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::getValidParameters(){
  15         try {
  16                 string Array[] =  {"fasta", "group", "list","shared","rabund", "name","sabund","size","groups","label","outputdir","inputdir"};
  17                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  18                 return myArray;
  19         }
  20         catch(exception& e) {
  21                 m->errorOut(e, "SubSampleCommand", "getValidParameters");
  22                 exit(1);
  23         }
  24 }
  25 //**********************************************************************************************************************
  26 SubSampleCommand::SubSampleCommand(){
  27         try {
  28                 abort = true;
  29                 //initialize outputTypes
  30                 vector<string> tempOutNames;
  31                 outputTypes["shared"] = tempOutNames;
  32                 outputTypes["list"] = tempOutNames;
  33                 outputTypes["rabund"] = tempOutNames;
  34                 outputTypes["sabund"] = tempOutNames;
  35                 outputTypes["fasta"] = tempOutNames;
  36                 outputTypes["name"] = tempOutNames;
  37                 outputTypes["group"] = tempOutNames;
  38         }
  39         catch(exception& e) {
  40                 m->errorOut(e, "SubSampleCommand", "GetRelAbundCommand");
  41                 exit(1);
  42         }
  43 }
  44 //**********************************************************************************************************************
  45 vector<string> SubSampleCommand::getRequiredParameters(){
  46         try {
  47                 string Array[] =  {"fasta","list","shared","rabund", "sabund","or"};
  48                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  49                 return myArray;
  50         }
  51         catch(exception& e) {
  52                 m->errorOut(e, "SubSampleCommand", "getRequiredParameters");
  53                 exit(1);
  54         }
  55 }
  56 //**********************************************************************************************************************
  57 vector<string> SubSampleCommand::getRequiredFiles(){
  58         try {
  59                 vector<string> myArray;
  60                 return myArray;
  61         }
  62         catch(exception& e) {
  63                 m->errorOut(e, "SubSampleCommand", "getRequiredFiles");
  64                 exit(1);
  65         }
  66 }
  67 //**********************************************************************************************************************
  68 SubSampleCommand::SubSampleCommand(string option) {
  69         try {
  70                 globaldata = GlobalData::getInstance();
  71                 abort = false;
  72                 allLines = 1;
  73                 labels.clear();
  74
  75                 //allow user to run help
  76                 if(option == "help") { help(); abort = true; }
  77
  78                 else {
  79                         //valid paramters for this command
  80                         string Array[] =  {"fasta", "group", "list","shared","rabund", "sabund","name","size","groups","label","outputdir","inputdir"};
  81                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  82
  83                         OptionParser parser(option);
  84                         map<string,string> parameters = parser.getParameters();
  85
  86                         ValidParameters validParameter;
  87
  88                         //check to make sure all parameters are valid for command
  89                         map<string,string>::iterator it;
  90                         for (it = parameters.begin(); it != parameters.end(); it++) {
  91                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  92                         }
  93
  94                         //initialize outputTypes
  95                         vector<string> tempOutNames;
  96                         outputTypes["shared"] = tempOutNames;
  97                         outputTypes["list"] = tempOutNames;
  98                         outputTypes["rabund"] = tempOutNames;
  99                         outputTypes["sabund"] = tempOutNames;
 100                         outputTypes["fasta"] = tempOutNames;
 101                         outputTypes["name"] = tempOutNames;
 102                         outputTypes["group"] = tempOutNames;
 103
 104                         //if the user changes the output directory command factory will send this info to us in the output parameter
 105                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
 106
 107                         //if the user changes the input directory command factory will send this info to us in the output parameter
 108                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 109                         if (inputDir == "not found"){   inputDir = "";          }
 110                         else {
 111                                 string path;
 112                                 it = parameters.find("list");
 113                                 //user has given a template file
 114                                 if(it != parameters.end()){
 115                                         path = m->hasPath(it->second);
 116                                         //if the user has not given a path then, add inputdir. else leave path alone.
 117                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 118                                 }
 119
 120                                 it = parameters.find("fasta");
 121                                 //user has given a template file
 122                                 if(it != parameters.end()){
 123                                         path = m->hasPath(it->second);
 124                                         //if the user has not given a path then, add inputdir. else leave path alone.
 125                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 126                                 }
 127
 128                                 it = parameters.find("shared");
 129                                 //user has given a template file
 130                                 if(it != parameters.end()){
 131                                         path = m->hasPath(it->second);
 132                                         //if the user has not given a path then, add inputdir. else leave path alone.
 133                                         if (path == "") {       parameters["shared"] = inputDir + it->second;           }
 134                                 }
 135
 136                                 it = parameters.find("group");
 137                                 //user has given a template file
 138                                 if(it != parameters.end()){
 139                                         path = m->hasPath(it->second);
 140                                         //if the user has not given a path then, add inputdir. else leave path alone.
 141                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 142                                 }
 143
 144                                 it = parameters.find("sabund");
 145                                 //user has given a template file
 146                                 if(it != parameters.end()){
 147                                         path = m->hasPath(it->second);
 148                                         //if the user has not given a path then, add inputdir. else leave path alone.
 149                                         if (path == "") {       parameters["sabund"] = inputDir + it->second;           }
 150                                 }
 151
 152                                 it = parameters.find("rabund");
 153                                 //user has given a template file
 154                                 if(it != parameters.end()){
 155                                         path = m->hasPath(it->second);
 156                                         //if the user has not given a path then, add inputdir. else leave path alone.
 157                                         if (path == "") {       parameters["rabund"] = inputDir + it->second;           }
 158                                 }
 159
 160                                 it = parameters.find("name");
 161                                 //user has given a template file
 162                                 if(it != parameters.end()){
 163                                         path = m->hasPath(it->second);
 164                                         //if the user has not given a path then, add inputdir. else leave path alone.
 165                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 166                                 }
 167                         }
 168
 169                         //check for required parameters
 170                         listfile = validParameter.validFile(parameters, "list", true);
 171                         if (listfile == "not open") { listfile = ""; abort = true; }
 172                         else if (listfile == "not found") { listfile = ""; }
 173
 174                         sabundfile = validParameter.validFile(parameters, "sabund", true);
 175                         if (sabundfile == "not open") { sabundfile = ""; abort = true; }
 176                         else if (sabundfile == "not found") { sabundfile = ""; }
 177
 178                         rabundfile = validParameter.validFile(parameters, "rabund", true);
 179                         if (rabundfile == "not open") { rabundfile = ""; abort = true; }
 180                         else if (rabundfile == "not found") { rabundfile = ""; }
 181
 182                         fastafile = validParameter.validFile(parameters, "fasta", true);
 183                         if (fastafile == "not open") { fastafile = ""; abort = true; }
 184                         else if (fastafile == "not found") { fastafile = ""; }
 185
 186                         sharedfile = validParameter.validFile(parameters, "shared", true);
 187                         if (sharedfile == "not open") { sharedfile = ""; abort = true; }
 188                         else if (sharedfile == "not found") { sharedfile = ""; }
 189
 190                         namefile = validParameter.validFile(parameters, "name", true);
 191                         if (namefile == "not open") { namefile = ""; abort = true; }
 192                         else if (namefile == "not found") { namefile = ""; }
 193
 194                         groupfile = validParameter.validFile(parameters, "group", true);
 195                         if (groupfile == "not open") { groupfile = ""; abort = true; }
 196                         else if (groupfile == "not found") { groupfile = ""; }
 197
 198
 199                         //check for optional parameter and set defaults
 200                         // ...at some point should added some additional type checking...
 201                         label = validParameter.validFile(parameters, "label", false);
 202                         if (label == "not found") { label = ""; }
 203                         else {
 204                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 205                                 else { allLines = 1;  }
 206                         }
 207
 208                         groups = validParameter.validFile(parameters, "groups", false);
 209                         if (groups == "not found") { groups = ""; pickedGroups = false; }
 210                         else {
 211                                 pickedGroups = true;
 212                                 m->splitAtDash(groups, Groups);
 213                                 globaldata->Groups = Groups;
 214                         }
 215
 216                         string temp = validParameter.validFile(parameters, "size", false);              if (temp == "not found"){       temp = "0";             }
 217                         convert(temp, size);
 218
 219                         if ((namefile != "") && (fastafile == "")) { m->mothurOut("You may only use a namefile with a fastafile."); m->mothurOutEndLine(); abort = true; }
 220
 221                         if ((fastafile == "") && (listfile == "") && (sabundfile == "") && (rabundfile == "") && (sharedfile == "")) {
 222                                 m->mothurOut("You must provide a fasta, list, sabund, rabund or shared file as an input file."); m->mothurOutEndLine(); abort = true; }
 223
 224                         if (pickedGroups && ((groupfile == "") && (sharedfile == ""))) {
 225                                 m->mothurOut("You cannot pick groups without a valid group file or shared file."); m->mothurOutEndLine(); abort = true; }
 226
 227                         if ((groupfile != "") && ((fastafile == "") && (listfile == ""))) {
 228                                 m->mothurOut("Group file only valid with listfile or fastafile."); m->mothurOutEndLine(); abort = true; }
 229
 230                         if ((groupfile != "") && ((fastafile != "") && (listfile != ""))) {
 231                                 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; }
 232
 233                 }
 234
 235         }
 236         catch(exception& e) {
 237                 m->errorOut(e, "SubSampleCommand", "SubSampleCommand");
 238                 exit(1);
 239         }
 240 }
 241
 242 //**********************************************************************************************************************
 243
 244 void SubSampleCommand::help(){
 245         try {
 246                 m->mothurOut("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");
 247                 m->mothurOut("The sub.sample command parameters are fasta, name, list, group, rabund, sabund, shared, groups, size and label.  You must provide a fasta, list, sabund, rabund or shared file as an input file.\n");
 248                 m->mothurOut("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");
 249                 m->mothurOut("The label parameter allows you to select what distance levels you would like, and are also separated by dashes.\n");
 250                 m->mothurOut("The size parameter allows you indicate the size of your subsample.\n");
 251                 m->mothurOut("The size parameter is not set: with shared file size=number of seqs in smallest sample, with all other files, 10% of number of seqs.\n");
 252                 m->mothurOut("The sub.sample command should be in the following format: sub.sample(list=yourListFile, group=yourGroupFile, groups=yourGroups, label=yourLabels).\n");
 253                 m->mothurOut("Example sub.sample(list=abrecovery.fn.list, group=abrecovery.groups, groups=B-C, size=20).\n");
 254                 m->mothurOut("The default value for groups is all the groups in your groupfile, and all labels in your inputfile will be used.\n");
 255                 m->mothurOut("The sub.sample command outputs a .subsample file.\n");
 256                 m->mothurOut("Note: No spaces between parameter labels (i.e. groups), '=' and parameters (i.e.yourGroups).\n\n");
 257
 258         }
 259         catch(exception& e) {
 260                 m->errorOut(e, "SubSampleCommand", "help");
 261                 exit(1);
 262         }
 263 }
 264
 265 //**********************************************************************************************************************
 266
 267 SubSampleCommand::~SubSampleCommand(){}
 268
 269 //**********************************************************************************************************************
 270
 271 int SubSampleCommand::execute(){
 272         try {
 273
 274                 if (abort == true) { return 0; }
 275
 276                 if (sharedfile != "")   {   getSubSampleShared();       }
 277                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       remove(outputNames[i].c_str()); return 0; } }
 278
 279                 if (listfile != "")             {   getSubSampleList();         }
 280                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       remove(outputNames[i].c_str()); return 0; } }
 281
 282                 if (rabundfile != "")   {   getSubSampleRabund();       }
 283                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       remove(outputNames[i].c_str()); return 0; } }
 284
 285                 if (sabundfile != "")   {   getSubSampleSabund();       }
 286                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       remove(outputNames[i].c_str()); return 0; } }
 287
 288                 if (fastafile != "")    {   getSubSampleFasta();        }
 289                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       remove(outputNames[i].c_str()); return 0; } }
 290
 291
 292                 m->mothurOutEndLine();
 293                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 294                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 295                 m->mothurOutEndLine();
 296
 297                 return 0;
 298         }
 299         catch(exception& e) {
 300                 m->errorOut(e, "SubSampleCommand", "execute");
 301                 exit(1);
 302         }
 303 }
 304 //**********************************************************************************************************************
 305 int SubSampleCommand::getSubSampleFasta() {
 306         try {
 307
 308                 if (namefile != "") { readNames(); }    //fills names with all names in namefile.
 309                 else { getNames(); }//no name file, so get list of names to pick from
 310
 311                 GroupMap* groupMap;
 312                 if (groupfile != "") {
 313
 314                         groupMap = new GroupMap(groupfile);
 315                         groupMap->readMap();
 316
 317                         //takes care of user setting groupNames that are invalid or setting groups=all
 318                         SharedUtil* util = new SharedUtil();
 319                         util->setGroups(Groups, groupMap->namesOfGroups);
 320                         delete util;
 321
 322                         //file mismatch quit
 323                         if (names.size() != groupMap->getNumSeqs()) {
 324                                 m->mothurOut("[ERROR]: your fasta file contains " + toString(names.size()) + " sequences, and your groupfile contains " + toString(groupMap->getNumSeqs()) + ", please correct.");
 325                                 m->mothurOutEndLine();
 326                                 delete groupMap;
 327                                 return 0;
 328                         }
 329                 }
 330
 331                 if (m->control_pressed) { return 0; }
 332
 333                 string thisOutputDir = outputDir;
 334                 if (outputDir == "") {  thisOutputDir += m->hasPath(fastafile);  }
 335                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(fastafile)) + "subsample" + m->getExtension(fastafile);
 336
 337                 ofstream out;
 338                 m->openOutputFile(outputFileName, out);
 339                 outputTypes["fasta"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 340
 341                 //make sure that if your picked groups size is not too big
 342
 343                 if (pickedGroups) {
 344                         int total = 0;
 345                         for(int i = 0; i < Groups.size(); i++) {
 346                                 total += groupMap->getNumSeqs(Groups[i]);
 347                         }
 348
 349                         if (size == 0) { //user has not set size, set size = 10% samples size
 350                                 size = int (total * 0.10);
 351                         }
 352
 353                         if (total < size) {
 354                                 if (size != 0) {
 355                                         m->mothurOut("Your size is too large for the number of groups you selected. Adjusting to " + toString(int (total * 0.10)) + "."); m->mothurOutEndLine();
 356                                 }
 357                                 size = int (total * 0.10);
 358                         }
 359
 360                         m->mothurOut("Sampling " + toString(size) + " from " + toString(total) + "."); m->mothurOutEndLine();
 361                 }
 362
 363                 if (size == 0) { //user has not set size, set size = 10% samples size
 364                         size = int (names.size() * 0.10);
 365                 }
 366
 367                 int thisSize = names.size();
 368                 if (size > thisSize) { m->mothurOut("Your fasta file only contains " + toString(thisSize) + " sequences. Setting size to " + toString(thisSize) + "."); m->mothurOutEndLine();
 369                         size = thisSize;
 370                 }
 371
 372                 random_shuffle(names.begin(), names.end());
 373
 374                 if (!pickedGroups) { m->mothurOut("Sampling " + toString(size) + " from " + toString(thisSize) + "."); m->mothurOutEndLine(); }
 375
 376                 //randomly select a subset of those names to include in the subsample
 377                 set<string> subset; //dont want repeat sequence names added
 378                 for (int j = 0; j < size; j++) {
 379
 380                         if (m->control_pressed) { return 0; }
 381
 382                         //get random sequence to add, making sure we have not already added it
 383                         bool done = false;
 384                         int myrand;
 385                         while (!done) {
 386                                 myrand = (int)((float)(rand()) / (RAND_MAX / (thisSize-1) + 1));
 387
 388                                 if (subset.count(names[myrand]) == 0)  {
 389
 390                                         if (groupfile != "") { //if there is a groupfile given fill in group info
 391                                                 string group = groupMap->getGroup(names[myrand]);
 392                                                 if (group == "not found") { m->mothurOut("[ERROR]: " + names[myrand] + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
 393
 394                                                 if (pickedGroups) { //if hte user picked groups, we only want to keep the names of sequences from those groups
 395                                                         if (m->inUsersGroups(group, Groups)) {
 396                                                                 subset.insert(names[myrand]); break;
 397                                                         }
 398                                                 }else{
 399                                                         subset.insert(names[myrand]); break;
 400                                                 }
 401                                         }else{ //save everyone, group
 402                                                 subset.insert(names[myrand]); break;
 403                                         }
 404                                 }
 405                         }
 406                 }
 407
 408                 //read through fasta file outputting only the names on the subsample list
 409                 ifstream in;
 410                 m->openInputFile(fastafile, in);
 411
 412                 string thisname;
 413                 int count = 0;
 414                 map<string, vector<string> >::iterator itNameMap;
 415
 416                 while(!in.eof()){
 417
 418                         if (m->control_pressed) { in.close(); out.close();  return 0; }
 419
 420                         Sequence currSeq(in);
 421                         thisname = currSeq.getName();
 422
 423                         if (thisname != "") {
 424
 425                                 //does the subset contain a sequence that this sequence represents
 426                                 itNameMap = nameMap.find(thisname);
 427                                 if (itNameMap != nameMap.end()) {
 428                                         vector<string> nameRepresents = itNameMap->second;
 429
 430                                         for (int i = 0; i < nameRepresents.size(); i++){
 431                                                 if (subset.count(nameRepresents[i]) != 0) {
 432                                                         out << ">" << nameRepresents[i] << endl << currSeq.getAligned() << endl;
 433                                                         count++;
 434                                                 }
 435                                         }
 436                                 }else{
 437                                         m->mothurOut("[ERROR]: " + thisname + " is not in your namefile, please correct."); m->mothurOutEndLine();
 438                                 }
 439                         }
 440                         m->gobble(in);
 441                 }
 442                 in.close();
 443                 out.close();
 444
 445                 if (count != subset.size()) {
 446                         m->mothurOut("[ERROR]: The subset selected contained " + toString(subset.size()) + " sequences, but I only found " + toString(count) + " of those in the fastafile."); m->mothurOutEndLine();
 447                 }
 448
 449                 //if a groupfile is provided read through the group file only outputting the names on the subsample list
 450                 if (groupfile != "") {
 451
 452                         string groupOutputDir = outputDir;
 453                         if (outputDir == "") {  groupOutputDir += m->hasPath(groupfile);  }
 454                         string groupOutputFileName = groupOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "subsample" + m->getExtension(groupfile);
 455
 456                         ofstream outGroup;
 457                         m->openOutputFile(groupOutputFileName, outGroup);
 458                         outputTypes["group"].push_back(groupOutputFileName);  outputNames.push_back(groupOutputFileName);
 459
 460                         ifstream inGroup;
 461                         m->openInputFile(groupfile, inGroup);
 462                         string name, group;
 463
 464                         while(!inGroup.eof()){
 465
 466                                 if (m->control_pressed) { inGroup.close(); outGroup.close(); return 0; }
 467
 468                                 inGroup >> name;        m->gobble(inGroup);                     //read from first column
 469                                 inGroup >> group;                       //read from second column
 470
 471                                 //if this name is in the accnos file
 472                                 if (subset.count(name) != 0) {
 473                                         outGroup << name << '\t' << group << endl;
 474                                         subset.erase(name);
 475                                 }
 476
 477                                 m->gobble(inGroup);
 478                         }
 479                         inGroup.close();
 480                         outGroup.close();
 481
 482                         //sanity check
 483                         if (subset.size() != 0) {
 484                                 m->mothurOut("Your groupfile does not match your fasta file."); m->mothurOutEndLine();
 485                                 for (set<string>::iterator it = subset.begin(); it != subset.end(); it++) {
 486                                         m->mothurOut("[ERROR]: " + *it + " is missing from your groupfile."); m->mothurOutEndLine();
 487                                 }
 488                         }
 489                 }
 490
 491                 return 0;
 492
 493         }
 494         catch(exception& e) {
 495                 m->errorOut(e, "SubSampleCommand", "getSubSampleFasta");
 496                 exit(1);
 497         }
 498 }
 499 //**********************************************************************************************************************
 500 int SubSampleCommand::getNames() {
 501         try {
 502
 503                 ifstream in;
 504                 m->openInputFile(fastafile, in);
 505
 506                 string thisname;
 507                 while(!in.eof()){
 508
 509                         if (m->control_pressed) { in.close(); return 0; }
 510
 511                         Sequence currSeq(in);
 512                         thisname = currSeq.getName();
 513
 514                         if (thisname != "") {
 515                                 vector<string> temp; temp.push_back(thisname);
 516                                 nameMap[thisname] = temp;
 517                                 names.push_back(thisname);
 518                         }
 519                         m->gobble(in);
 520                 }
 521                 in.close();
 522
 523         }
 524         catch(exception& e) {
 525                 m->errorOut(e, "SubSampleCommand", "getNames");
 526                 exit(1);
 527         }
 528 }
 529 //**********************************************************************************************************************
 530 int SubSampleCommand::readNames() {
 531         try {
 532
 533                 ifstream in;
 534                 m->openInputFile(namefile, in);
 535
 536                 string thisname, repnames;
 537                 map<string, vector<string> >::iterator it;
 538
 539                 while(!in.eof()){
 540
 541                         if (m->control_pressed) { in.close(); return 0; }
 542
 543                         in >> thisname;         m->gobble(in);          //read from first column
 544                         in >> repnames;                 //read from second column
 545
 546                         it = nameMap.find(thisname);
 547                         if (it == nameMap.end()) {
 548
 549                                 vector<string> splitRepNames;
 550                                 m->splitAtComma(repnames, splitRepNames);
 551
 552                                 nameMap[thisname] = splitRepNames;
 553                                 for (int i = 0; i < splitRepNames.size(); i++) { names.push_back(splitRepNames[i]); }
 554
 555                         }else{  m->mothurOut(thisname + " is already in namesfile. I will use first definition."); m->mothurOutEndLine();  }
 556
 557                         m->gobble(in);
 558                 }
 559                 in.close();
 560
 561         }
 562         catch(exception& e) {
 563                 m->errorOut(e, "SubSampleCommand", "readNames");
 564                 exit(1);
 565         }
 566 }
 567 //**********************************************************************************************************************
 568 int SubSampleCommand::getSubSampleShared() {
 569         try {
 570
 571                 string thisOutputDir = outputDir;
 572                 if (outputDir == "") {  thisOutputDir += m->hasPath(sharedfile);  }
 573                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(sharedfile)) + "subsample" + m->getExtension(sharedfile);
 574
 575                 ofstream out;
 576                 m->openOutputFile(outputFileName, out);
 577                 outputTypes["shared"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 578
 579                 InputData* input = new InputData(sharedfile, "sharedfile");
 580                 vector<SharedRAbundVector*> lookup = input->getSharedRAbundVectors();
 581                 string lastLabel = lookup[0]->getLabel();
 582
 583                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 584                 set<string> processedLabels;
 585                 set<string> userLabels = labels;
 586
 587                 if (size == 0) { //user has not set size, set size = smallest samples size
 588                         size = lookup[0]->getNumSeqs();
 589                         for (int i = 1; i < lookup.size(); i++) {
 590                                 int thisSize = lookup[i]->getNumSeqs();
 591
 592                                 if (thisSize < size) {  size = thisSize;        }
 593                         }
 594                 }
 595
 596                 m->mothurOut("Sampling " + toString(size) + " from " + toString(lookup[0]->getNumSeqs()) + "."); m->mothurOutEndLine();
 597
 598                 //as long as you are not at the end of the file or done wih the lines you want
 599                 while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 600                         if (m->control_pressed) {  delete input; for (int i = 0; i < lookup.size(); i++) {  delete lookup[i]; lookup[i] = NULL; } out.close(); return 0;  }
 601
 602                         if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){
 603
 604                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 605
 606                                 processShared(lookup, out);
 607
 608                                 processedLabels.insert(lookup[0]->getLabel());
 609                                 userLabels.erase(lookup[0]->getLabel());
 610                         }
 611
 612                         if ((m->anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 613                                 string saveLabel = lookup[0]->getLabel();
 614
 615                                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 616
 617                                 lookup = input->getSharedRAbundVectors(lastLabel);
 618                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 619
 620                                 processShared(lookup, out);
 621
 622                                 processedLabels.insert(lookup[0]->getLabel());
 623                                 userLabels.erase(lookup[0]->getLabel());
 624
 625                                 //restore real lastlabel to save below
 626                                 lookup[0]->setLabel(saveLabel);
 627                         }
 628
 629                         lastLabel = lookup[0]->getLabel();
 630                         //prevent memory leak
 631                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i]; lookup[i] = NULL; }
 632
 633                         //get next line to process
 634                         lookup = input->getSharedRAbundVectors();
 635                 }
 636
 637
 638                 if (m->control_pressed) {  out.close(); return 0;  }
 639
 640                 //output error messages about any remaining user labels
 641                 set<string>::iterator it;
 642                 bool needToRun = false;
 643                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 644                         m->mothurOut("Your file does not include the label " + *it);
 645                         if (processedLabels.count(lastLabel) != 1) {
 646                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 647                                 needToRun = true;
 648                         }else {
 649                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 650                         }
 651                 }
 652
 653                 //run last label if you need to
 654                 if (needToRun == true)  {
 655                         for (int i = 0; i < lookup.size(); i++) { if (lookup[i] != NULL) { delete lookup[i]; } }
 656                         lookup = input->getSharedRAbundVectors(lastLabel);
 657
 658                         m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 659
 660                         processShared(lookup, out);
 661
 662                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 663                 }
 664
 665                 delete input;
 666                 out.close();
 667
 668                 return 0;
 669
 670         }
 671         catch(exception& e) {
 672                 m->errorOut(e, "SubSampleCommand", "getSubSampleShared");
 673                 exit(1);
 674         }
 675 }
 676 //**********************************************************************************************************************
 677 int SubSampleCommand::processShared(vector<SharedRAbundVector*>& thislookup, ofstream& out) {
 678         try {
 679
 680                 int numBins = thislookup[0]->getNumBins();
 681                 for (int i = 0; i < thislookup.size(); i++) {
 682                         int thisSize = thislookup[i]->getNumSeqs();
 683
 684                         if (thisSize != size) {
 685
 686                                 string thisgroup = thislookup[i]->getGroup();
 687
 688                                 OrderVector* order = new OrderVector();
 689                                 for(int p=0;p<numBins;p++){
 690                                         for(int j=0;j<thislookup[i]->getAbundance(p);j++){
 691                                                 order->push_back(p);
 692                                         }
 693                                 }
 694                                 random_shuffle(order->begin(), order->end());
 695
 696                                 SharedRAbundVector* temp = new SharedRAbundVector(numBins);
 697                                 temp->setLabel(thislookup[i]->getLabel());
 698                                 temp->setGroup(thislookup[i]->getGroup());
 699
 700                                 delete thislookup[i];
 701                                 thislookup[i] = temp;
 702
 703
 704                                 for (int j = 0; j < size; j++) {
 705
 706                                         if (m->control_pressed) { delete order; return 0; }
 707
 708                                         //get random number to sample from order between 0 and thisSize-1.
 709                                         int myrand = (int)((float)(rand()) / (RAND_MAX / (thisSize-1) + 1));
 710
 711                                         int bin = order->get(myrand);
 712
 713                                         int abund = thislookup[i]->getAbundance(bin);
 714                                         thislookup[i]->set(bin, (abund+1), thisgroup);
 715                                 }
 716                                 delete order;
 717                         }
 718                 }
 719
 720                 //subsampling may have created some otus with no sequences in them
 721                 eliminateZeroOTUS(thislookup);
 722
 723                 if (m->control_pressed) { return 0; }
 724
 725                 for (int i = 0; i < thislookup.size(); i++) {
 726                         out << thislookup[i]->getLabel() << '\t' << thislookup[i]->getGroup() << '\t';
 727                         thislookup[i]->print(out);
 728                 }
 729
 730                 return 0;
 731
 732         }
 733         catch(exception& e) {
 734                 m->errorOut(e, "SubSampleCommand", "processShared");
 735                 exit(1);
 736         }
 737 }
 738 //**********************************************************************************************************************
 739 int SubSampleCommand::getSubSampleList() {
 740         try {
 741
 742                 string thisOutputDir = outputDir;
 743                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 744                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "subsample" + m->getExtension(listfile);
 745
 746                 ofstream out;
 747                 m->openOutputFile(outputFileName, out);
 748                 outputTypes["list"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 749
 750                 InputData* input = new InputData(listfile, "list");
 751                 ListVector* list = input->getListVector();
 752                 string lastLabel = list->getLabel();
 753
 754                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 755                 set<string> processedLabels;
 756                 set<string> userLabels = labels;
 757
 758                 ofstream outGroup;
 759                 GroupMap* groupMap;
 760                 if (groupfile != "") {
 761
 762                         groupMap = new GroupMap(groupfile);
 763                         groupMap->readMap();
 764
 765                         //takes care of user setting groupNames that are invalid or setting groups=all
 766                         SharedUtil* util = new SharedUtil();
 767                         util->setGroups(Groups, groupMap->namesOfGroups);
 768                         delete util;
 769
 770                         //create outputfiles
 771                         string groupOutputDir = outputDir;
 772                         if (outputDir == "") {  groupOutputDir += m->hasPath(groupfile);  }
 773                         string groupOutputFileName = groupOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "subsample" + m->getExtension(groupfile);
 774
 775                         m->openOutputFile(groupOutputFileName, outGroup);
 776                         outputTypes["group"].push_back(groupOutputFileName);  outputNames.push_back(groupOutputFileName);
 777
 778                         //file mismatch quit
 779                         if (list->getNumSeqs() != groupMap->getNumSeqs()) {
 780                                 m->mothurOut("[ERROR]: your list file contains " + toString(list->getNumSeqs()) + " sequences, and your groupfile contains " + toString(groupMap->getNumSeqs()) + ", please correct.");
 781                                 m->mothurOutEndLine();
 782                                 delete groupMap;
 783                                 delete list;
 784                                 delete input;
 785                                 out.close();
 786                                 outGroup.close();
 787                                 return 0;
 788                         }
 789                 }
 790
 791                 //make sure that if your picked groups size is not too big
 792                 if (pickedGroups) {
 793                         int total = 0;
 794                         for(int i = 0; i < Groups.size(); i++) {
 795                                 total += groupMap->getNumSeqs(Groups[i]);
 796                         }
 797
 798                         if (size == 0) { //user has not set size, set size = 10% samples size
 799                                 size = int (total * 0.10);
 800                         }
 801
 802                         if (total < size) {
 803                                 m->mothurOut("Your size is too large for the number of groups you selected. Adjusting to " + toString(int (total * 0.10)) + "."); m->mothurOutEndLine();
 804                                 size = int (total * 0.10);
 805                         }
 806
 807                         m->mothurOut("Sampling " + toString(size) + " from " + toString(total) + "."); m->mothurOutEndLine();
 808                 }else{
 809
 810                         if (size == 0) { //user has not set size, set size = 10% samples size
 811                                 size = int (list->getNumSeqs() * 0.10);
 812                         }
 813
 814                         int thisSize = list->getNumSeqs();
 815                         if (size > thisSize) { m->mothurOut("Your list file only contains " + toString(thisSize) + " sequences. Setting size to " + toString(thisSize) + "."); m->mothurOutEndLine();
 816                                 size = thisSize;
 817                         }
 818
 819                         m->mothurOut("Sampling " + toString(size) + " from " + toString(list->getNumSeqs()) + "."); m->mothurOutEndLine();
 820                 }
 821
 822
 823                 for (int i = 0; i < list->getNumBins(); i++) {
 824                         string binnames = list->get(i);
 825
 826                         //parse names
 827                         string individual = "";
 828                         int length = binnames.length();
 829                         for(int j=0;j<length;j++){
 830                                 if(binnames[j] == ','){
 831
 832                                         if (groupfile != "") { //if there is a groupfile given fill in group info
 833                                                 string group = groupMap->getGroup(individual);
 834                                                 if (group == "not found") { m->mothurOut("[ERROR]: " + individual + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
 835
 836                                                 if (pickedGroups) { //if hte user picked groups, we only want to keep the names of sequences from those groups
 837                                                         if (m->inUsersGroups(group, Groups)) {
 838                                                                 names.push_back(individual);
 839                                                         }
 840                                                 }else{
 841                                                         names.push_back(individual);
 842                                                 }
 843                                         }else{ //save everyone, group
 844                                                 names.push_back(individual);
 845                                         }
 846                                         individual = "";
 847                                 }
 848                                 else{
 849                                         individual += binnames[j];
 850                                 }
 851                         }
 852                         //save last name
 853                         if (groupfile != "") { //if there is a groupfile given fill in group info
 854                                 string group = groupMap->getGroup(individual);
 855                                 if (group == "not found") { m->mothurOut("[ERROR]: " + individual + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
 856
 857                                 if (pickedGroups) { //if hte user picked groups, we only want to keep the names of sequences from those groups
 858                                         if (m->inUsersGroups(group, Groups)) {
 859                                                 names.push_back(individual);
 860                                         }
 861                                 }else{
 862                                         names.push_back(individual);
 863                                 }
 864                         }else{ //save everyone, group
 865                                 names.push_back(individual);
 866                         }
 867                 }
 868
 869                 random_shuffle(names.begin(), names.end());
 870
 871                 //randomly select a subset of those names to include in the subsample
 872                 set<string> subset; //dont want repeat sequence names added
 873                 for (int j = 0; j < size; j++) {
 874
 875                         if (m->control_pressed) { break; }
 876
 877                         //get random sequence to add, making sure we have not already added it
 878                         bool done = false;
 879                         int myrand;
 880                         while (!done) {
 881                                 myrand = (int)((float)(rand()) / (RAND_MAX / (names.size()-1) + 1));
 882
 883                                 if (subset.count(names[myrand]) == 0)  { subset.insert(names[myrand]); break;   }
 884                         }
 885                 }
 886
 887                 if (groupfile != "") {
 888                         //write out new groupfile
 889                         for (set<string>::iterator it = subset.begin(); it != subset.end(); it++) {
 890                                 string group = groupMap->getGroup(*it);
 891                                 if (group == "not found") { group = "NOTFOUND"; }
 892
 893                                 outGroup << *it << '\t' << group << endl;
 894                         }
 895                         outGroup.close(); delete groupMap;
 896                 }
 897
 898
 899                 //as long as you are not at the end of the file or done wih the lines you want
 900                 while((list != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 901
 902                         if (m->control_pressed) {  delete list; delete input; out.close();  return 0;  }
 903
 904                         if(allLines == 1 || labels.count(list->getLabel()) == 1){
 905
 906                                 m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 907
 908                                 processList(list, out, subset);
 909
 910                                 processedLabels.insert(list->getLabel());
 911                                 userLabels.erase(list->getLabel());
 912                         }
 913
 914                         if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 915                                 string saveLabel = list->getLabel();
 916
 917                                 delete list;
 918
 919                                 list = input->getListVector(lastLabel);
 920                                 m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 921
 922                                 processList(list, out, subset);
 923
 924                                 processedLabels.insert(list->getLabel());
 925                                 userLabels.erase(list->getLabel());
 926
 927                                 //restore real lastlabel to save below
 928                                 list->setLabel(saveLabel);
 929                         }
 930
 931                         lastLabel = list->getLabel();
 932
 933                         delete list; list = NULL;
 934
 935                         //get next line to process
 936                         list = input->getListVector();
 937                 }
 938
 939
 940                 if (m->control_pressed) {  if (list != NULL) { delete list; } delete input; out.close(); return 0;  }
 941
 942                 //output error messages about any remaining user labels
 943                 set<string>::iterator it;
 944                 bool needToRun = false;
 945                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 946                         m->mothurOut("Your file does not include the label " + *it);
 947                         if (processedLabels.count(lastLabel) != 1) {
 948                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 949                                 needToRun = true;
 950                         }else {
 951                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 952                         }
 953                 }
 954
 955                 //run last label if you need to
 956                 if (needToRun == true)  {
 957                         if (list != NULL) { delete list; }
 958
 959                         list = input->getListVector(lastLabel);
 960
 961                         m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 962
 963                         processList(list, out, subset);
 964
 965                         delete list; list = NULL;
 966                 }
 967
 968                 out.close();
 969                 if (list != NULL) { delete list; }
 970                 delete input;
 971
 972                 return 0;
 973
 974         }
 975         catch(exception& e) {
 976                 m->errorOut(e, "SubSampleCommand", "getSubSampleList");
 977                 exit(1);
 978         }
 979 }
 980 //**********************************************************************************************************************
 981 int SubSampleCommand::processList(ListVector*& list, ofstream& out, set<string>& subset) {
 982         try {
 983
 984                 int numBins = list->getNumBins();
 985
 986                 ListVector* temp = new ListVector();
 987                 temp->setLabel(list->getLabel());
 988
 989                 for (int i = 0; i < numBins; i++) {
 990
 991                         if (m->control_pressed) { break; }
 992
 993                         string binnames = list->get(i);
 994
 995                         //parse names
 996                         string individual = "";
 997                         string newNames = "";
 998                         int length = binnames.length();
 999                         for(int j=0;j<length;j++){
1000                                 if(binnames[j] == ','){
1001                                         if (subset.count(individual) != 0) {  newNames += individual + ",";  }
1002                                         individual = "";
1003                                 }else{
1004                                         individual += binnames[j];
1005                                 }
1006                         }
1007                         if (subset.count(individual) != 0) {  newNames += individual;  }
1008
1009
1010                         //if there are names in this bin add to new list
1011                         if (newNames != "") {
1012                                 newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
1013                                 temp->push_back(newNames);
1014                         }
1015                 }
1016
1017                 delete list;
1018                 list = temp;
1019
1020                 if (m->control_pressed) { return 0; }
1021
1022                 list->print(out);
1023
1024                 return 0;
1025
1026         }
1027         catch(exception& e) {
1028                 m->errorOut(e, "SubSampleCommand", "processList");
1029                 exit(1);
1030         }
1031 }
1032 //**********************************************************************************************************************
1033 int SubSampleCommand::getSubSampleRabund() {
1034         try {
1035
1036                 string thisOutputDir = outputDir;
1037                 if (outputDir == "") {  thisOutputDir += m->hasPath(rabundfile);  }
1038                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(rabundfile)) + "subsample" + m->getExtension(rabundfile);
1039
1040                 ofstream out;
1041                 m->openOutputFile(outputFileName, out);
1042                 outputTypes["rabund"].push_back(outputFileName);  outputNames.push_back(outputFileName);
1043
1044                 InputData* input = new InputData(rabundfile, "rabund");
1045                 RAbundVector* rabund = input->getRAbundVector();
1046                 string lastLabel = rabund->getLabel();
1047
1048                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
1049                 set<string> processedLabels;
1050                 set<string> userLabels = labels;
1051
1052                 if (size == 0) { //user has not set size, set size = 10%
1053                         size = int((rabund->getNumSeqs()) * 0.10);
1054                 }
1055
1056                 m->mothurOut("Sampling " + toString(size) + " from " + toString(rabund->getNumSeqs()) + "."); m->mothurOutEndLine();
1057
1058                 //as long as you are not at the end of the file or done wih the lines you want
1059                 while((rabund != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
1060                         if (m->control_pressed) {  delete input; delete rabund; out.close(); return 0;  }
1061
1062                         if(allLines == 1 || labels.count(rabund->getLabel()) == 1){
1063
1064                                 m->mothurOut(rabund->getLabel()); m->mothurOutEndLine();
1065
1066                                 processRabund(rabund, out);
1067
1068                                 processedLabels.insert(rabund->getLabel());
1069                                 userLabels.erase(rabund->getLabel());
1070                         }
1071
1072                         if ((m->anyLabelsToProcess(rabund->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
1073                                 string saveLabel = rabund->getLabel();
1074
1075                                 delete rabund;
1076
1077                                 rabund = input->getRAbundVector(lastLabel);
1078                                 m->mothurOut(rabund->getLabel()); m->mothurOutEndLine();
1079
1080                                 processRabund(rabund, out);
1081
1082                                 processedLabels.insert(rabund->getLabel());
1083                                 userLabels.erase(rabund->getLabel());
1084
1085                                 //restore real lastlabel to save below
1086                                 rabund->setLabel(saveLabel);
1087                         }
1088
1089                         lastLabel = rabund->getLabel();
1090
1091                         //prevent memory leak
1092                         delete rabund; rabund = NULL;
1093
1094                         //get next line to process
1095                         rabund = input->getRAbundVector();
1096                 }
1097
1098
1099                 if (m->control_pressed) {  out.close(); return 0;  }
1100
1101                 //output error messages about any remaining user labels
1102                 set<string>::iterator it;
1103                 bool needToRun = false;
1104                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
1105                         m->mothurOut("Your file does not include the label " + *it);
1106                         if (processedLabels.count(lastLabel) != 1) {
1107                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
1108                                 needToRun = true;
1109                         }else {
1110                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
1111                         }
1112                 }
1113
1114                 //run last label if you need to
1115                 if (needToRun == true)  {
1116                         if (rabund != NULL) { delete rabund; }
1117
1118                         rabund = input->getRAbundVector(lastLabel);
1119
1120                         m->mothurOut(rabund->getLabel()); m->mothurOutEndLine();
1121
1122                         processRabund(rabund, out);
1123
1124                         delete rabund;
1125                 }
1126
1127                 delete input;
1128                 out.close();
1129
1130                 return 0;
1131
1132         }
1133         catch(exception& e) {
1134                 m->errorOut(e, "SubSampleCommand", "getSubSampleRabund");
1135                 exit(1);
1136         }
1137 }
1138 //**********************************************************************************************************************
1139 int SubSampleCommand::processRabund(RAbundVector*& rabund, ofstream& out) {
1140         try {
1141
1142                 int numBins = rabund->getNumBins();
1143                 int thisSize = rabund->getNumSeqs();
1144
1145                 if (thisSize != size) {
1146
1147                         OrderVector* order = new OrderVector();
1148                         for(int p=0;p<numBins;p++){
1149                                 for(int j=0;j<rabund->get(p);j++){
1150                                         order->push_back(p);
1151                                 }
1152                         }
1153                         random_shuffle(order->begin(), order->end());
1154
1155                         RAbundVector* temp = new RAbundVector(numBins);
1156                         temp->setLabel(rabund->getLabel());
1157
1158                         delete rabund;
1159                         rabund = temp;
1160
1161                         for (int j = 0; j < size; j++) {
1162
1163                                 if (m->control_pressed) { delete order; return 0; }
1164
1165                                 //get random number to sample from order between 0 and thisSize-1.
1166                                 int myrand = (int)((float)(rand()) / (RAND_MAX / (thisSize-1) + 1));
1167
1168                                 int bin = order->get(myrand);
1169
1170                                 int abund = rabund->get(bin);
1171                                 rabund->set(bin, (abund+1));
1172                         }
1173
1174                         delete order;
1175                 }
1176
1177                 if (m->control_pressed) { return 0; }
1178
1179                 rabund->print(out);
1180
1181                 return 0;
1182
1183         }
1184         catch(exception& e) {
1185                 m->errorOut(e, "SubSampleCommand", "processRabund");
1186                 exit(1);
1187         }
1188 }
1189 //**********************************************************************************************************************
1190 int SubSampleCommand::getSubSampleSabund() {
1191         try {
1192
1193                 string thisOutputDir = outputDir;
1194                 if (outputDir == "") {  thisOutputDir += m->hasPath(sabundfile);  }
1195                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(sabundfile)) + "subsample" + m->getExtension(sabundfile);
1196
1197                 ofstream out;
1198                 m->openOutputFile(outputFileName, out);
1199                 outputTypes["sabund"].push_back(outputFileName);  outputNames.push_back(outputFileName);
1200
1201                 InputData* input = new InputData(sabundfile, "sabund");
1202                 SAbundVector* sabund = input->getSAbundVector();
1203                 string lastLabel = sabund->getLabel();
1204
1205                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
1206                 set<string> processedLabels;
1207                 set<string> userLabels = labels;
1208
1209                 if (size == 0) { //user has not set size, set size = 10%
1210                         size = int((sabund->getNumSeqs()) * 0.10);
1211                 }
1212
1213                 m->mothurOut("Sampling " + toString(size) + " from " + toString(sabund->getNumSeqs()) + "."); m->mothurOutEndLine();
1214
1215                 //as long as you are not at the end of the file or done wih the lines you want
1216                 while((sabund != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
1217                         if (m->control_pressed) {  delete input; delete sabund; out.close(); return 0;  }
1218
1219                         if(allLines == 1 || labels.count(sabund->getLabel()) == 1){
1220
1221                                 m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
1222
1223                                 processSabund(sabund, out);
1224
1225                                 processedLabels.insert(sabund->getLabel());
1226                                 userLabels.erase(sabund->getLabel());
1227                         }
1228
1229                         if ((m->anyLabelsToProcess(sabund->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
1230                                 string saveLabel = sabund->getLabel();
1231
1232                                 delete sabund;
1233
1234                                 sabund = input->getSAbundVector(lastLabel);
1235                                 m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
1236
1237                                 processSabund(sabund, out);
1238
1239                                 processedLabels.insert(sabund->getLabel());
1240                                 userLabels.erase(sabund->getLabel());
1241
1242                                 //restore real lastlabel to save below
1243                                 sabund->setLabel(saveLabel);
1244                         }
1245
1246                         lastLabel = sabund->getLabel();
1247
1248                         //prevent memory leak
1249                         delete sabund; sabund = NULL;
1250
1251                         //get next line to process
1252                         sabund = input->getSAbundVector();
1253                 }
1254
1255
1256                 if (m->control_pressed) {  out.close(); return 0;  }
1257
1258                 //output error messages about any remaining user labels
1259                 set<string>::iterator it;
1260                 bool needToRun = false;
1261                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
1262                         m->mothurOut("Your file does not include the label " + *it);
1263                         if (processedLabels.count(lastLabel) != 1) {
1264                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
1265                                 needToRun = true;
1266                         }else {
1267                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
1268                         }
1269                 }
1270
1271                 //run last label if you need to
1272                 if (needToRun == true)  {
1273                         if (sabund != NULL) { delete sabund; }
1274
1275                         sabund = input->getSAbundVector(lastLabel);
1276
1277                         m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
1278
1279                         processSabund(sabund, out);
1280
1281                         delete sabund;
1282                 }
1283
1284                 delete input;
1285                 out.close();
1286
1287                 return 0;
1288
1289         }
1290         catch(exception& e) {
1291                 m->errorOut(e, "SubSampleCommand", "getSubSampleSabund");
1292                 exit(1);
1293         }
1294 }
1295 //**********************************************************************************************************************
1296 int SubSampleCommand::processSabund(SAbundVector*& sabund, ofstream& out) {
1297         try {
1298
1299                 RAbundVector* rabund = new RAbundVector();
1300                 *rabund = sabund->getRAbundVector();
1301
1302                 int numBins = rabund->getNumBins();
1303                 int thisSize = rabund->getNumSeqs();
1304
1305                 if (thisSize != size) {
1306
1307                         OrderVector* order = new OrderVector();
1308                         for(int p=0;p<numBins;p++){
1309                                 for(int j=0;j<rabund->get(p);j++){
1310                                         order->push_back(p);
1311                                 }
1312                         }
1313                         random_shuffle(order->begin(), order->end());
1314
1315                         RAbundVector* temp = new RAbundVector(numBins);
1316                         temp->setLabel(rabund->getLabel());
1317
1318                         delete rabund;
1319                         rabund = temp;
1320
1321                         for (int j = 0; j < size; j++) {
1322
1323                                 if (m->control_pressed) { delete order; return 0; }
1324
1325                                 //get random number to sample from order between 0 and thisSize-1.
1326                                 int myrand = (int)((float)(rand()) / (RAND_MAX / (thisSize-1) + 1));
1327
1328                                 int bin = order->get(myrand);
1329
1330                                 int abund = rabund->get(bin);
1331                                 rabund->set(bin, (abund+1));
1332                         }
1333
1334                         delete order;
1335                 }
1336
1337                 if (m->control_pressed) { return 0; }
1338
1339                 delete sabund;
1340                 sabund = new SAbundVector();
1341                 *sabund = rabund->getSAbundVector();
1342                 delete rabund;
1343
1344                 sabund->print(out);
1345
1346                 return 0;
1347
1348         }
1349         catch(exception& e) {
1350                 m->errorOut(e, "SubSampleCommand", "processSabund");
1351                 exit(1);
1352         }
1353 }
1354 //**********************************************************************************************************************
1355 int SubSampleCommand::eliminateZeroOTUS(vector<SharedRAbundVector*>& thislookup) {
1356         try {
1357
1358                 vector<SharedRAbundVector*> newLookup;
1359                 for (int i = 0; i < thislookup.size(); i++) {
1360                         SharedRAbundVector* temp = new SharedRAbundVector();
1361                         temp->setLabel(thislookup[i]->getLabel());
1362                         temp->setGroup(thislookup[i]->getGroup());
1363                         newLookup.push_back(temp);
1364                 }
1365
1366                 //for each bin
1367                 for (int i = 0; i < thislookup[0]->getNumBins(); i++) {
1368                         if (m->control_pressed) { for (int j = 0; j < newLookup.size(); j++) {  delete newLookup[j];  } return 0; }
1369
1370                         //look at each sharedRabund and make sure they are not all zero
1371                         bool allZero = true;
1372                         for (int j = 0; j < thislookup.size(); j++) {
1373                                 if (thislookup[j]->getAbundance(i) != 0) { allZero = false;  break;  }
1374                         }
1375
1376                         //if they are not all zero add this bin
1377                         if (!allZero) {
1378                                 for (int j = 0; j < thislookup.size(); j++) {
1379                                         newLookup[j]->push_back(thislookup[j]->getAbundance(i), thislookup[j]->getGroup());
1380                                 }
1381                         }
1382                 }
1383
1384                 for (int j = 0; j < thislookup.size(); j++) {  delete thislookup[j];  }
1385                 thislookup.clear();
1386
1387                 thislookup = newLookup;
1388
1389                 return 0;
1390
1391         }
1392         catch(exception& e) {
1393                 m->errorOut(e, "SubSampleCommand", "eliminateZeroOTUS");
1394                 exit(1);
1395         }
1396 }
1397
1398 //**********************************************************************************************************************
1399
1400
1401