Groups[i])  { subset.insert(names[myrand]); break;     }
1009                                                 }
1010                                         }
1011                                 }
1012                         }
1013                 }else{
1014                         for (int j = 0; j < size; j++) {
1015
1016                                 if (m->control_pressed) { break; }
1017
1018                                 //get random sequence to add, making sure we have not already added it
1019                                 bool done = false;
1020                                 int myrand;
1021                                 while (!done) {
1022                                         myrand = int((float)(names.size()) * (float)(rand()) / ((float)RAND_MAX+1.0));
1023
1024                                         if (subset.count(names[myrand]) == 0)  { subset.insert(names[myrand]); break;   }
1025                                 }
1026                         }
1027                 }
1028
1029                 if (groupfile != "") {
1030                         //write out new groupfile
1031                         for (set<string>::iterator it = subset.begin(); it != subset.end(); it++) {
1032                                 string group = groupMap->getGroup(*it);
1033                                 if (group == "not found") { group = "NOTFOUND"; }
1034
1035                                 outGroup << *it << '\t' << group << endl;
1036                         }
1037                         outGroup.close(); delete groupMap;
1038                 }
1039
1040
1041                 //as long as you are not at the end of the file or done wih the lines you want
1042                 while((list != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
1043
1044                         if (m->control_pressed) {  delete list; delete input; out.close();  return 0;  }
1045
1046                         if(allLines == 1 || labels.count(list->getLabel()) == 1){
1047
1048                                 m->mothurOut(list->getLabel()); m->mothurOutEndLine();
1049
1050                                 processList(list, out, subset);
1051
1052                                 processedLabels.insert(list->getLabel());
1053                                 userLabels.erase(list->getLabel());
1054                         }
1055
1056                         if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
1057                                 string saveLabel = list->getLabel();
1058
1059                                 delete list;
1060
1061                                 list = input->getListVector(lastLabel);
1062                                 m->mothurOut(list->getLabel()); m->mothurOutEndLine();
1063
1064                                 processList(list, out, subset);
1065
1066                                 processedLabels.insert(list->getLabel());
1067                                 userLabels.erase(list->getLabel());
1068
1069                                 //restore real lastlabel to save below
1070                                 list->setLabel(saveLabel);
1071                         }
1072
1073                         lastLabel = list->getLabel();
1074
1075                         delete list; list = NULL;
1076
1077                         //get next line to process
1078                         list = input->getListVector();
1079                 }
1080
1081
1082                 if (m->control_pressed) {  if (list != NULL) { delete list; } delete input; out.close(); return 0;  }
1083
1084                 //output error messages about any remaining user labels
1085                 set<string>::iterator it;
1086                 bool needToRun = false;
1087                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
1088                         m->mothurOut("Your file does not include the label " + *it);
1089                         if (processedLabels.count(lastLabel) != 1) {
1090                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
1091                                 needToRun = true;
1092                         }else {
1093                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
1094                         }
1095                 }
1096
1097                 //run last label if you need to
1098                 if (needToRun == true)  {
1099                         if (list != NULL) { delete list; }
1100
1101                         list = input->getListVector(lastLabel);
1102
1103                         m->mothurOut(list->getLabel()); m->mothurOutEndLine();
1104
1105                         processList(list, out, subset);
1106
1107                         delete list; list = NULL;
1108                 }
1109
1110                 out.close();
1111                 if (list != NULL) { delete list; }
1112                 delete input;
1113
1114                 return 0;
1115
1116         }
1117         catch(exception& e) {
1118                 m->errorOut(e, "SubSampleCommand", "getSubSampleList");
1119                 exit(1);
1120         }
1121 }
1122 //**********************************************************************************************************************
1123 int SubSampleCommand::processList(ListVector*& list, ofstream& out, set<string>& subset) {
1124         try {
1125
1126                 int numBins = list->getNumBins();
1127
1128                 ListVector* temp = new ListVector();
1129                 temp->setLabel(list->getLabel());
1130
1131                 for (int i = 0; i < numBins; i++) {
1132
1133                         if (m->control_pressed) { break; }
1134
1135                         string binnames = list->get(i);
1136
1137                         //parse names
1138                         string individual = "";
1139                         string newNames = "";
1140                         int length = binnames.length();
1141                         for(int j=0;j<length;j++){
1142                                 if(binnames[j] == ','){
1143                                         if (subset.count(individual) != 0) {  newNames += individual + ",";  }
1144                                         individual = "";
1145                                 }else{
1146                                         individual += binnames[j];
1147                                 }
1148                         }
1149                         if (subset.count(individual) != 0) {  newNames += individual + ",";  }
1150
1151
1152                         //if there are names in this bin add to new list
1153                         if (newNames != "") {
1154                                 newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
1155                                 temp->push_back(newNames);
1156                         }
1157                 }
1158
1159                 delete list;
1160                 list = temp;
1161
1162                 if (m->control_pressed) { return 0; }
1163
1164                 list->print(out);
1165
1166                 return 0;
1167
1168         }
1169         catch(exception& e) {
1170                 m->errorOut(e, "SubSampleCommand", "processList");
1171                 exit(1);
1172         }
1173 }
1174 //**********************************************************************************************************************
1175 int SubSampleCommand::getSubSampleRabund() {
1176         try {
1177                 InputData* input = new InputData(rabundfile, "rabund");
1178                 RAbundVector* rabund = input->getRAbundVector();
1179                 string lastLabel = rabund->getLabel();
1180
1181                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
1182                 set<string> processedLabels;
1183                 set<string> userLabels = labels;
1184
1185                 if (size == 0) { //user has not set size, set size = 10%
1186                         size = int((rabund->getNumSeqs()) * 0.10);
1187                 }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; }
1188
1189                 m->mothurOut("Sampling " + toString(size) + " from " + toString(rabund->getNumSeqs()) + "."); m->mothurOutEndLine();
1190
1191                 string thisOutputDir = outputDir;
1192                 if (outputDir == "") {  thisOutputDir += m->hasPath(rabundfile);  }
1193                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(rabundfile)) + "subsample" + m->getExtension(rabundfile);
1194
1195                 ofstream out;
1196                 m->openOutputFile(outputFileName, out);
1197                 outputTypes["rabund"].push_back(outputFileName);  outputNames.push_back(outputFileName);
1198
1199                 //as long as you are not at the end of the file or done wih the lines you want
1200                 while((rabund != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
1201                         if (m->control_pressed) {  delete input; delete rabund; out.close(); return 0;  }
1202
1203                         if(allLines == 1 || labels.count(rabund->getLabel()) == 1){
1204
1205                                 m->mothurOut(rabund->getLabel()); m->mothurOutEndLine();
1206
1207                                 processRabund(rabund, out);
1208
1209                                 processedLabels.insert(rabund->getLabel());
1210                                 userLabels.erase(rabund->getLabel());
1211                         }
1212
1213                         if ((m->anyLabelsToProcess(rabund->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
1214                                 string saveLabel = rabund->getLabel();
1215
1216                                 delete rabund;
1217
1218                                 rabund = input->getRAbundVector(lastLabel);
1219                                 m->mothurOut(rabund->getLabel()); m->mothurOutEndLine();
1220
1221                                 processRabund(rabund, out);
1222
1223                                 processedLabels.insert(rabund->getLabel());
1224                                 userLabels.erase(rabund->getLabel());
1225
1226                                 //restore real lastlabel to save below
1227                                 rabund->setLabel(saveLabel);
1228                         }
1229
1230                         lastLabel = rabund->getLabel();
1231
1232                         //prevent memory leak
1233                         delete rabund; rabund = NULL;
1234
1235                         //get next line to process
1236                         rabund = input->getRAbundVector();
1237                 }
1238
1239
1240                 if (m->control_pressed) {  out.close(); return 0;  }
1241
1242                 //output error messages about any remaining user labels
1243                 set<string>::iterator it;
1244                 bool needToRun = false;
1245                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
1246                         m->mothurOut("Your file does not include the label " + *it);
1247                         if (processedLabels.count(lastLabel) != 1) {
1248                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
1249                                 needToRun = true;
1250                         }else {
1251                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
1252                         }
1253                 }
1254
1255                 //run last label if you need to
1256                 if (needToRun == true)  {
1257                         if (rabund != NULL) { delete rabund; }
1258
1259                         rabund = input->getRAbundVector(lastLabel);
1260
1261                         m->mothurOut(rabund->getLabel()); m->mothurOutEndLine();
1262
1263                         processRabund(rabund, out);
1264
1265                         delete rabund;
1266                 }
1267
1268                 delete input;
1269                 out.close();
1270
1271                 return 0;
1272
1273         }
1274         catch(exception& e) {
1275                 m->errorOut(e, "SubSampleCommand", "getSubSampleRabund");
1276                 exit(1);
1277         }
1278 }
1279 //**********************************************************************************************************************
1280 int SubSampleCommand::processRabund(RAbundVector*& rabund, ofstream& out) {
1281         try {
1282
1283                 int numBins = rabund->getNumBins();
1284                 int thisSize = rabund->getNumSeqs();
1285
1286                 if (thisSize != size) {
1287
1288                         OrderVector* order = new OrderVector();
1289                         for(int p=0;p<numBins;p++){
1290                                 for(int j=0;j<rabund->get(p);j++){
1291                                         order->push_back(p);
1292                                 }
1293                         }
1294                         random_shuffle(order->begin(), order->end());
1295
1296                         RAbundVector* temp = new RAbundVector(numBins);
1297                         temp->setLabel(rabund->getLabel());
1298
1299                         delete rabund;
1300                         rabund = temp;
1301
1302                         for (int j = 0; j < size; j++) {
1303
1304                                 if (m->control_pressed) { delete order; return 0; }
1305
1306                                 //get random number to sample from order between 0 and thisSize-1.
1307                                 int myrand = int((float)(thisSize) * (float)(rand()) / ((float)RAND_MAX+1.0));
1308
1309                                 int bin = order->get(myrand);
1310
1311                                 int abund = rabund->get(bin);
1312                                 rabund->set(bin, (abund+1));
1313                         }
1314
1315                         delete order;
1316                 }
1317
1318                 if (m->control_pressed) { return 0; }
1319
1320                 rabund->print(out);
1321
1322                 return 0;
1323
1324         }
1325         catch(exception& e) {
1326                 m->errorOut(e, "SubSampleCommand", "processRabund");
1327                 exit(1);
1328         }
1329 }
1330 //**********************************************************************************************************************
1331 int SubSampleCommand::getSubSampleSabund() {
1332         try {
1333
1334                 InputData* input = new InputData(sabundfile, "sabund");
1335                 SAbundVector* sabund = input->getSAbundVector();
1336                 string lastLabel = sabund->getLabel();
1337
1338                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
1339                 set<string> processedLabels;
1340                 set<string> userLabels = labels;
1341
1342                 if (size == 0) { //user has not set size, set size = 10%
1343                         size = int((sabund->getNumSeqs()) * 0.10);
1344                 }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; }
1345
1346
1347                 m->mothurOut("Sampling " + toString(size) + " from " + toString(sabund->getNumSeqs()) + "."); m->mothurOutEndLine();
1348
1349                 string thisOutputDir = outputDir;
1350                 if (outputDir == "") {  thisOutputDir += m->hasPath(sabundfile);  }
1351                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(sabundfile)) + "subsample" + m->getExtension(sabundfile);
1352
1353                 ofstream out;
1354                 m->openOutputFile(outputFileName, out);
1355                 outputTypes["sabund"].push_back(outputFileName);  outputNames.push_back(outputFileName);
1356
1357
1358                 //as long as you are not at the end of the file or done wih the lines you want
1359                 while((sabund != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
1360                         if (m->control_pressed) {  delete input; delete sabund; out.close(); return 0;  }
1361
1362                         if(allLines == 1 || labels.count(sabund->getLabel()) == 1){
1363
1364                                 m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
1365
1366                                 processSabund(sabund, out);
1367
1368                                 processedLabels.insert(sabund->getLabel());
1369                                 userLabels.erase(sabund->getLabel());
1370                         }
1371
1372                         if ((m->anyLabelsToProcess(sabund->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
1373                                 string saveLabel = sabund->getLabel();
1374
1375                                 delete sabund;
1376
1377                                 sabund = input->getSAbundVector(lastLabel);
1378                                 m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
1379
1380                                 processSabund(sabund, out);
1381
1382                                 processedLabels.insert(sabund->getLabel());
1383                                 userLabels.erase(sabund->getLabel());
1384
1385                                 //restore real lastlabel to save below
1386                                 sabund->setLabel(saveLabel);
1387                         }
1388
1389                         lastLabel = sabund->getLabel();
1390
1391                         //prevent memory leak
1392                         delete sabund; sabund = NULL;
1393
1394                         //get next line to process
1395                         sabund = input->getSAbundVector();
1396                 }
1397
1398
1399                 if (m->control_pressed) {  out.close(); return 0;  }
1400
1401                 //output error messages about any remaining user labels
1402                 set<string>::iterator it;
1403                 bool needToRun = false;
1404                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
1405                         m->mothurOut("Your file does not include the label " + *it);
1406                         if (processedLabels.count(lastLabel) != 1) {
1407                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
1408                                 needToRun = true;
1409                         }else {
1410                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
1411                         }
1412                 }
1413
1414                 //run last label if you need to
1415                 if (needToRun == true)  {
1416                         if (sabund != NULL) { delete sabund; }
1417
1418                         sabund = input->getSAbundVector(lastLabel);
1419
1420                         m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
1421
1422                         processSabund(sabund, out);
1423
1424                         delete sabund;
1425                 }
1426
1427                 delete input;
1428                 out.close();
1429
1430                 return 0;
1431
1432         }
1433         catch(exception& e) {
1434                 m->errorOut(e, "SubSampleCommand", "getSubSampleSabund");
1435                 exit(1);
1436         }
1437 }
1438 //**********************************************************************************************************************
1439 int SubSampleCommand::processSabund(SAbundVector*& sabund, ofstream& out) {
1440         try {
1441
1442                 RAbundVector* rabund = new RAbundVector();
1443                 *rabund = sabund->getRAbundVector();
1444
1445                 int numBins = rabund->getNumBins();
1446                 int thisSize = rabund->getNumSeqs();
1447
1448                 if (thisSize != size) {
1449
1450                         OrderVector* order = new OrderVector();
1451                         for(int p=0;p<numBins;p++){
1452                                 for(int j=0;j<rabund->get(p);j++){
1453                                         order->push_back(p);
1454                                 }
1455                         }
1456                         random_shuffle(order->begin(), order->end());
1457
1458                         RAbundVector* temp = new RAbundVector(numBins);
1459                         temp->setLabel(rabund->getLabel());
1460
1461                         delete rabund;
1462                         rabund = temp;
1463
1464                         for (int j = 0; j < size; j++) {
1465
1466                                 if (m->control_pressed) { delete order; return 0; }
1467
1468                                 //get random number to sample from order between 0 and thisSize-1.
1469                                 int myrand = int((float)(thisSize) * (float)(rand()) / ((float)RAND_MAX+1.0));
1470
1471                                 int bin = order->get(myrand);
1472
1473                                 int abund = rabund->get(bin);
1474                                 rabund->set(bin, (abund+1));
1475                         }
1476
1477                         delete order;
1478                 }
1479
1480                 if (m->control_pressed) { return 0; }
1481
1482                 delete sabund;
1483                 sabund = new SAbundVector();
1484                 *sabund = rabund->getSAbundVector();
1485                 delete rabund;
1486
1487                 sabund->print(out);
1488
1489                 return 0;
1490
1491         }
1492         catch(exception& e) {
1493                 m->errorOut(e, "SubSampleCommand", "processSabund");
1494                 exit(1);
1495         }
1496 }
1497 //**********************************************************************************************************************
1498 int SubSampleCommand::eliminateZeroOTUS(vector<SharedRAbundVector*>& thislookup) {
1499         try {
1500
1501                 vector<SharedRAbundVector*> newLookup;
1502                 for (int i = 0; i < thislookup.size(); i++) {
1503                         SharedRAbundVector* temp = new SharedRAbundVector();
1504                         temp->setLabel(thislookup[i]->getLabel());
1505                         temp->setGroup(thislookup[i]->getGroup());
1506                         newLookup.push_back(temp);
1507                 }
1508
1509                 //for each bin
1510                 for (int i = 0; i < thislookup[0]->getNumBins(); i++) {
1511                         if (m->control_pressed) { for (int j = 0; j < newLookup.size(); j++) {  delete newLookup[j];  } return 0; }
1512
1513                         //look at each sharedRabund and make sure they are not all zero
1514                         bool allZero = true;
1515                         for (int j = 0; j < thislookup.size(); j++) {
1516                                 if (thislookup[j]->getAbundance(i) != 0) { allZero = false;  break;  }
1517                         }
1518
1519                         //if they are not all zero add this bin
1520                         if (!allZero) {
1521                                 for (int j = 0; j < thislookup.size(); j++) {
1522                                         newLookup[j]->push_back(thislookup[j]->getAbundance(i), thislookup[j]->getGroup());
1523                                 }
1524                         }
1525                 }
1526
1527                 for (int j = 0; j < thislookup.size(); j++) {  delete thislookup[j];  }
1528                 thislookup.clear();
1529
1530                 thislookup = newLookup;
1531
1532                 return 0;
1533
1534         }
1535         catch(exception& e) {
1536                 m->errorOut(e, "SubSampleCommand", "eliminateZeroOTUS");
1537                 exit(1);
1538         }
1539 }
1540
1541 //**********************************************************************************************************************
1542
1543
1544