subsamplecommand.cpp

   1 /*
   2  *  subsamplecommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 10/27/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "subsamplecommand.h"
  11 #include "sharedutilities.h"
  12
  13 //**********************************************************************************************************************
  14 vector<string> SubSampleCommand::setParameters(){
  15         try {
  16                 CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "FLSSR", "none",false,false); parameters.push_back(pfasta);
  17                 CommandParameter pname("name", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pname);
  18                 CommandParameter pgroup("group", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pgroup);
  19                 CommandParameter plist("list", "InputTypes", "", "", "none", "FLSSR", "none",false,false); parameters.push_back(plist);
  20                 CommandParameter pshared("shared", "InputTypes", "", "", "none", "FLSSR", "none",false,false); parameters.push_back(pshared);
  21                 CommandParameter prabund("rabund", "InputTypes", "", "", "none", "FLSSR", "none",false,false); parameters.push_back(prabund);
  22                 CommandParameter psabund("sabund", "InputTypes", "", "", "none", "FLSSR", "none",false,false); parameters.push_back(psabund);
  23                 CommandParameter plabel("label", "String", "", "", "", "", "",false,false); parameters.push_back(plabel);
  24                 CommandParameter pgroups("groups", "String", "", "", "", "", "",false,false); parameters.push_back(pgroups);
  25                 CommandParameter psize("size", "Number", "", "0", "", "", "",false,false); parameters.push_back(psize);
  26                 CommandParameter ppersample("persample", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(ppersample);
  27                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  28                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  29
  30                 vector<string> myArray;
  31                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  32                 return myArray;
  33         }
  34         catch(exception& e) {
  35                 m->errorOut(e, "SubSampleCommand", "setParameters");
  36                 exit(1);
  37         }
  38 }
  39 //**********************************************************************************************************************
  40 string SubSampleCommand::getHelpString(){
  41         try {
  42                 string helpString = "";
  43                 helpString += "The sub.sample command is designed to be used as a way to normalize your data, or create a smaller set from your original set.\n";
  44                 helpString += "The sub.sample command parameters are fasta, name, list, group, rabund, sabund, shared, groups, size, persample and label.  You must provide a fasta, list, sabund, rabund or shared file as an input file.\n";
  45                 helpString += "The namefile is only used with the fasta file, not with the listfile, because the list file should contain all sequences.\n";
  46                 helpString += "The groups parameter allows you to specify which of the groups in your groupfile you would like included. The group names are separated by dashes.\n";
  47                 helpString += "The label parameter allows you to select what distance levels you would like, and are also separated by dashes.\n";
  48                 helpString += "The size parameter allows you indicate the size of your subsample.\n";
  49                 helpString += "The persample parameter allows you indicate you want to select subsample of the same size from each of your groups, default=false. It is only used with the list and fasta files if a groupfile is given.\n";
  50                 helpString += "persample=false will select a random set of sequences of the size you select, but the number of seqs from each group may differ.\n";
  51                 helpString += "The size parameter is not set: with shared file size=number of seqs in smallest sample, with all other files if a groupfile is given and persample=true, then size=number of seqs in smallest sample, otherwise size=10% of number of seqs.\n";
  52                 helpString += "The sub.sample command should be in the following format: sub.sample(list=yourListFile, group=yourGroupFile, groups=yourGroups, label=yourLabels).\n";
  53                 helpString += "Example sub.sample(list=abrecovery.fn.list, group=abrecovery.groups, groups=B-C, size=20).\n";
  54                 helpString += "The default value for groups is all the groups in your groupfile, and all labels in your inputfile will be used.\n";
  55                 helpString += "The sub.sample command outputs a .subsample file.\n";
  56                 helpString += "Note: No spaces between parameter labels (i.e. groups), '=' and parameters (i.e.yourGroups).\n";
  57                 return helpString;
  58         }
  59         catch(exception& e) {
  60                 m->errorOut(e, "SubSampleCommand", "getHelpString");
  61                 exit(1);
  62         }
  63 }
  64 //**********************************************************************************************************************
  65 SubSampleCommand::SubSampleCommand(){
  66         try {
  67                 abort = true; calledHelp = true;
  68                 setParameters();
  69                 vector<string> tempOutNames;
  70                 outputTypes["shared"] = tempOutNames;
  71                 outputTypes["list"] = tempOutNames;
  72                 outputTypes["rabund"] = tempOutNames;
  73                 outputTypes["sabund"] = tempOutNames;
  74                 outputTypes["fasta"] = tempOutNames;
  75                 outputTypes["name"] = tempOutNames;
  76                 outputTypes["group"] = tempOutNames;
  77         }
  78         catch(exception& e) {
  79                 m->errorOut(e, "SubSampleCommand", "GetRelAbundCommand");
  80                 exit(1);
  81         }
  82 }
  83 //**********************************************************************************************************************
  84 SubSampleCommand::SubSampleCommand(string option) {
  85         try {
  86                 abort = false; calledHelp = false;
  87                 allLines = 1;
  88
  89                 //allow user to run help
  90                 if(option == "help") { help(); abort = true; calledHelp = true; }
  91                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  92
  93                 else {
  94                         vector<string> myArray = setParameters();
  95
  96                         OptionParser parser(option);
  97                         map<string,string> parameters = parser.getParameters();
  98
  99                         ValidParameters validParameter;
 100
 101                         //check to make sure all parameters are valid for command
 102                         map<string,string>::iterator it;
 103                         for (it = parameters.begin(); it != parameters.end(); it++) {
 104                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 105                         }
 106
 107                         //initialize outputTypes
 108                         vector<string> tempOutNames;
 109                         outputTypes["shared"] = tempOutNames;
 110                         outputTypes["list"] = tempOutNames;
 111                         outputTypes["rabund"] = tempOutNames;
 112                         outputTypes["sabund"] = tempOutNames;
 113                         outputTypes["fasta"] = tempOutNames;
 114                         outputTypes["name"] = tempOutNames;
 115                         outputTypes["group"] = tempOutNames;
 116
 117                         //if the user changes the output directory command factory will send this info to us in the output parameter
 118                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
 119
 120                         //if the user changes the input directory command factory will send this info to us in the output parameter
 121                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 122                         if (inputDir == "not found"){   inputDir = "";          }
 123                         else {
 124                                 string path;
 125                                 it = parameters.find("list");
 126                                 //user has given a template file
 127                                 if(it != parameters.end()){
 128                                         path = m->hasPath(it->second);
 129                                         //if the user has not given a path then, add inputdir. else leave path alone.
 130                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 131                                 }
 132
 133                                 it = parameters.find("fasta");
 134                                 //user has given a template file
 135                                 if(it != parameters.end()){
 136                                         path = m->hasPath(it->second);
 137                                         //if the user has not given a path then, add inputdir. else leave path alone.
 138                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 139                                 }
 140
 141                                 it = parameters.find("shared");
 142                                 //user has given a template file
 143                                 if(it != parameters.end()){
 144                                         path = m->hasPath(it->second);
 145                                         //if the user has not given a path then, add inputdir. else leave path alone.
 146                                         if (path == "") {       parameters["shared"] = inputDir + it->second;           }
 147                                 }
 148
 149                                 it = parameters.find("group");
 150                                 //user has given a template file
 151                                 if(it != parameters.end()){
 152                                         path = m->hasPath(it->second);
 153                                         //if the user has not given a path then, add inputdir. else leave path alone.
 154                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 155                                 }
 156
 157                                 it = parameters.find("sabund");
 158                                 //user has given a template file
 159                                 if(it != parameters.end()){
 160                                         path = m->hasPath(it->second);
 161                                         //if the user has not given a path then, add inputdir. else leave path alone.
 162                                         if (path == "") {       parameters["sabund"] = inputDir + it->second;           }
 163                                 }
 164
 165                                 it = parameters.find("rabund");
 166                                 //user has given a template file
 167                                 if(it != parameters.end()){
 168                                         path = m->hasPath(it->second);
 169                                         //if the user has not given a path then, add inputdir. else leave path alone.
 170                                         if (path == "") {       parameters["rabund"] = inputDir + it->second;           }
 171                                 }
 172
 173                                 it = parameters.find("name");
 174                                 //user has given a template file
 175                                 if(it != parameters.end()){
 176                                         path = m->hasPath(it->second);
 177                                         //if the user has not given a path then, add inputdir. else leave path alone.
 178                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 179                                 }
 180                         }
 181
 182                         //check for required parameters
 183                         listfile = validParameter.validFile(parameters, "list", true);
 184                         if (listfile == "not open") { listfile = ""; abort = true; }
 185                         else if (listfile == "not found") { listfile = ""; }
 186                         else { m->setListFile(listfile); }
 187
 188                         sabundfile = validParameter.validFile(parameters, "sabund", true);
 189                         if (sabundfile == "not open") { sabundfile = ""; abort = true; }
 190                         else if (sabundfile == "not found") { sabundfile = ""; }
 191                         else { m->setSabundFile(sabundfile); }
 192
 193                         rabundfile = validParameter.validFile(parameters, "rabund", true);
 194                         if (rabundfile == "not open") { rabundfile = ""; abort = true; }
 195                         else if (rabundfile == "not found") { rabundfile = ""; }
 196                         else { m->setRabundFile(rabundfile); }
 197
 198                         fastafile = validParameter.validFile(parameters, "fasta", true);
 199                         if (fastafile == "not open") { fastafile = ""; abort = true; }
 200                         else if (fastafile == "not found") { fastafile = ""; }
 201                         else { m->setFastaFile(fastafile); }
 202
 203                         sharedfile = validParameter.validFile(parameters, "shared", true);
 204                         if (sharedfile == "not open") { sharedfile = ""; abort = true; }
 205                         else if (sharedfile == "not found") { sharedfile = ""; }
 206                         else { m->setSharedFile(sharedfile); }
 207
 208                         namefile = validParameter.validFile(parameters, "name", true);
 209                         if (namefile == "not open") { namefile = ""; abort = true; }
 210                         else if (namefile == "not found") { namefile = ""; }
 211                         else { m->setNameFile(namefile); }
 212
 213                         groupfile = validParameter.validFile(parameters, "group", true);
 214                         if (groupfile == "not open") { groupfile = ""; abort = true; }
 215                         else if (groupfile == "not found") { groupfile = ""; }
 216                         else { m->setGroupFile(groupfile); }
 217
 218                         //check for optional parameter and set defaults
 219                         // ...at some point should added some additional type checking...
 220                         label = validParameter.validFile(parameters, "label", false);
 221                         if (label == "not found") { label = ""; }
 222                         else {
 223                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 224                                 else { allLines = 1;  }
 225                         }
 226
 227                         groups = validParameter.validFile(parameters, "groups", false);
 228                         if (groups == "not found") { groups = ""; pickedGroups = false; }
 229                         else {
 230                                 pickedGroups = true;
 231                                 m->splitAtDash(groups, Groups);
 232                                 m->setGroups(Groups);
 233                         }
 234
 235                         string temp = validParameter.validFile(parameters, "size", false);              if (temp == "not found"){       temp = "0";             }
 236                         convert(temp, size);
 237
 238                         temp = validParameter.validFile(parameters, "persample", false);                if (temp == "not found"){       temp = "f";             }
 239                         persample = m->isTrue(temp);
 240
 241                         if (groupfile == "") { persample = false; }
 242
 243                         if ((namefile != "") && (fastafile == "")) { m->mothurOut("You may only use a namefile with a fastafile."); m->mothurOutEndLine(); abort = true; }
 244
 245                         if ((fastafile == "") && (listfile == "") && (sabundfile == "") && (rabundfile == "") && (sharedfile == "")) {
 246                                 m->mothurOut("You must provide a fasta, list, sabund, rabund or shared file as an input file."); m->mothurOutEndLine(); abort = true; }
 247
 248                         if (pickedGroups && ((groupfile == "") && (sharedfile == ""))) {
 249                                 m->mothurOut("You cannot pick groups without a valid group file or shared file."); m->mothurOutEndLine(); abort = true; }
 250
 251                         if ((groupfile != "") && ((fastafile == "") && (listfile == ""))) {
 252                                 m->mothurOut("Group file only valid with listfile or fastafile."); m->mothurOutEndLine(); abort = true; }
 253
 254                         if ((groupfile != "") && ((fastafile != "") && (listfile != ""))) {
 255                                 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; }
 256
 257                 }
 258
 259         }
 260         catch(exception& e) {
 261                 m->errorOut(e, "SubSampleCommand", "SubSampleCommand");
 262                 exit(1);
 263         }
 264 }
 265 //**********************************************************************************************************************
 266
 267 int SubSampleCommand::execute(){
 268         try {
 269
 270                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 271
 272                 if (sharedfile != "")   {   getSubSampleShared();       }
 273                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]); return 0; } }
 274
 275                 if (listfile != "")             {   getSubSampleList();         }
 276                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]); return 0; } }
 277
 278                 if (rabundfile != "")   {   getSubSampleRabund();       }
 279                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]); return 0; } }
 280
 281                 if (sabundfile != "")   {   getSubSampleSabund();       }
 282                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]); return 0; } }
 283
 284                 if (fastafile != "")    {   getSubSampleFasta();        }
 285                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]); return 0; } }
 286
 287                 //set fasta file as new current fastafile
 288                 string current = "";
 289                 itTypes = outputTypes.find("fasta");
 290                 if (itTypes != outputTypes.end()) {
 291                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 292                 }
 293
 294                 itTypes = outputTypes.find("name");
 295                 if (itTypes != outputTypes.end()) {
 296                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
 297                 }
 298
 299                 itTypes = outputTypes.find("group");
 300                 if (itTypes != outputTypes.end()) {
 301                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setGroupFile(current); }
 302                 }
 303
 304                 itTypes = outputTypes.find("list");
 305                 if (itTypes != outputTypes.end()) {
 306                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 307                 }
 308
 309                 itTypes = outputTypes.find("shared");
 310                 if (itTypes != outputTypes.end()) {
 311                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setSharedFile(current); }
 312                 }
 313
 314                 itTypes = outputTypes.find("rabund");
 315                 if (itTypes != outputTypes.end()) {
 316                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setRabundFile(current); }
 317                 }
 318
 319                 itTypes = outputTypes.find("sabund");
 320                 if (itTypes != outputTypes.end()) {
 321                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setSabundFile(current); }
 322                 }
 323
 324
 325                 m->mothurOutEndLine();
 326                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 327                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 328                 m->mothurOutEndLine();
 329
 330                 return 0;
 331         }
 332         catch(exception& e) {
 333                 m->errorOut(e, "SubSampleCommand", "execute");
 334                 exit(1);
 335         }
 336 }
 337 //**********************************************************************************************************************
 338 int SubSampleCommand::getSubSampleFasta() {
 339         try {
 340
 341                 if (namefile != "") { readNames(); }    //fills names with all names in namefile.
 342                 else { getNames(); }//no name file, so get list of names to pick from
 343
 344                 GroupMap* groupMap;
 345                 if (groupfile != "") {
 346
 347                         groupMap = new GroupMap(groupfile);
 348                         groupMap->readMap();
 349
 350                         //takes care of user setting groupNames that are invalid or setting groups=all
 351                         SharedUtil* util = new SharedUtil();
 352                         vector<string> namesGroups = groupMap->getNamesOfGroups();
 353                         util->setGroups(Groups, namesGroups);
 354                         delete util;
 355
 356                         //file mismatch quit
 357                         if (names.size() != groupMap->getNumSeqs()) {
 358                                 m->mothurOut("[ERROR]: your fasta file contains " + toString(names.size()) + " sequences, and your groupfile contains " + toString(groupMap->getNumSeqs()) + ", please correct.");
 359                                 m->mothurOutEndLine();
 360                                 delete groupMap;
 361                                 return 0;
 362                         }
 363                 }
 364
 365                 if (m->control_pressed) { return 0; }
 366
 367
 368                 //make sure that if your picked groups size is not too big
 369                 int thisSize = names.size();
 370                 if (persample) {
 371                         if (size == 0) { //user has not set size, set size = smallest samples size
 372                                 size = groupMap->getNumSeqs(Groups[0]);
 373                                 for (int i = 1; i < Groups.size(); i++) {
 374                                         int thisSize = groupMap->getNumSeqs(Groups[i]);
 375
 376                                         if (thisSize < size) {  size = thisSize;        }
 377                                 }
 378                         }else { //make sure size is not too large
 379                                 vector<string> newGroups;
 380                                 for (int i = 0; i < Groups.size(); i++) {
 381                                         int thisSize = groupMap->getNumSeqs(Groups[i]);
 382
 383                                         if (thisSize >= size) { newGroups.push_back(Groups[i]); }
 384                                         else {  m->mothurOut("You have selected a size that is larger than " + Groups[i] + " number of sequences, removing " + Groups[i] + "."); m->mothurOutEndLine(); }
 385                                 }
 386                                 Groups = newGroups;
 387                         }
 388
 389                         m->mothurOut("Sampling " + toString(size) + " from each group."); m->mothurOutEndLine();
 390                 }else {
 391                         if (pickedGroups) {
 392                                 int total = 0;
 393                                 for(int i = 0; i < Groups.size(); i++) {
 394                                         total += groupMap->getNumSeqs(Groups[i]);
 395                                 }
 396
 397                                 if (size == 0) { //user has not set size, set size = 10% samples size
 398                                         size = int (total * 0.10);
 399                                 }
 400
 401                                 if (total < size) {
 402                                         if (size != 0) {
 403                                                 m->mothurOut("Your size is too large for the number of groups you selected. Adjusting to " + toString(int (total * 0.10)) + "."); m->mothurOutEndLine();
 404                                         }
 405                                         size = int (total * 0.10);
 406                                 }
 407
 408                                 m->mothurOut("Sampling " + toString(size) + " from " + toString(total) + "."); m->mothurOutEndLine();
 409                         }
 410
 411                         if (size == 0) { //user has not set size, set size = 10% samples size
 412                                 size = int (names.size() * 0.10);
 413                         }
 414
 415                         if (size > thisSize) { m->mothurOut("Your fasta file only contains " + toString(thisSize) + " sequences. Setting size to " + toString(thisSize) + "."); m->mothurOutEndLine();
 416                                 size = thisSize;
 417                         }
 418
 419                         if (!pickedGroups) { m->mothurOut("Sampling " + toString(size) + " from " + toString(thisSize) + "."); m->mothurOutEndLine(); }
 420
 421                 }
 422                 random_shuffle(names.begin(), names.end());
 423
 424                 set<string> subset; //dont want repeat sequence names added
 425                 if (persample) {
 426                         for (int i = 0; i < Groups.size(); i++) {
 427
 428                                 //randomly select a subset of those names from this group to include in the subsample
 429                                 for (int j = 0; j < size; j++) {
 430
 431                                         if (m->control_pressed) { return 0; }
 432
 433                                         //get random sequence to add, making sure we have not already added it
 434                                         bool done = false;
 435                                         int myrand;
 436                                         while (!done) {
 437                                                 myrand = int((float)(thisSize) * (float)(rand()) / ((float)RAND_MAX+1.0));
 438
 439                                                 if (subset.count(names[myrand]) == 0)  {
 440
 441                                                         string group = groupMap->getGroup(names[myrand]);
 442                                                         if (group == "not found") { m->mothurOut("[ERROR]: " + names[myrand] + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
 443
 444                                                         if (group == Groups[i]) {       subset.insert(names[myrand]); break;    }
 445                                                 }
 446                                         }
 447                                 }
 448                         }
 449                 }else {
 450
 451                         //randomly select a subset of those names to include in the subsample
 452                         for (int j = 0; j < size; j++) {
 453
 454                                 if (m->control_pressed) { return 0; }
 455
 456                                 //get random sequence to add, making sure we have not already added it
 457                                 bool done = false;
 458                                 int myrand;
 459                                 while (!done) {
 460                                         myrand = int((float)(thisSize) * (float)(rand()) / ((float)RAND_MAX+1.0));
 461
 462                                         if (subset.count(names[myrand]) == 0)  {
 463
 464                                                 if (groupfile != "") { //if there is a groupfile given fill in group info
 465                                                         string group = groupMap->getGroup(names[myrand]);
 466                                                         if (group == "not found") { m->mothurOut("[ERROR]: " + names[myrand] + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
 467
 468                                                         if (pickedGroups) { //if hte user picked groups, we only want to keep the names of sequences from those groups
 469                                                                 if (m->inUsersGroups(group, Groups)) {
 470                                                                         subset.insert(names[myrand]); break;
 471                                                                 }
 472                                                         }else{
 473                                                                 subset.insert(names[myrand]); break;
 474                                                         }
 475                                                 }else{ //save everyone, group
 476                                                         subset.insert(names[myrand]); break;
 477                                                 }
 478                                         }
 479                                 }
 480                         }
 481                 }
 482
 483                 if (subset.size() == 0) {  m->mothurOut("The size you selected is too large, skipping fasta file."); m->mothurOutEndLine();  return 0; }
 484
 485                 string thisOutputDir = outputDir;
 486                 if (outputDir == "") {  thisOutputDir += m->hasPath(fastafile);  }
 487                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(fastafile)) + "subsample" + m->getExtension(fastafile);
 488
 489                 ofstream out;
 490                 m->openOutputFile(outputFileName, out);
 491                 outputTypes["fasta"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 492
 493                 //read through fasta file outputting only the names on the subsample list
 494                 ifstream in;
 495                 m->openInputFile(fastafile, in);
 496
 497                 string thisname;
 498                 int count = 0;
 499                 map<string, vector<string> >::iterator itNameMap;
 500
 501                 while(!in.eof()){
 502
 503                         if (m->control_pressed) { in.close(); out.close();  return 0; }
 504
 505                         Sequence currSeq(in);
 506                         thisname = currSeq.getName();
 507
 508                         if (thisname != "") {
 509
 510                                 //does the subset contain a sequence that this sequence represents
 511                                 itNameMap = nameMap.find(thisname);
 512                                 if (itNameMap != nameMap.end()) {
 513                                         vector<string> nameRepresents = itNameMap->second;
 514
 515                                         for (int i = 0; i < nameRepresents.size(); i++){
 516                                                 if (subset.count(nameRepresents[i]) != 0) {
 517                                                         out << ">" << nameRepresents[i] << endl << currSeq.getAligned() << endl;
 518                                                         count++;
 519                                                 }
 520                                         }
 521                                 }else{
 522                                         m->mothurOut("[ERROR]: " + thisname + " is not in your namefile, please correct."); m->mothurOutEndLine();
 523                                 }
 524                         }
 525                         m->gobble(in);
 526                 }
 527                 in.close();
 528                 out.close();
 529
 530                 if (count != subset.size()) {
 531                         m->mothurOut("[ERROR]: The subset selected contained " + toString(subset.size()) + " sequences, but I only found " + toString(count) + " of those in the fastafile."); m->mothurOutEndLine();
 532                 }
 533
 534                 //if a groupfile is provided read through the group file only outputting the names on the subsample list
 535                 if (groupfile != "") {
 536
 537                         string groupOutputDir = outputDir;
 538                         if (outputDir == "") {  groupOutputDir += m->hasPath(groupfile);  }
 539                         string groupOutputFileName = groupOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "subsample" + m->getExtension(groupfile);
 540
 541                         ofstream outGroup;
 542                         m->openOutputFile(groupOutputFileName, outGroup);
 543                         outputTypes["group"].push_back(groupOutputFileName);  outputNames.push_back(groupOutputFileName);
 544
 545                         ifstream inGroup;
 546                         m->openInputFile(groupfile, inGroup);
 547                         string name, group;
 548
 549                         while(!inGroup.eof()){
 550
 551                                 if (m->control_pressed) { inGroup.close(); outGroup.close(); return 0; }
 552
 553                                 inGroup >> name;        m->gobble(inGroup);                     //read from first column
 554                                 inGroup >> group;                       //read from second column
 555
 556                                 //if this name is in the accnos file
 557                                 if (subset.count(name) != 0) {
 558                                         outGroup << name << '\t' << group << endl;
 559                                         subset.erase(name);
 560                                 }
 561
 562                                 m->gobble(inGroup);
 563                         }
 564                         inGroup.close();
 565                         outGroup.close();
 566
 567                         //sanity check
 568                         if (subset.size() != 0) {
 569                                 m->mothurOut("Your groupfile does not match your fasta file."); m->mothurOutEndLine();
 570                                 for (set<string>::iterator it = subset.begin(); it != subset.end(); it++) {
 571                                         m->mothurOut("[ERROR]: " + *it + " is missing from your groupfile."); m->mothurOutEndLine();
 572                                 }
 573                         }
 574                 }
 575
 576                 return 0;
 577
 578         }
 579         catch(exception& e) {
 580                 m->errorOut(e, "SubSampleCommand", "getSubSampleFasta");
 581                 exit(1);
 582         }
 583 }
 584 //**********************************************************************************************************************
 585 int SubSampleCommand::getNames() {
 586         try {
 587
 588                 ifstream in;
 589                 m->openInputFile(fastafile, in);
 590
 591                 string thisname;
 592                 while(!in.eof()){
 593
 594                         if (m->control_pressed) { in.close(); return 0; }
 595
 596                         Sequence currSeq(in);
 597                         thisname = currSeq.getName();
 598
 599                         if (thisname != "") {
 600                                 vector<string> temp; temp.push_back(thisname);
 601                                 nameMap[thisname] = temp;
 602                                 names.push_back(thisname);
 603                         }
 604                         m->gobble(in);
 605                 }
 606                 in.close();
 607
 608                 return 0;
 609
 610         }
 611         catch(exception& e) {
 612                 m->errorOut(e, "SubSampleCommand", "getNames");
 613                 exit(1);
 614         }
 615 }
 616 //**********************************************************************************************************************
 617 int SubSampleCommand::readNames() {
 618         try {
 619
 620                 ifstream in;
 621                 m->openInputFile(namefile, in);
 622
 623                 string thisname, repnames;
 624                 map<string, vector<string> >::iterator it;
 625
 626                 while(!in.eof()){
 627
 628                         if (m->control_pressed) { in.close(); return 0; }
 629
 630                         in >> thisname;         m->gobble(in);          //read from first column
 631                         in >> repnames;                 //read from second column
 632
 633                         it = nameMap.find(thisname);
 634                         if (it == nameMap.end()) {
 635
 636                                 vector<string> splitRepNames;
 637                                 m->splitAtComma(repnames, splitRepNames);
 638
 639                                 nameMap[thisname] = splitRepNames;
 640                                 for (int i = 0; i < splitRepNames.size(); i++) { names.push_back(splitRepNames[i]); }
 641
 642                         }else{  m->mothurOut(thisname + " is already in namesfile. I will use first definition."); m->mothurOutEndLine();  }
 643
 644                         m->gobble(in);
 645                 }
 646                 in.close();
 647
 648                 return 0;
 649
 650         }
 651         catch(exception& e) {
 652                 m->errorOut(e, "SubSampleCommand", "readNames");
 653                 exit(1);
 654         }
 655 }
 656 //**********************************************************************************************************************
 657 int SubSampleCommand::getSubSampleShared() {
 658         try {
 659
 660                 InputData* input = new InputData(sharedfile, "sharedfile");
 661                 vector<SharedRAbundVector*> lookup = input->getSharedRAbundVectors();
 662                 string lastLabel = lookup[0]->getLabel();
 663
 664                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 665                 set<string> processedLabels;
 666                 set<string> userLabels = labels;
 667
 668                 if (size == 0) { //user has not set size, set size = smallest samples size
 669                         size = lookup[0]->getNumSeqs();
 670                         for (int i = 1; i < lookup.size(); i++) {
 671                                 int thisSize = lookup[i]->getNumSeqs();
 672
 673                                 if (thisSize < size) {  size = thisSize;        }
 674                         }
 675                 }else {
 676                         m->clearGroups();
 677                         Groups.clear();
 678                         vector<SharedRAbundVector*> temp;
 679                         for (int i = 0; i < lookup.size(); i++) {
 680                                 if (lookup[i]->getNumSeqs() < size) {
 681                                         m->mothurOut(lookup[i]->getGroup() + " contains " + toString(lookup[i]->getNumSeqs()) + ". Eliminating."); m->mothurOutEndLine();
 682                                         delete lookup[i];
 683                                 }else {
 684                                         Groups.push_back(lookup[i]->getGroup());
 685                                         temp.push_back(lookup[i]);
 686                                 }
 687                         }
 688                         lookup = temp;
 689                         m->setGroups(Groups);
 690                 }
 691
 692                 if (lookup.size() == 0) {  m->mothurOut("The size you selected is too large, skipping shared file."); m->mothurOutEndLine(); delete input; return 0; }
 693
 694                 m->mothurOut("Sampling " + toString(size) + " from each group."); m->mothurOutEndLine();
 695
 696                 //as long as you are not at the end of the file or done wih the lines you want
 697                 while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 698                         if (m->control_pressed) {  delete input; for (int i = 0; i < lookup.size(); i++) {  delete lookup[i]; lookup[i] = NULL; }  return 0;  }
 699
 700                         if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){
 701
 702                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 703
 704                                 processShared(lookup);
 705
 706                                 processedLabels.insert(lookup[0]->getLabel());
 707                                 userLabels.erase(lookup[0]->getLabel());
 708                         }
 709
 710                         if ((m->anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 711                                 string saveLabel = lookup[0]->getLabel();
 712
 713                                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 714
 715                                 lookup = input->getSharedRAbundVectors(lastLabel);
 716                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 717
 718                                 processShared(lookup);
 719
 720                                 processedLabels.insert(lookup[0]->getLabel());
 721                                 userLabels.erase(lookup[0]->getLabel());
 722
 723                                 //restore real lastlabel to save below
 724                                 lookup[0]->setLabel(saveLabel);
 725                         }
 726
 727                         lastLabel = lookup[0]->getLabel();
 728                         //prevent memory leak
 729                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i]; lookup[i] = NULL; }
 730
 731                         //get next line to process
 732                         lookup = input->getSharedRAbundVectors();
 733                 }
 734
 735
 736                 if (m->control_pressed) {   return 0;  }
 737
 738                 //output error messages about any remaining user labels
 739                 set<string>::iterator it;
 740                 bool needToRun = false;
 741                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 742                         m->mothurOut("Your file does not include the label " + *it);
 743                         if (processedLabels.count(lastLabel) != 1) {
 744                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 745                                 needToRun = true;
 746                         }else {
 747                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 748                         }
 749                 }
 750
 751                 //run last label if you need to
 752                 if (needToRun == true)  {
 753                         for (int i = 0; i < lookup.size(); i++) { if (lookup[i] != NULL) { delete lookup[i]; } }
 754                         lookup = input->getSharedRAbundVectors(lastLabel);
 755
 756                         m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 757
 758                         processShared(lookup);
 759
 760                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 761                 }
 762
 763                 delete input;
 764
 765                 return 0;
 766
 767         }
 768         catch(exception& e) {
 769                 m->errorOut(e, "SubSampleCommand", "getSubSampleShared");
 770                 exit(1);
 771         }
 772 }
 773 //**********************************************************************************************************************
 774 int SubSampleCommand::processShared(vector<SharedRAbundVector*>& thislookup) {
 775         try {
 776
 777                 //save mothurOut's binLabels to restore for next label
 778                 vector<string> saveBinLabels = m->currentBinLabels;
 779
 780                 string thisOutputDir = outputDir;
 781                 if (outputDir == "") {  thisOutputDir += m->hasPath(sharedfile);  }
 782                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(sharedfile)) + thislookup[0]->getLabel() + ".subsample" + m->getExtension(sharedfile);
 783
 784
 785                 ofstream out;
 786                 m->openOutputFile(outputFileName, out);
 787                 outputTypes["shared"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 788
 789                 int numBins = thislookup[0]->getNumBins();
 790                 for (int i = 0; i < thislookup.size(); i++) {
 791                         int thisSize = thislookup[i]->getNumSeqs();
 792
 793                         if (thisSize != size) {
 794
 795                                 string thisgroup = thislookup[i]->getGroup();
 796
 797                                 OrderVector* order = new OrderVector();
 798                                 for(int p=0;p<numBins;p++){
 799                                         for(int j=0;j<thislookup[i]->getAbundance(p);j++){
 800                                                 order->push_back(p);
 801                                         }
 802                                 }
 803                                 random_shuffle(order->begin(), order->end());
 804
 805                                 SharedRAbundVector* temp = new SharedRAbundVector(numBins);
 806                                 temp->setLabel(thislookup[i]->getLabel());
 807                                 temp->setGroup(thislookup[i]->getGroup());
 808
 809                                 delete thislookup[i];
 810                                 thislookup[i] = temp;
 811
 812
 813                                 for (int j = 0; j < size; j++) {
 814
 815                                         if (m->control_pressed) { delete order; out.close(); return 0; }
 816
 817                                         //get random number to sample from order between 0 and thisSize-1.
 818                                         int myrand = int((float)(thisSize) * (float)(rand()) / ((float)RAND_MAX+1.0));
 819
 820                                         int bin = order->get(myrand);
 821
 822                                         int abund = thislookup[i]->getAbundance(bin);
 823                                         thislookup[i]->set(bin, (abund+1), thisgroup);
 824                                 }
 825                                 delete order;
 826                         }
 827                 }
 828
 829                 //subsampling may have created some otus with no sequences in them
 830                 eliminateZeroOTUS(thislookup);
 831
 832                 if (m->control_pressed) { out.close(); return 0; }
 833
 834                 thislookup[0]->printHeaders(out);
 835
 836                 for (int i = 0; i < thislookup.size(); i++) {
 837                         out << thislookup[i]->getLabel() << '\t' << thislookup[i]->getGroup() << '\t';
 838                         thislookup[i]->print(out);
 839                 }
 840
 841                 out.close();
 842
 843                 //save mothurOut's binLabels to restore for next label
 844                 m->currentBinLabels = saveBinLabels;
 845
 846                 return 0;
 847
 848         }
 849         catch(exception& e) {
 850                 m->errorOut(e, "SubSampleCommand", "processShared");
 851                 exit(1);
 852         }
 853 }
 854 //**********************************************************************************************************************
 855 int SubSampleCommand::getSubSampleList() {
 856         try {
 857
 858                 string thisOutputDir = outputDir;
 859                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 860                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "subsample" + m->getExtension(listfile);
 861
 862                 ofstream out;
 863                 m->openOutputFile(outputFileName, out);
 864                 outputTypes["list"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 865
 866                 InputData* input = new InputData(listfile, "list");
 867                 ListVector* list = input->getListVector();
 868                 string lastLabel = list->getLabel();
 869
 870                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 871                 set<string> processedLabels;
 872                 set<string> userLabels = labels;
 873
 874                 ofstream outGroup;
 875                 GroupMap* groupMap;
 876                 if (groupfile != "") {
 877
 878                         groupMap = new GroupMap(groupfile);
 879                         groupMap->readMap();
 880
 881                         //takes care of user setting groupNames that are invalid or setting groups=all
 882                         SharedUtil* util = new SharedUtil();
 883                         vector<string> namesGroups = groupMap->getNamesOfGroups();
 884                         util->setGroups(Groups, namesGroups);
 885                         delete util;
 886
 887                         //create outputfiles
 888                         string groupOutputDir = outputDir;
 889                         if (outputDir == "") {  groupOutputDir += m->hasPath(groupfile);  }
 890                         string groupOutputFileName = groupOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "subsample" + m->getExtension(groupfile);
 891
 892                         m->openOutputFile(groupOutputFileName, outGroup);
 893                         outputTypes["group"].push_back(groupOutputFileName);  outputNames.push_back(groupOutputFileName);
 894
 895                         //file mismatch quit
 896                         if (list->getNumSeqs() != groupMap->getNumSeqs()) {
 897                                 m->mothurOut("[ERROR]: your list file contains " + toString(list->getNumSeqs()) + " sequences, and your groupfile contains " + toString(groupMap->getNumSeqs()) + ", please correct.");
 898                                 m->mothurOutEndLine();
 899                                 delete groupMap;
 900                                 delete list;
 901                                 delete input;
 902                                 out.close();
 903                                 outGroup.close();
 904                                 return 0;
 905                         }
 906                 }
 907
 908                 //make sure that if your picked groups size is not too big
 909                 if (persample) {
 910                         if (size == 0) { //user has not set size, set size = smallest samples size
 911                                 size = groupMap->getNumSeqs(Groups[0]);
 912                                 for (int i = 1; i < Groups.size(); i++) {
 913                                         int thisSize = groupMap->getNumSeqs(Groups[i]);
 914
 915                                         if (thisSize < size) {  size = thisSize;        }
 916                                 }
 917                         }else { //make sure size is not too large
 918                                 vector<string> newGroups;
 919                                 for (int i = 0; i < Groups.size(); i++) {
 920                                         int thisSize = groupMap->getNumSeqs(Groups[i]);
 921
 922                                         if (thisSize >= size) { newGroups.push_back(Groups[i]); }
 923                                         else {  m->mothurOut("You have selected a size that is larger than " + Groups[i] + " number of sequences, removing " + Groups[i] + "."); m->mothurOutEndLine(); }
 924                                 }
 925                                 Groups = newGroups;
 926                         }
 927
 928                         m->mothurOut("Sampling " + toString(size) + " from each group."); m->mothurOutEndLine();
 929                 }else{
 930                         if (pickedGroups) {
 931                                 int total = 0;
 932                                 for(int i = 0; i < Groups.size(); i++) {
 933                                         total += groupMap->getNumSeqs(Groups[i]);
 934                                 }
 935
 936                                 if (size == 0) { //user has not set size, set size = 10% samples size
 937                                         size = int (total * 0.10);
 938                                 }
 939
 940                                 if (total < size) {
 941                                         m->mothurOut("Your size is too large for the number of groups you selected. Adjusting to " + toString(int (total * 0.10)) + "."); m->mothurOutEndLine();
 942                                         size = int (total * 0.10);
 943                                 }
 944
 945                                 m->mothurOut("Sampling " + toString(size) + " from " + toString(total) + "."); m->mothurOutEndLine();
 946                         }else{
 947
 948                                 if (size == 0) { //user has not set size, set size = 10% samples size
 949                                         size = int (list->getNumSeqs() * 0.10);
 950                                 }
 951
 952                                 int thisSize = list->getNumSeqs();
 953                                 if (size > thisSize) { m->mothurOut("Your list file only contains " + toString(thisSize) + " sequences. Setting size to " + toString(thisSize) + "."); m->mothurOutEndLine();
 954                                         size = thisSize;
 955                                 }
 956
 957                                 m->mothurOut("Sampling " + toString(size) + " from " + toString(list->getNumSeqs()) + "."); m->mothurOutEndLine();
 958                         }
 959                 }
 960
 961
 962                 //fill names
 963                 for (int i = 0; i < list->getNumBins(); i++) {
 964                         string binnames = list->get(i);
 965
 966                         //parse names
 967                         string individual = "";
 968                         int length = binnames.length();
 969                         for(int j=0;j<length;j++){
 970                                 if(binnames[j] == ','){
 971
 972                                         if (groupfile != "") { //if there is a groupfile given fill in group info
 973                                                 string group = groupMap->getGroup(individual);
 974                                                 if (group == "not found") { m->mothurOut("[ERROR]: " + individual + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
 975
 976                                                 if (pickedGroups) { //if hte user picked groups, we only want to keep the names of sequences from those groups
 977                                                         if (m->inUsersGroups(group, Groups)) {
 978                                                                 names.push_back(individual);
 979                                                         }
 980                                                 }else{
 981                                                         names.push_back(individual);
 982                                                 }
 983                                         }else{ //save everyone, group
 984                                                 names.push_back(individual);
 985                                         }
 986                                         individual = "";
 987                                 }
 988                                 else{
 989                                         individual += binnames[j];
 990                                 }
 991                         }
 992                         //save last name
 993                         if (groupfile != "") { //if there is a groupfile given fill in group info
 994                                 string group = groupMap->getGroup(individual);
 995                                 if (group == "not found") { m->mothurOut("[ERROR]: " + individual + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
 996
 997                                 if (pickedGroups) { //if hte user picked groups, we only want to keep the names of sequences from those groups
 998                                         if (m->inUsersGroups(group, Groups)) {
 999                                                 names.push_back(individual);
1000                                         }
1001                                 }else{
1002                                         names.push_back(individual);
1003                                 }
1004                         }else{ //save everyone, group
1005                                 names.push_back(individual);
1006                         }
1007                 }
1008
1009                 random_shuffle(names.begin(), names.end());
1010
1011                 //randomly select a subset of those names to include in the subsample
1012                 set<string> subset; //dont want repeat sequence names added
1013                 if (persample) {
1014                         for (int i = 0; i < Groups.size(); i++) {
1015
1016                                 for (int j = 0; j < size; j++) {
1017
1018                                         if (m->control_pressed) { break; }
1019
1020                                         //get random sequence to add, making sure we have not already added it
1021                                         bool done = false;
1022                                         int myrand;
1023                                         while (!done) {
1024                                                 myrand = int((float)(names.size()) * (float)(rand()) / ((float)RAND_MAX+1.0));
1025
1026                                                 if (subset.count(names[myrand]) == 0) { //you are not already added
1027                                                         if (groupMap->getGroup(names[myrand]) == Groups[i])  { subset.insert(names[myrand]); break;     }
1028                                                 }
1029                                         }
1030                                 }
1031                         }
1032                 }else{
1033                         for (int j = 0; j < size; j++) {
1034
1035                                 if (m->control_pressed) { break; }
1036
1037                                 //get random sequence to add, making sure we have not already added it
1038                                 bool done = false;
1039                                 int myrand;
1040                                 while (!done) {
1041                                         myrand = int((float)(names.size()) * (float)(rand()) / ((float)RAND_MAX+1.0));
1042
1043                                         if (subset.count(names[myrand]) == 0)  { subset.insert(names[myrand]); break;   }
1044                                 }
1045                         }
1046                 }
1047
1048                 if (groupfile != "") {
1049                         //write out new groupfile
1050                         for (set<string>::iterator it = subset.begin(); it != subset.end(); it++) {
1051                                 string group = groupMap->getGroup(*it);
1052                                 if (group == "not found") { group = "NOTFOUND"; }
1053
1054                                 outGroup << *it << '\t' << group << endl;
1055                         }
1056                         outGroup.close(); delete groupMap;
1057                 }
1058
1059
1060                 //as long as you are not at the end of the file or done wih the lines you want
1061                 while((list != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
1062
1063                         if (m->control_pressed) {  delete list; delete input; out.close();  return 0;  }
1064
1065                         if(allLines == 1 || labels.count(list->getLabel()) == 1){
1066
1067                                 m->mothurOut(list->getLabel()); m->mothurOutEndLine();
1068
1069                                 processList(list, out, subset);
1070
1071                                 processedLabels.insert(list->getLabel());
1072                                 userLabels.erase(list->getLabel());
1073                         }
1074
1075                         if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
1076                                 string saveLabel = list->getLabel();
1077
1078                                 delete list;
1079
1080                                 list = input->getListVector(lastLabel);
1081                                 m->mothurOut(list->getLabel()); m->mothurOutEndLine();
1082
1083                                 processList(list, out, subset);
1084
1085                                 processedLabels.insert(list->getLabel());
1086                                 userLabels.erase(list->getLabel());
1087
1088                                 //restore real lastlabel to save below
1089                                 list->setLabel(saveLabel);
1090                         }
1091
1092                         lastLabel = list->getLabel();
1093
1094                         delete list; list = NULL;
1095
1096                         //get next line to process
1097                         list = input->getListVector();
1098                 }
1099
1100
1101                 if (m->control_pressed) {  if (list != NULL) { delete list; } delete input; out.close(); return 0;  }
1102
1103                 //output error messages about any remaining user labels
1104                 set<string>::iterator it;
1105                 bool needToRun = false;
1106                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
1107                         m->mothurOut("Your file does not include the label " + *it);
1108                         if (processedLabels.count(lastLabel) != 1) {
1109                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
1110                                 needToRun = true;
1111                         }else {
1112                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
1113                         }
1114                 }
1115
1116                 //run last label if you need to
1117                 if (needToRun == true)  {
1118                         if (list != NULL) { delete list; }
1119
1120                         list = input->getListVector(lastLabel);
1121
1122                         m->mothurOut(list->getLabel()); m->mothurOutEndLine();
1123
1124                         processList(list, out, subset);
1125
1126                         delete list; list = NULL;
1127                 }
1128
1129                 out.close();
1130                 if (list != NULL) { delete list; }
1131                 delete input;
1132
1133                 return 0;
1134
1135         }
1136         catch(exception& e) {
1137                 m->errorOut(e, "SubSampleCommand", "getSubSampleList");
1138                 exit(1);
1139         }
1140 }
1141 //**********************************************************************************************************************
1142 int SubSampleCommand::processList(ListVector*& list, ofstream& out, set<string>& subset) {
1143         try {
1144
1145                 int numBins = list->getNumBins();
1146
1147                 ListVector* temp = new ListVector();
1148                 temp->setLabel(list->getLabel());
1149
1150                 for (int i = 0; i < numBins; i++) {
1151
1152                         if (m->control_pressed) { break; }
1153
1154                         string binnames = list->get(i);
1155
1156                         //parse names
1157                         string individual = "";
1158                         string newNames = "";
1159                         int length = binnames.length();
1160                         for(int j=0;j<length;j++){
1161                                 if(binnames[j] == ','){
1162                                         if (subset.count(individual) != 0) {  newNames += individual + ",";  }
1163                                         individual = "";
1164                                 }else{
1165                                         individual += binnames[j];
1166                                 }
1167                         }
1168                         if (subset.count(individual) != 0) {  newNames += individual + ",";  }
1169
1170
1171                         //if there are names in this bin add to new list
1172                         if (newNames != "") {
1173                                 newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
1174                                 temp->push_back(newNames);
1175                         }
1176                 }
1177
1178                 delete list;
1179                 list = temp;
1180
1181                 if (m->control_pressed) { return 0; }
1182
1183                 list->print(out);
1184
1185                 return 0;
1186
1187         }
1188         catch(exception& e) {
1189                 m->errorOut(e, "SubSampleCommand", "processList");
1190                 exit(1);
1191         }
1192 }
1193 //**********************************************************************************************************************
1194 int SubSampleCommand::getSubSampleRabund() {
1195         try {
1196                 InputData* input = new InputData(rabundfile, "rabund");
1197                 RAbundVector* rabund = input->getRAbundVector();
1198                 string lastLabel = rabund->getLabel();
1199
1200                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
1201                 set<string> processedLabels;
1202                 set<string> userLabels = labels;
1203
1204                 if (size == 0) { //user has not set size, set size = 10%
1205                         size = int((rabund->getNumSeqs()) * 0.10);
1206                 }else if (size > rabund->getNumSeqs()) { m->mothurOut("The size you selected is too large, skipping rabund file."); m->mothurOutEndLine(); delete input; delete rabund; return 0; }
1207
1208                 m->mothurOut("Sampling " + toString(size) + " from " + toString(rabund->getNumSeqs()) + "."); m->mothurOutEndLine();
1209
1210                 string thisOutputDir = outputDir;
1211                 if (outputDir == "") {  thisOutputDir += m->hasPath(rabundfile);  }
1212                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(rabundfile)) + "subsample" + m->getExtension(rabundfile);
1213
1214                 ofstream out;
1215                 m->openOutputFile(outputFileName, out);
1216                 outputTypes["rabund"].push_back(outputFileName);  outputNames.push_back(outputFileName);
1217
1218                 //as long as you are not at the end of the file or done wih the lines you want
1219                 while((rabund != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
1220                         if (m->control_pressed) {  delete input; delete rabund; out.close(); return 0;  }
1221
1222                         if(allLines == 1 || labels.count(rabund->getLabel()) == 1){
1223
1224                                 m->mothurOut(rabund->getLabel()); m->mothurOutEndLine();
1225
1226                                 processRabund(rabund, out);
1227
1228                                 processedLabels.insert(rabund->getLabel());
1229                                 userLabels.erase(rabund->getLabel());
1230                         }
1231
1232                         if ((m->anyLabelsToProcess(rabund->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
1233                                 string saveLabel = rabund->getLabel();
1234
1235                                 delete rabund;
1236
1237                                 rabund = input->getRAbundVector(lastLabel);
1238                                 m->mothurOut(rabund->getLabel()); m->mothurOutEndLine();
1239
1240                                 processRabund(rabund, out);
1241
1242                                 processedLabels.insert(rabund->getLabel());
1243                                 userLabels.erase(rabund->getLabel());
1244
1245                                 //restore real lastlabel to save below
1246                                 rabund->setLabel(saveLabel);
1247                         }
1248
1249                         lastLabel = rabund->getLabel();
1250
1251                         //prevent memory leak
1252                         delete rabund; rabund = NULL;
1253
1254                         //get next line to process
1255                         rabund = input->getRAbundVector();
1256                 }
1257
1258
1259                 if (m->control_pressed) {  out.close(); return 0;  }
1260
1261                 //output error messages about any remaining user labels
1262                 set<string>::iterator it;
1263                 bool needToRun = false;
1264                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
1265                         m->mothurOut("Your file does not include the label " + *it);
1266                         if (processedLabels.count(lastLabel) != 1) {
1267                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
1268                                 needToRun = true;
1269                         }else {
1270                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
1271                         }
1272                 }
1273
1274                 //run last label if you need to
1275                 if (needToRun == true)  {
1276                         if (rabund != NULL) { delete rabund; }
1277
1278                         rabund = input->getRAbundVector(lastLabel);
1279
1280                         m->mothurOut(rabund->getLabel()); m->mothurOutEndLine();
1281
1282                         processRabund(rabund, out);
1283
1284                         delete rabund;
1285                 }
1286
1287                 delete input;
1288                 out.close();
1289
1290                 return 0;
1291
1292         }
1293         catch(exception& e) {
1294                 m->errorOut(e, "SubSampleCommand", "getSubSampleRabund");
1295                 exit(1);
1296         }
1297 }
1298 //**********************************************************************************************************************
1299 int SubSampleCommand::processRabund(RAbundVector*& rabund, ofstream& out) {
1300         try {
1301
1302                 int numBins = rabund->getNumBins();
1303                 int thisSize = rabund->getNumSeqs();
1304
1305                 if (thisSize != size) {
1306
1307                         OrderVector* order = new OrderVector();
1308                         for(int p=0;p<numBins;p++){
1309                                 for(int j=0;j<rabund->get(p);j++){
1310                                         order->push_back(p);
1311                                 }
1312                         }
1313                         random_shuffle(order->begin(), order->end());
1314
1315                         RAbundVector* temp = new RAbundVector(numBins);
1316                         temp->setLabel(rabund->getLabel());
1317
1318                         delete rabund;
1319                         rabund = temp;
1320
1321                         for (int j = 0; j < size; j++) {
1322
1323                                 if (m->control_pressed) { delete order; return 0; }
1324
1325                                 //get random number to sample from order between 0 and thisSize-1.
1326                                 int myrand = int((float)(thisSize) * (float)(rand()) / ((float)RAND_MAX+1.0));
1327
1328                                 int bin = order->get(myrand);
1329
1330                                 int abund = rabund->get(bin);
1331                                 rabund->set(bin, (abund+1));
1332                         }
1333
1334                         delete order;
1335                 }
1336
1337                 if (m->control_pressed) { return 0; }
1338
1339                 rabund->print(out);
1340
1341                 return 0;
1342
1343         }
1344         catch(exception& e) {
1345                 m->errorOut(e, "SubSampleCommand", "processRabund");
1346                 exit(1);
1347         }
1348 }
1349 //**********************************************************************************************************************
1350 int SubSampleCommand::getSubSampleSabund() {
1351         try {
1352
1353                 InputData* input = new InputData(sabundfile, "sabund");
1354                 SAbundVector* sabund = input->getSAbundVector();
1355                 string lastLabel = sabund->getLabel();
1356
1357                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
1358                 set<string> processedLabels;
1359                 set<string> userLabels = labels;
1360
1361                 if (size == 0) { //user has not set size, set size = 10%
1362                         size = int((sabund->getNumSeqs()) * 0.10);
1363                 }else if (size > sabund->getNumSeqs()) { m->mothurOut("The size you selected is too large, skipping sabund file."); m->mothurOutEndLine(); delete input; delete sabund; return 0; }
1364
1365
1366                 m->mothurOut("Sampling " + toString(size) + " from " + toString(sabund->getNumSeqs()) + "."); m->mothurOutEndLine();
1367
1368                 string thisOutputDir = outputDir;
1369                 if (outputDir == "") {  thisOutputDir += m->hasPath(sabundfile);  }
1370                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(sabundfile)) + "subsample" + m->getExtension(sabundfile);
1371
1372                 ofstream out;
1373                 m->openOutputFile(outputFileName, out);
1374                 outputTypes["sabund"].push_back(outputFileName);  outputNames.push_back(outputFileName);
1375
1376
1377                 //as long as you are not at the end of the file or done wih the lines you want
1378                 while((sabund != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
1379                         if (m->control_pressed) {  delete input; delete sabund; out.close(); return 0;  }
1380
1381                         if(allLines == 1 || labels.count(sabund->getLabel()) == 1){
1382
1383                                 m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
1384
1385                                 processSabund(sabund, out);
1386
1387                                 processedLabels.insert(sabund->getLabel());
1388                                 userLabels.erase(sabund->getLabel());
1389                         }
1390
1391                         if ((m->anyLabelsToProcess(sabund->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
1392                                 string saveLabel = sabund->getLabel();
1393
1394                                 delete sabund;
1395
1396                                 sabund = input->getSAbundVector(lastLabel);
1397                                 m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
1398
1399                                 processSabund(sabund, out);
1400
1401                                 processedLabels.insert(sabund->getLabel());
1402                                 userLabels.erase(sabund->getLabel());
1403
1404                                 //restore real lastlabel to save below
1405                                 sabund->setLabel(saveLabel);
1406                         }
1407
1408                         lastLabel = sabund->getLabel();
1409
1410                         //prevent memory leak
1411                         delete sabund; sabund = NULL;
1412
1413                         //get next line to process
1414                         sabund = input->getSAbundVector();
1415                 }
1416
1417
1418                 if (m->control_pressed) {  out.close(); return 0;  }
1419
1420                 //output error messages about any remaining user labels
1421                 set<string>::iterator it;
1422                 bool needToRun = false;
1423                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
1424                         m->mothurOut("Your file does not include the label " + *it);
1425                         if (processedLabels.count(lastLabel) != 1) {
1426                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
1427                                 needToRun = true;
1428                         }else {
1429                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
1430                         }
1431                 }
1432
1433                 //run last label if you need to
1434                 if (needToRun == true)  {
1435                         if (sabund != NULL) { delete sabund; }
1436
1437                         sabund = input->getSAbundVector(lastLabel);
1438
1439                         m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
1440
1441                         processSabund(sabund, out);
1442
1443                         delete sabund;
1444                 }
1445
1446                 delete input;
1447                 out.close();
1448
1449                 return 0;
1450
1451         }
1452         catch(exception& e) {
1453                 m->errorOut(e, "SubSampleCommand", "getSubSampleSabund");
1454                 exit(1);
1455         }
1456 }
1457 //**********************************************************************************************************************
1458 int SubSampleCommand::processSabund(SAbundVector*& sabund, ofstream& out) {
1459         try {
1460
1461                 RAbundVector* rabund = new RAbundVector();
1462                 *rabund = sabund->getRAbundVector();
1463
1464                 int numBins = rabund->getNumBins();
1465                 int thisSize = rabund->getNumSeqs();
1466
1467                 if (thisSize != size) {
1468
1469                         OrderVector* order = new OrderVector();
1470                         for(int p=0;p<numBins;p++){
1471                                 for(int j=0;j<rabund->get(p);j++){
1472                                         order->push_back(p);
1473                                 }
1474                         }
1475                         random_shuffle(order->begin(), order->end());
1476
1477                         RAbundVector* temp = new RAbundVector(numBins);
1478                         temp->setLabel(rabund->getLabel());
1479
1480                         delete rabund;
1481                         rabund = temp;
1482
1483                         for (int j = 0; j < size; j++) {
1484
1485                                 if (m->control_pressed) { delete order; return 0; }
1486
1487                                 //get random number to sample from order between 0 and thisSize-1.
1488                                 int myrand = int((float)(thisSize) * (float)(rand()) / ((float)RAND_MAX+1.0));
1489
1490                                 int bin = order->get(myrand);
1491
1492                                 int abund = rabund->get(bin);
1493                                 rabund->set(bin, (abund+1));
1494                         }
1495
1496                         delete order;
1497                 }
1498
1499                 if (m->control_pressed) { return 0; }
1500
1501                 delete sabund;
1502                 sabund = new SAbundVector();
1503                 *sabund = rabund->getSAbundVector();
1504                 delete rabund;
1505
1506                 sabund->print(out);
1507
1508                 return 0;
1509
1510         }
1511         catch(exception& e) {
1512                 m->errorOut(e, "SubSampleCommand", "processSabund");
1513                 exit(1);
1514         }
1515 }
1516 //**********************************************************************************************************************
1517 int SubSampleCommand::eliminateZeroOTUS(vector<SharedRAbundVector*>& thislookup) {
1518         try {
1519
1520                 vector<SharedRAbundVector*> newLookup;
1521                 for (int i = 0; i < thislookup.size(); i++) {
1522                         SharedRAbundVector* temp = new SharedRAbundVector();
1523                         temp->setLabel(thislookup[i]->getLabel());
1524                         temp->setGroup(thislookup[i]->getGroup());
1525                         newLookup.push_back(temp);
1526                 }
1527
1528                 //for each bin
1529                 vector<string> newBinLabels;
1530                 for (int i = 0; i < thislookup[0]->getNumBins(); i++) {
1531                         if (m->control_pressed) { for (int j = 0; j < newLookup.size(); j++) {  delete newLookup[j];  } return 0; }
1532
1533                         //look at each sharedRabund and make sure they are not all zero
1534                         bool allZero = true;
1535                         for (int j = 0; j < thislookup.size(); j++) {
1536                                 if (thislookup[j]->getAbundance(i) != 0) { allZero = false;  break;  }
1537                         }
1538
1539                         //if they are not all zero add this bin
1540                         if (!allZero) {
1541                                 for (int j = 0; j < thislookup.size(); j++) {
1542                                         newLookup[j]->push_back(thislookup[j]->getAbundance(i), thislookup[j]->getGroup());
1543                                 }
1544                                 //if there is a bin label use it otherwise make one
1545                                 string binLabel = "Otu" + toString(i+1);
1546                                 if (i < m->currentBinLabels.size()) {  binLabel = m->currentBinLabels[i]; }
1547
1548                                 newBinLabels.push_back(binLabel);
1549                         }
1550                 }
1551
1552                 for (int j = 0; j < thislookup.size(); j++) {  delete thislookup[j];  }
1553                 thislookup.clear();
1554
1555                 thislookup = newLookup;
1556                 m->currentBinLabels = newBinLabels;
1557
1558                 return 0;
1559
1560         }
1561         catch(exception& e) {
1562                 m->errorOut(e, "SubSampleCommand", "eliminateZeroOTUS");
1563                 exit(1);
1564         }
1565 }
1566
1567 //**********************************************************************************************************************
1568
1569
1570