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