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