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->Groups = 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++) {       remove(outputNames[i].c_str()); return 0; } }
 274
 275                 if (listfile != "")             {   getSubSampleList();         }
 276                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       remove(outputNames[i].c_str()); return 0; } }
 277
 278                 if (rabundfile != "")   {   getSubSampleRabund();       }
 279                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       remove(outputNames[i].c_str()); return 0; } }
 280
 281                 if (sabundfile != "")   {   getSubSampleSabund();       }
 282                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       remove(outputNames[i].c_str()); return 0; } }
 283
 284                 if (fastafile != "")    {   getSubSampleFasta();        }
 285                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       remove(outputNames[i].c_str()); 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                         util->setGroups(Groups, groupMap->namesOfGroups);
 353                         delete util;
 354
 355                         //file mismatch quit
 356                         if (names.size() != groupMap->getNumSeqs()) {
 357                                 m->mothurOut("[ERROR]: your fasta file contains " + toString(names.size()) + " sequences, and your groupfile contains " + toString(groupMap->getNumSeqs()) + ", please correct.");
 358                                 m->mothurOutEndLine();
 359                                 delete groupMap;
 360                                 return 0;
 361                         }
 362                 }
 363
 364                 if (m->control_pressed) { return 0; }
 365
 366
 367                 //make sure that if your picked groups size is not too big
 368                 int thisSize = names.size();
 369                 if (persample) {
 370                         if (size == 0) { //user has not set size, set size = smallest samples size
 371                                 size = groupMap->getNumSeqs(Groups[0]);
 372                                 for (int i = 1; i < Groups.size(); i++) {
 373                                         int thisSize = groupMap->getNumSeqs(Groups[i]);
 374
 375                                         if (thisSize < size) {  size = thisSize;        }
 376                                 }
 377                         }else { //make sure size is not too large
 378                                 vector<string> newGroups;
 379                                 for (int i = 0; i < Groups.size(); i++) {
 380                                         int thisSize = groupMap->getNumSeqs(Groups[i]);
 381
 382                                         if (thisSize >= size) { newGroups.push_back(Groups[i]); }
 383                                         else {  m->mothurOut("You have selected a size that is larger than " + Groups[i] + " number of sequences, removing " + Groups[i] + "."); m->mothurOutEndLine(); }
 384                                 }
 385                                 Groups = newGroups;
 386                         }
 387
 388                         m->mothurOut("Sampling " + toString(size) + " from each group."); m->mothurOutEndLine();
 389                 }else {
 390                         if (pickedGroups) {
 391                                 int total = 0;
 392                                 for(int i = 0; i < Groups.size(); i++) {
 393                                         total += groupMap->getNumSeqs(Groups[i]);
 394                                 }
 395
 396                                 if (size == 0) { //user has not set size, set size = 10% samples size
 397                                         size = int (total * 0.10);
 398                                 }
 399
 400                                 if (total < size) {
 401                                         if (size != 0) {
 402                                                 m->mothurOut("Your size is too large for the number of groups you selected. Adjusting to " + toString(int (total * 0.10)) + "."); m->mothurOutEndLine();
 403                                         }
 404                                         size = int (total * 0.10);
 405                                 }
 406
 407                                 m->mothurOut("Sampling " + toString(size) + " from " + toString(total) + "."); m->mothurOutEndLine();
 408                         }
 409
 410                         if (size == 0) { //user has not set size, set size = 10% samples size
 411                                 size = int (names.size() * 0.10);
 412                         }
 413
 414                         if (size > thisSize) { m->mothurOut("Your fasta file only contains " + toString(thisSize) + " sequences. Setting size to " + toString(thisSize) + "."); m->mothurOutEndLine();
 415                                 size = thisSize;
 416                         }
 417
 418                         if (!pickedGroups) { m->mothurOut("Sampling " + toString(size) + " from " + toString(thisSize) + "."); m->mothurOutEndLine(); }
 419
 420                 }
 421                 random_shuffle(names.begin(), names.end());
 422
 423                 set<string> subset; //dont want repeat sequence names added
 424                 if (persample) {
 425                         for (int i = 0; i < Groups.size(); i++) {
 426
 427                                 //randomly select a subset of those names from this group to include in the subsample
 428                                 for (int j = 0; j < size; j++) {
 429
 430                                         if (m->control_pressed) { return 0; }
 431
 432                                         //get random sequence to add, making sure we have not already added it
 433                                         bool done = false;
 434                                         int myrand;
 435                                         while (!done) {
 436                                                 myrand = int((float)(thisSize) * (float)(rand()) / ((float)RAND_MAX+1.0));
 437
 438                                                 if (subset.count(names[myrand]) == 0)  {
 439
 440                                                         string group = groupMap->getGroup(names[myrand]);
 441                                                         if (group == "not found") { m->mothurOut("[ERROR]: " + names[myrand] + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
 442
 443                                                         if (group == Groups[i]) {       subset.insert(names[myrand]); break;    }
 444                                                 }
 445                                         }
 446                                 }
 447                         }
 448                 }else {
 449
 450                         //randomly select a subset of those names to include in the subsample
 451                         for (int j = 0; j < size; j++) {
 452
 453                                 if (m->control_pressed) { return 0; }
 454
 455                                 //get random sequence to add, making sure we have not already added it
 456                                 bool done = false;
 457                                 int myrand;
 458                                 while (!done) {
 459                                         myrand = int((float)(thisSize) * (float)(rand()) / ((float)RAND_MAX+1.0));
 460
 461                                         if (subset.count(names[myrand]) == 0)  {
 462
 463                                                 if (groupfile != "") { //if there is a groupfile given fill in group info
 464                                                         string group = groupMap->getGroup(names[myrand]);
 465                                                         if (group == "not found") { m->mothurOut("[ERROR]: " + names[myrand] + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
 466
 467                                                         if (pickedGroups) { //if hte user picked groups, we only want to keep the names of sequences from those groups
 468                                                                 if (m->inUsersGroups(group, Groups)) {
 469                                                                         subset.insert(names[myrand]); break;
 470                                                                 }
 471                                                         }else{
 472                                                                 subset.insert(names[myrand]); break;
 473                                                         }
 474                                                 }else{ //save everyone, group
 475                                                         subset.insert(names[myrand]); break;
 476                                                 }
 477                                         }
 478                                 }
 479                         }
 480                 }
 481
 482                 if (subset.size() == 0) {  m->mothurOut("The size you selected is too large, skipping fasta file."); m->mothurOutEndLine();  return 0; }
 483
 484                 string thisOutputDir = outputDir;
 485                 if (outputDir == "") {  thisOutputDir += m->hasPath(fastafile);  }
 486                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(fastafile)) + "subsample" + m->getExtension(fastafile);
 487
 488                 ofstream out;
 489                 m->openOutputFile(outputFileName, out);
 490                 outputTypes["fasta"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 491
 492                 //read through fasta file outputting only the names on the subsample list
 493                 ifstream in;
 494                 m->openInputFile(fastafile, in);
 495
 496                 string thisname;
 497                 int count = 0;
 498                 map<string, vector<string> >::iterator itNameMap;
 499
 500                 while(!in.eof()){
 501
 502                         if (m->control_pressed) { in.close(); out.close();  return 0; }
 503
 504                         Sequence currSeq(in);
 505                         thisname = currSeq.getName();
 506
 507                         if (thisname != "") {
 508
 509                                 //does the subset contain a sequence that this sequence represents
 510                                 itNameMap = nameMap.find(thisname);
 511                                 if (itNameMap != nameMap.end()) {
 512                                         vector<string> nameRepresents = itNameMap->second;
 513
 514                                         for (int i = 0; i < nameRepresents.size(); i++){
 515                                                 if (subset.count(nameRepresents[i]) != 0) {
 516                                                         out << ">" << nameRepresents[i] << endl << currSeq.getAligned() << endl;
 517                                                         count++;
 518                                                 }
 519                                         }
 520                                 }else{
 521                                         m->mothurOut("[ERROR]: " + thisname + " is not in your namefile, please correct."); m->mothurOutEndLine();
 522                                 }
 523                         }
 524                         m->gobble(in);
 525                 }
 526                 in.close();
 527                 out.close();
 528
 529                 if (count != subset.size()) {
 530                         m->mothurOut("[ERROR]: The subset selected contained " + toString(subset.size()) + " sequences, but I only found " + toString(count) + " of those in the fastafile."); m->mothurOutEndLine();
 531                 }
 532
 533                 //if a groupfile is provided read through the group file only outputting the names on the subsample list
 534                 if (groupfile != "") {
 535
 536                         string groupOutputDir = outputDir;
 537                         if (outputDir == "") {  groupOutputDir += m->hasPath(groupfile);  }
 538                         string groupOutputFileName = groupOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "subsample" + m->getExtension(groupfile);
 539
 540                         ofstream outGroup;
 541                         m->openOutputFile(groupOutputFileName, outGroup);
 542                         outputTypes["group"].push_back(groupOutputFileName);  outputNames.push_back(groupOutputFileName);
 543
 544                         ifstream inGroup;
 545                         m->openInputFile(groupfile, inGroup);
 546                         string name, group;
 547
 548                         while(!inGroup.eof()){
 549
 550                                 if (m->control_pressed) { inGroup.close(); outGroup.close(); return 0; }
 551
 552                                 inGroup >> name;        m->gobble(inGroup);                     //read from first column
 553                                 inGroup >> group;                       //read from second column
 554
 555                                 //if this name is in the accnos file
 556                                 if (subset.count(name) != 0) {
 557                                         outGroup << name << '\t' << group << endl;
 558                                         subset.erase(name);
 559                                 }
 560
 561                                 m->gobble(inGroup);
 562                         }
 563                         inGroup.close();
 564                         outGroup.close();
 565
 566                         //sanity check
 567                         if (subset.size() != 0) {
 568                                 m->mothurOut("Your groupfile does not match your fasta file."); m->mothurOutEndLine();
 569                                 for (set<string>::iterator it = subset.begin(); it != subset.end(); it++) {
 570                                         m->mothurOut("[ERROR]: " + *it + " is missing from your groupfile."); m->mothurOutEndLine();
 571                                 }
 572                         }
 573                 }
 574
 575                 return 0;
 576
 577         }
 578         catch(exception& e) {
 579                 m->errorOut(e, "SubSampleCommand", "getSubSampleFasta");
 580                 exit(1);
 581         }
 582 }
 583 //**********************************************************************************************************************
 584 int SubSampleCommand::getNames() {
 585         try {
 586
 587                 ifstream in;
 588                 m->openInputFile(fastafile, in);
 589
 590                 string thisname;
 591                 while(!in.eof()){
 592
 593                         if (m->control_pressed) { in.close(); return 0; }
 594
 595                         Sequence currSeq(in);
 596                         thisname = currSeq.getName();
 597
 598                         if (thisname != "") {
 599                                 vector<string> temp; temp.push_back(thisname);
 600                                 nameMap[thisname] = temp;
 601                                 names.push_back(thisname);
 602                         }
 603                         m->gobble(in);
 604                 }
 605                 in.close();
 606
 607                 return 0;
 608
 609         }
 610         catch(exception& e) {
 611                 m->errorOut(e, "SubSampleCommand", "getNames");
 612                 exit(1);
 613         }
 614 }
 615 //**********************************************************************************************************************
 616 int SubSampleCommand::readNames() {
 617         try {
 618
 619                 ifstream in;
 620                 m->openInputFile(namefile, in);
 621
 622                 string thisname, repnames;
 623                 map<string, vector<string> >::iterator it;
 624
 625                 while(!in.eof()){
 626
 627                         if (m->control_pressed) { in.close(); return 0; }
 628
 629                         in >> thisname;         m->gobble(in);          //read from first column
 630                         in >> repnames;                 //read from second column
 631
 632                         it = nameMap.find(thisname);
 633                         if (it == nameMap.end()) {
 634
 635                                 vector<string> splitRepNames;
 636                                 m->splitAtComma(repnames, splitRepNames);
 637
 638                                 nameMap[thisname] = splitRepNames;
 639                                 for (int i = 0; i < splitRepNames.size(); i++) { names.push_back(splitRepNames[i]); }
 640
 641                         }else{  m->mothurOut(thisname + " is already in namesfile. I will use first definition."); m->mothurOutEndLine();  }
 642
 643                         m->gobble(in);
 644                 }
 645                 in.close();
 646
 647                 return 0;
 648
 649         }
 650         catch(exception& e) {
 651                 m->errorOut(e, "SubSampleCommand", "readNames");
 652                 exit(1);
 653         }
 654 }
 655 //**********************************************************************************************************************
 656 int SubSampleCommand::getSubSampleShared() {
 657         try {
 658
 659                 InputData* input = new InputData(sharedfile, "sharedfile");
 660                 vector<SharedRAbundVector*> lookup = input->getSharedRAbundVectors();
 661                 string lastLabel = lookup[0]->getLabel();
 662
 663                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 664                 set<string> processedLabels;
 665                 set<string> userLabels = labels;
 666
 667                 if (size == 0) { //user has not set size, set size = smallest samples size
 668                         size = lookup[0]->getNumSeqs();
 669                         for (int i = 1; i < lookup.size(); i++) {
 670                                 int thisSize = lookup[i]->getNumSeqs();
 671
 672                                 if (thisSize < size) {  size = thisSize;        }
 673                         }
 674                 }else {
 675                         m->Groups.clear();
 676                         vector<SharedRAbundVector*> temp;
 677                         for (int i = 0; i < lookup.size(); i++) {
 678                                 if (lookup[i]->getNumSeqs() < size) {
 679                                         m->mothurOut(lookup[i]->getGroup() + " contains " + toString(lookup[i]->getNumSeqs()) + ". Eliminating."); m->mothurOutEndLine();
 680                                         delete lookup[i];
 681                                 }else {
 682                                         m->Groups.push_back(lookup[i]->getGroup());
 683                                         temp.push_back(lookup[i]);
 684                                 }
 685                         }
 686                         lookup = temp;
 687                         Groups = m->Groups;
 688                 }
 689
 690                 if (lookup.size() == 0) {  m->mothurOut("The size you selected is too large, skipping shared file."); m->mothurOutEndLine(); delete input; return 0; }
 691
 692                 string thisOutputDir = outputDir;
 693                 if (outputDir == "") {  thisOutputDir += m->hasPath(sharedfile);  }
 694                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(sharedfile)) + "subsample" + m->getExtension(sharedfile);
 695
 696                 ofstream out;
 697                 m->openOutputFile(outputFileName, out);
 698                 outputTypes["shared"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 699
 700
 701                 m->mothurOut("Sampling " + toString(size) + " from each group."); m->mothurOutEndLine();
 702
 703                 //as long as you are not at the end of the file or done wih the lines you want
 704                 while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 705                         if (m->control_pressed) {  delete input; for (int i = 0; i < lookup.size(); i++) {  delete lookup[i]; lookup[i] = NULL; } out.close(); return 0;  }
 706
 707                         if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){
 708
 709                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 710
 711                                 if (!m->printedHeaders) { lookup[0]->printHeaders(out); }
 712                                 processShared(lookup, out);
 713
 714                                 processedLabels.insert(lookup[0]->getLabel());
 715                                 userLabels.erase(lookup[0]->getLabel());
 716                         }
 717
 718                         if ((m->anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 719                                 string saveLabel = lookup[0]->getLabel();
 720
 721                                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 722
 723                                 lookup = input->getSharedRAbundVectors(lastLabel);
 724                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 725
 726                                 if (!m->printedHeaders) { lookup[0]->printHeaders(out); }
 727                                 processShared(lookup, out);
 728
 729                                 processedLabels.insert(lookup[0]->getLabel());
 730                                 userLabels.erase(lookup[0]->getLabel());
 731
 732                                 //restore real lastlabel to save below
 733                                 lookup[0]->setLabel(saveLabel);
 734                         }
 735
 736                         lastLabel = lookup[0]->getLabel();
 737                         //prevent memory leak
 738                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i]; lookup[i] = NULL; }
 739
 740                         //get next line to process
 741                         lookup = input->getSharedRAbundVectors();
 742                 }
 743
 744
 745                 if (m->control_pressed) {  out.close(); return 0;  }
 746
 747                 //output error messages about any remaining user labels
 748                 set<string>::iterator it;
 749                 bool needToRun = false;
 750                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 751                         m->mothurOut("Your file does not include the label " + *it);
 752                         if (processedLabels.count(lastLabel) != 1) {
 753                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 754                                 needToRun = true;
 755                         }else {
 756                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 757                         }
 758                 }
 759
 760                 //run last label if you need to
 761                 if (needToRun == true)  {
 762                         for (int i = 0; i < lookup.size(); i++) { if (lookup[i] != NULL) { delete lookup[i]; } }
 763                         lookup = input->getSharedRAbundVectors(lastLabel);
 764
 765                         m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 766
 767                         if (!m->printedHeaders) { lookup[0]->printHeaders(out); }
 768                         processShared(lookup, out);
 769
 770                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 771                 }
 772
 773                 delete input;
 774                 out.close();
 775
 776                 return 0;
 777
 778         }
 779         catch(exception& e) {
 780                 m->errorOut(e, "SubSampleCommand", "getSubSampleShared");
 781                 exit(1);
 782         }
 783 }
 784 //**********************************************************************************************************************
 785 int SubSampleCommand::processShared(vector<SharedRAbundVector*>& thislookup, ofstream& out) {
 786         try {
 787
 788                 int numBins = thislookup[0]->getNumBins();
 789                 for (int i = 0; i < thislookup.size(); i++) {
 790                         int thisSize = thislookup[i]->getNumSeqs();
 791
 792                         if (thisSize != size) {
 793
 794                                 string thisgroup = thislookup[i]->getGroup();
 795
 796                                 OrderVector* order = new OrderVector();
 797                                 for(int p=0;p<numBins;p++){
 798                                         for(int j=0;j<thislookup[i]->getAbundance(p);j++){
 799                                                 order->push_back(p);
 800                                         }
 801                                 }
 802                                 random_shuffle(order->begin(), order->end());
 803
 804                                 SharedRAbundVector* temp = new SharedRAbundVector(numBins);
 805                                 temp->setLabel(thislookup[i]->getLabel());
 806                                 temp->setGroup(thislookup[i]->getGroup());
 807
 808                                 delete thislookup[i];
 809                                 thislookup[i] = temp;
 810
 811
 812                                 for (int j = 0; j < size; j++) {
 813
 814                                         if (m->control_pressed) { delete order; return 0; }
 815
 816                                         //get random number to sample from order between 0 and thisSize-1.
 817                                         int myrand = int((float)(thisSize) * (float)(rand()) / ((float)RAND_MAX+1.0));
 818
 819                                         int bin = order->get(myrand);
 820
 821                                         int abund = thislookup[i]->getAbundance(bin);
 822                                         thislookup[i]->set(bin, (abund+1), thisgroup);
 823                                 }
 824                                 delete order;
 825                         }
 826                 }
 827
 828                 //subsampling may have created some otus with no sequences in them
 829                 eliminateZeroOTUS(thislookup);
 830
 831                 if (m->control_pressed) { return 0; }
 832
 833                 for (int i = 0; i < thislookup.size(); i++) {
 834                         out << thislookup[i]->getLabel() << '\t' << thislookup[i]->getGroup() << '\t';
 835                         thislookup[i]->print(out);
 836                 }
 837
 838                 return 0;
 839
 840         }
 841         catch(exception& e) {
 842                 m->errorOut(e, "SubSampleCommand", "processShared");
 843                 exit(1);
 844         }
 845 }
 846 //**********************************************************************************************************************
 847 int SubSampleCommand::getSubSampleList() {
 848         try {
 849
 850                 string thisOutputDir = outputDir;
 851                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 852                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "subsample" + m->getExtension(listfile);
 853
 854                 ofstream out;
 855                 m->openOutputFile(outputFileName, out);
 856                 outputTypes["list"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 857
 858                 InputData* input = new InputData(listfile, "list");
 859                 ListVector* list = input->getListVector();
 860                 string lastLabel = list->getLabel();
 861
 862                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 863                 set<string> processedLabels;
 864                 set<string> userLabels = labels;
 865
 866                 ofstream outGroup;
 867                 GroupMap* groupMap;
 868                 if (groupfile != "") {
 869
 870                         groupMap = new GroupMap(groupfile);
 871                         groupMap->readMap();
 872
 873                         //takes care of user setting groupNames that are invalid or setting groups=all
 874                         SharedUtil* util = new SharedUtil();
 875                         util->setGroups(Groups, groupMap->namesOfGroups);
 876                         delete util;
 877
 878                         //create outputfiles
 879                         string groupOutputDir = outputDir;
 880                         if (outputDir == "") {  groupOutputDir += m->hasPath(groupfile);  }
 881                         string groupOutputFileName = groupOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "subsample" + m->getExtension(groupfile);
 882
 883                         m->openOutputFile(groupOutputFileName, outGroup);
 884                         outputTypes["group"].push_back(groupOutputFileName);  outputNames.push_back(groupOutputFileName);
 885
 886                         //file mismatch quit
 887                         if (list->getNumSeqs() != groupMap->getNumSeqs()) {
 888                                 m->mothurOut("[ERROR]: your list file contains " + toString(list->getNumSeqs()) + " sequences, and your groupfile contains " + toString(groupMap->getNumSeqs()) + ", please correct.");
 889                                 m->mothurOutEndLine();
 890                                 delete groupMap;
 891                                 delete list;
 892                                 delete input;
 893                                 out.close();
 894                                 outGroup.close();
 895                                 return 0;
 896                         }
 897                 }
 898
 899                 //make sure that if your picked groups size is not too big
 900                 if (persample) {
 901                         if (size == 0) { //user has not set size, set size = smallest samples size
 902                                 size = groupMap->getNumSeqs(Groups[0]);
 903                                 for (int i = 1; i < Groups.size(); i++) {
 904                                         int thisSize = groupMap->getNumSeqs(Groups[i]);
 905
 906                                         if (thisSize < size) {  size = thisSize;        }
 907                                 }
 908                         }else { //make sure size is not too large
 909                                 vector<string> newGroups;
 910                                 for (int i = 0; i < Groups.size(); i++) {
 911                                         int thisSize = groupMap->getNumSeqs(Groups[i]);
 912
 913                                         if (thisSize >= size) { newGroups.push_back(Groups[i]); }
 914                                         else {  m->mothurOut("You have selected a size that is larger than " + Groups[i] + " number of sequences, removing " + Groups[i] + "."); m->mothurOutEndLine(); }
 915                                 }
 916                                 Groups = newGroups;
 917                         }
 918
 919                         m->mothurOut("Sampling " + toString(size) + " from each group."); m->mothurOutEndLine();
 920                 }else{
 921                         if (pickedGroups) {
 922                                 int total = 0;
 923                                 for(int i = 0; i < Groups.size(); i++) {
 924                                         total += groupMap->getNumSeqs(Groups[i]);
 925                                 }
 926
 927                                 if (size == 0) { //user has not set size, set size = 10% samples size
 928                                         size = int (total * 0.10);
 929                                 }
 930
 931                                 if (total < size) {
 932                                         m->mothurOut("Your size is too large for the number of groups you selected. Adjusting to " + toString(int (total * 0.10)) + "."); m->mothurOutEndLine();
 933                                         size = int (total * 0.10);
 934                                 }
 935
 936                                 m->mothurOut("Sampling " + toString(size) + " from " + toString(total) + "."); m->mothurOutEndLine();
 937                         }else{
 938
 939                                 if (size == 0) { //user has not set size, set size = 10% samples size
 940                                         size = int (list->getNumSeqs() * 0.10);
 941                                 }
 942
 943                                 int thisSize = list->getNumSeqs();
 944                                 if (size > thisSize) { m->mothurOut("Your list file only contains " + toString(thisSize) + " sequences. Setting size to " + toString(thisSize) + "."); m->mothurOutEndLine();
 945                                         size = thisSize;
 946                                 }
 947
 948                                 m->mothurOut("Sampling " + toString(size) + " from " + toString(list->getNumSeqs()) + "."); m->mothurOutEndLine();
 949                         }
 950                 }
 951
 952
 953                 //fill names
 954                 for (int i = 0; i < list->getNumBins(); i++) {
 955                         string binnames = list->get(i);
 956
 957                         //parse names
 958                         string individual = "";
 959                         int length = binnames.length();
 960                         for(int j=0;j<length;j++){
 961                                 if(binnames[j] == ','){
 962
 963                                         if (groupfile != "") { //if there is a groupfile given fill in group info
 964                                                 string group = groupMap->getGroup(individual);
 965                                                 if (group == "not found") { m->mothurOut("[ERROR]: " + individual + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
 966
 967                                                 if (pickedGroups) { //if hte user picked groups, we only want to keep the names of sequences from those groups
 968                                                         if (m->inUsersGroups(group, Groups)) {
 969                                                                 names.push_back(individual);
 970                                                         }
 971                                                 }else{
 972                                                         names.push_back(individual);
 973                                                 }
 974                                         }else{ //save everyone, group
 975                                                 names.push_back(individual);
 976                                         }
 977                                         individual = "";
 978                                 }
 979                                 else{
 980                                         individual += binnames[j];
 981                                 }
 982                         }
 983                         //save last name
 984                         if (groupfile != "") { //if there is a groupfile given fill in group info
 985                                 string group = groupMap->getGroup(individual);
 986                                 if (group == "not found") { m->mothurOut("[ERROR]: " + individual + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
 987
 988                                 if (pickedGroups) { //if hte user picked groups, we only want to keep the names of sequences from those groups
 989                                         if (m->inUsersGroups(group, Groups)) {
 990                                                 names.push_back(individual);
 991                                         }
 992                                 }else{
 993                                         names.push_back(individual);
 994                                 }
 995                         }else{ //save everyone, group
 996                                 names.push_back(individual);
 997                         }
 998                 }
 999
1000                 random_shuffle(names.begin(), names.end());
1001
1002                 //randomly select a subset of those names to include in the subsample
1003                 set<string> subset; //dont want repeat sequence names added
1004                 if (persample) {
1005                         for (int i = 0; i < Groups.size(); i++) {
1006
1007                                 for (int j = 0; j < size; j++) {
1008
1009                                         if (m->control_pressed) { break; }
1010
1011                                         //get random sequence to add, making sure we have not already added it
1012                                         bool done = false;
1013                                         int myrand;
1014                                         while (!done) {
1015                                                 myrand = int((float)(names.size()) * (float)(rand()) / ((float)RAND_MAX+1.0));
1016
1017                                                 if (subset.count(names[myrand]) == 0) { //you are not already added
1018                                                         if (groupMap->getGroup(names[myrand]) == Groups[i])  { subset.insert(names[myrand]); break;     }
1019                                                 }
1020                                         }
1021                                 }
1022                         }
1023                 }else{
1024                         for (int j = 0; j < size; j++) {
1025
1026                                 if (m->control_pressed) { break; }
1027
1028                                 //get random sequence to add, making sure we have not already added it
1029                                 bool done = false;
1030                                 int myrand;
1031                                 while (!done) {
1032                                         myrand = int((float)(names.size()) * (float)(rand()) / ((float)RAND_MAX+1.0));
1033
1034                                         if (subset.count(names[myrand]) == 0)  { subset.insert(names[myrand]); break;   }
1035                                 }
1036                         }
1037                 }
1038
1039                 if (groupfile != "") {
1040                         //write out new groupfile
1041                         for (set<string>::iterator it = subset.begin(); it != subset.end(); it++) {
1042                                 string group = groupMap->getGroup(*it);
1043                                 if (group == "not found") { group = "NOTFOUND"; }
1044
1045                                 outGroup << *it << '\t' << group << endl;
1046                         }
1047                         outGroup.close(); delete groupMap;
1048                 }
1049
1050
1051                 //as long as you are not at the end of the file or done wih the lines you want
1052                 while((list != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
1053
1054                         if (m->control_pressed) {  delete list; delete input; out.close();  return 0;  }
1055
1056                         if(allLines == 1 || labels.count(list->getLabel()) == 1){
1057
1058                                 m->mothurOut(list->getLabel()); m->mothurOutEndLine();
1059
1060                                 processList(list, out, subset);
1061
1062                                 processedLabels.insert(list->getLabel());
1063                                 userLabels.erase(list->getLabel());
1064                         }
1065
1066                         if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
1067                                 string saveLabel = list->getLabel();
1068
1069                                 delete list;
1070
1071                                 list = input->getListVector(lastLabel);
1072                                 m->mothurOut(list->getLabel()); m->mothurOutEndLine();
1073
1074                                 processList(list, out, subset);
1075
1076                                 processedLabels.insert(list->getLabel());
1077                                 userLabels.erase(list->getLabel());
1078
1079                                 //restore real lastlabel to save below
1080                                 list->setLabel(saveLabel);
1081                         }
1082
1083                         lastLabel = list->getLabel();
1084
1085                         delete list; list = NULL;
1086
1087                         //get next line to process
1088                         list = input->getListVector();
1089                 }
1090
1091
1092                 if (m->control_pressed) {  if (list != NULL) { delete list; } delete input; out.close(); return 0;  }
1093
1094                 //output error messages about any remaining user labels
1095                 set<string>::iterator it;
1096                 bool needToRun = false;
1097                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
1098                         m->mothurOut("Your file does not include the label " + *it);
1099                         if (processedLabels.count(lastLabel) != 1) {
1100                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
1101                                 needToRun = true;
1102                         }else {
1103                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
1104                         }
1105                 }
1106
1107                 //run last label if you need to
1108                 if (needToRun == true)  {
1109                         if (list != NULL) { delete list; }
1110
1111                         list = input->getListVector(lastLabel);
1112
1113                         m->mothurOut(list->getLabel()); m->mothurOutEndLine();
1114
1115                         processList(list, out, subset);
1116
1117                         delete list; list = NULL;
1118                 }
1119
1120                 out.close();
1121                 if (list != NULL) { delete list; }
1122                 delete input;
1123
1124                 return 0;
1125
1126         }
1127         catch(exception& e) {
1128                 m->errorOut(e, "SubSampleCommand", "getSubSampleList");
1129                 exit(1);
1130         }
1131 }
1132 //**********************************************************************************************************************
1133 int SubSampleCommand::processList(ListVector*& list, ofstream& out, set<string>& subset) {
1134         try {
1135
1136                 int numBins = list->getNumBins();
1137
1138                 ListVector* temp = new ListVector();
1139                 temp->setLabel(list->getLabel());
1140
1141                 for (int i = 0; i < numBins; i++) {
1142
1143                         if (m->control_pressed) { break; }
1144
1145                         string binnames = list->get(i);
1146
1147                         //parse names
1148                         string individual = "";
1149                         string newNames = "";
1150                         int length = binnames.length();
1151                         for(int j=0;j<length;j++){
1152                                 if(binnames[j] == ','){
1153                                         if (subset.count(individual) != 0) {  newNames += individual + ",";  }
1154                                         individual = "";
1155                                 }else{
1156                                         individual += binnames[j];
1157                                 }
1158                         }
1159                         if (subset.count(individual) != 0) {  newNames += individual + ",";  }
1160
1161
1162                         //if there are names in this bin add to new list
1163                         if (newNames != "") {
1164                                 newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
1165                                 temp->push_back(newNames);
1166                         }
1167                 }
1168
1169                 delete list;
1170                 list = temp;
1171
1172                 if (m->control_pressed) { return 0; }
1173
1174                 list->print(out);
1175
1176                 return 0;
1177
1178         }
1179         catch(exception& e) {
1180                 m->errorOut(e, "SubSampleCommand", "processList");
1181                 exit(1);
1182         }
1183 }
1184 //**********************************************************************************************************************
1185 int SubSampleCommand::getSubSampleRabund() {
1186         try {
1187                 InputData* input = new InputData(rabundfile, "rabund");
1188                 RAbundVector* rabund = input->getRAbundVector();
1189                 string lastLabel = rabund->getLabel();
1190
1191                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
1192                 set<string> processedLabels;
1193                 set<string> userLabels = labels;
1194
1195                 if (size == 0) { //user has not set size, set size = 10%
1196                         size = int((rabund->getNumSeqs()) * 0.10);
1197                 }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; }
1198
1199                 m->mothurOut("Sampling " + toString(size) + " from " + toString(rabund->getNumSeqs()) + "."); m->mothurOutEndLine();
1200
1201                 string thisOutputDir = outputDir;
1202                 if (outputDir == "") {  thisOutputDir += m->hasPath(rabundfile);  }
1203                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(rabundfile)) + "subsample" + m->getExtension(rabundfile);
1204
1205                 ofstream out;
1206                 m->openOutputFile(outputFileName, out);
1207                 outputTypes["rabund"].push_back(outputFileName);  outputNames.push_back(outputFileName);
1208
1209                 //as long as you are not at the end of the file or done wih the lines you want
1210                 while((rabund != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
1211                         if (m->control_pressed) {  delete input; delete rabund; out.close(); return 0;  }
1212
1213                         if(allLines == 1 || labels.count(rabund->getLabel()) == 1){
1214
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
1223                         if ((m->anyLabelsToProcess(rabund->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
1224                                 string saveLabel = rabund->getLabel();
1225
1226                                 delete rabund;
1227
1228                                 rabund = input->getRAbundVector(lastLabel);
1229                                 m->mothurOut(rabund->getLabel()); m->mothurOutEndLine();
1230
1231                                 processRabund(rabund, out);
1232
1233                                 processedLabels.insert(rabund->getLabel());
1234                                 userLabels.erase(rabund->getLabel());
1235
1236                                 //restore real lastlabel to save below
1237                                 rabund->setLabel(saveLabel);
1238                         }
1239
1240                         lastLabel = rabund->getLabel();
1241
1242                         //prevent memory leak
1243                         delete rabund; rabund = NULL;
1244
1245                         //get next line to process
1246                         rabund = input->getRAbundVector();
1247                 }
1248
1249
1250                 if (m->control_pressed) {  out.close(); return 0;  }
1251
1252                 //output error messages about any remaining user labels
1253                 set<string>::iterator it;
1254                 bool needToRun = false;
1255                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
1256                         m->mothurOut("Your file does not include the label " + *it);
1257                         if (processedLabels.count(lastLabel) != 1) {
1258                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
1259                                 needToRun = true;
1260                         }else {
1261                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
1262                         }
1263                 }
1264
1265                 //run last label if you need to
1266                 if (needToRun == true)  {
1267                         if (rabund != NULL) { delete rabund; }
1268
1269                         rabund = input->getRAbundVector(lastLabel);
1270
1271                         m->mothurOut(rabund->getLabel()); m->mothurOutEndLine();
1272
1273                         processRabund(rabund, out);
1274
1275                         delete rabund;
1276                 }
1277
1278                 delete input;
1279                 out.close();
1280
1281                 return 0;
1282
1283         }
1284         catch(exception& e) {
1285                 m->errorOut(e, "SubSampleCommand", "getSubSampleRabund");
1286                 exit(1);
1287         }
1288 }
1289 //**********************************************************************************************************************
1290 int SubSampleCommand::processRabund(RAbundVector*& rabund, ofstream& out) {
1291         try {
1292
1293                 int numBins = rabund->getNumBins();
1294                 int thisSize = rabund->getNumSeqs();
1295
1296                 if (thisSize != size) {
1297
1298                         OrderVector* order = new OrderVector();
1299                         for(int p=0;p<numBins;p++){
1300                                 for(int j=0;j<rabund->get(p);j++){
1301                                         order->push_back(p);
1302                                 }
1303                         }
1304                         random_shuffle(order->begin(), order->end());
1305
1306                         RAbundVector* temp = new RAbundVector(numBins);
1307                         temp->setLabel(rabund->getLabel());
1308
1309                         delete rabund;
1310                         rabund = temp;
1311
1312                         for (int j = 0; j < size; j++) {
1313
1314                                 if (m->control_pressed) { delete order; return 0; }
1315
1316                                 //get random number to sample from order between 0 and thisSize-1.
1317                                 int myrand = int((float)(thisSize) * (float)(rand()) / ((float)RAND_MAX+1.0));
1318
1319                                 int bin = order->get(myrand);
1320
1321                                 int abund = rabund->get(bin);
1322                                 rabund->set(bin, (abund+1));
1323                         }
1324
1325                         delete order;
1326                 }
1327
1328                 if (m->control_pressed) { return 0; }
1329
1330                 rabund->print(out);
1331
1332                 return 0;
1333
1334         }
1335         catch(exception& e) {
1336                 m->errorOut(e, "SubSampleCommand", "processRabund");
1337                 exit(1);
1338         }
1339 }
1340 //**********************************************************************************************************************
1341 int SubSampleCommand::getSubSampleSabund() {
1342         try {
1343
1344                 InputData* input = new InputData(sabundfile, "sabund");
1345                 SAbundVector* sabund = input->getSAbundVector();
1346                 string lastLabel = sabund->getLabel();
1347
1348                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
1349                 set<string> processedLabels;
1350                 set<string> userLabels = labels;
1351
1352                 if (size == 0) { //user has not set size, set size = 10%
1353                         size = int((sabund->getNumSeqs()) * 0.10);
1354                 }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; }
1355
1356
1357                 m->mothurOut("Sampling " + toString(size) + " from " + toString(sabund->getNumSeqs()) + "."); m->mothurOutEndLine();
1358
1359                 string thisOutputDir = outputDir;
1360                 if (outputDir == "") {  thisOutputDir += m->hasPath(sabundfile);  }
1361                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(sabundfile)) + "subsample" + m->getExtension(sabundfile);
1362
1363                 ofstream out;
1364                 m->openOutputFile(outputFileName, out);
1365                 outputTypes["sabund"].push_back(outputFileName);  outputNames.push_back(outputFileName);
1366
1367
1368                 //as long as you are not at the end of the file or done wih the lines you want
1369                 while((sabund != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
1370                         if (m->control_pressed) {  delete input; delete sabund; out.close(); return 0;  }
1371
1372                         if(allLines == 1 || labels.count(sabund->getLabel()) == 1){
1373
1374                                 m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
1375
1376                                 processSabund(sabund, out);
1377
1378                                 processedLabels.insert(sabund->getLabel());
1379                                 userLabels.erase(sabund->getLabel());
1380                         }
1381
1382                         if ((m->anyLabelsToProcess(sabund->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
1383                                 string saveLabel = sabund->getLabel();
1384
1385                                 delete sabund;
1386
1387                                 sabund = input->getSAbundVector(lastLabel);
1388                                 m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
1389
1390                                 processSabund(sabund, out);
1391
1392                                 processedLabels.insert(sabund->getLabel());
1393                                 userLabels.erase(sabund->getLabel());
1394
1395                                 //restore real lastlabel to save below
1396                                 sabund->setLabel(saveLabel);
1397                         }
1398
1399                         lastLabel = sabund->getLabel();
1400
1401                         //prevent memory leak
1402                         delete sabund; sabund = NULL;
1403
1404                         //get next line to process
1405                         sabund = input->getSAbundVector();
1406                 }
1407
1408
1409                 if (m->control_pressed) {  out.close(); return 0;  }
1410
1411                 //output error messages about any remaining user labels
1412                 set<string>::iterator it;
1413                 bool needToRun = false;
1414                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
1415                         m->mothurOut("Your file does not include the label " + *it);
1416                         if (processedLabels.count(lastLabel) != 1) {
1417                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
1418                                 needToRun = true;
1419                         }else {
1420                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
1421                         }
1422                 }
1423
1424                 //run last label if you need to
1425                 if (needToRun == true)  {
1426                         if (sabund != NULL) { delete sabund; }
1427
1428                         sabund = input->getSAbundVector(lastLabel);
1429
1430                         m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
1431
1432                         processSabund(sabund, out);
1433
1434                         delete sabund;
1435                 }
1436
1437                 delete input;
1438                 out.close();
1439
1440                 return 0;
1441
1442         }
1443         catch(exception& e) {
1444                 m->errorOut(e, "SubSampleCommand", "getSubSampleSabund");
1445                 exit(1);
1446         }
1447 }
1448 //**********************************************************************************************************************
1449 int SubSampleCommand::processSabund(SAbundVector*& sabund, ofstream& out) {
1450         try {
1451
1452                 RAbundVector* rabund = new RAbundVector();
1453                 *rabund = sabund->getRAbundVector();
1454
1455                 int numBins = rabund->getNumBins();
1456                 int thisSize = rabund->getNumSeqs();
1457
1458                 if (thisSize != size) {
1459
1460                         OrderVector* order = new OrderVector();
1461                         for(int p=0;p<numBins;p++){
1462                                 for(int j=0;j<rabund->get(p);j++){
1463                                         order->push_back(p);
1464                                 }
1465                         }
1466                         random_shuffle(order->begin(), order->end());
1467
1468                         RAbundVector* temp = new RAbundVector(numBins);
1469                         temp->setLabel(rabund->getLabel());
1470
1471                         delete rabund;
1472                         rabund = temp;
1473
1474                         for (int j = 0; j < size; j++) {
1475
1476                                 if (m->control_pressed) { delete order; return 0; }
1477
1478                                 //get random number to sample from order between 0 and thisSize-1.
1479                                 int myrand = int((float)(thisSize) * (float)(rand()) / ((float)RAND_MAX+1.0));
1480
1481                                 int bin = order->get(myrand);
1482
1483                                 int abund = rabund->get(bin);
1484                                 rabund->set(bin, (abund+1));
1485                         }
1486
1487                         delete order;
1488                 }
1489
1490                 if (m->control_pressed) { return 0; }
1491
1492                 delete sabund;
1493                 sabund = new SAbundVector();
1494                 *sabund = rabund->getSAbundVector();
1495                 delete rabund;
1496
1497                 sabund->print(out);
1498
1499                 return 0;
1500
1501         }
1502         catch(exception& e) {
1503                 m->errorOut(e, "SubSampleCommand", "processSabund");
1504                 exit(1);
1505         }
1506 }
1507 //**********************************************************************************************************************
1508 int SubSampleCommand::eliminateZeroOTUS(vector<SharedRAbundVector*>& thislookup) {
1509         try {
1510
1511                 vector<SharedRAbundVector*> newLookup;
1512                 for (int i = 0; i < thislookup.size(); i++) {
1513                         SharedRAbundVector* temp = new SharedRAbundVector();
1514                         temp->setLabel(thislookup[i]->getLabel());
1515                         temp->setGroup(thislookup[i]->getGroup());
1516                         newLookup.push_back(temp);
1517                 }
1518
1519                 //for each bin
1520                 for (int i = 0; i < thislookup[0]->getNumBins(); i++) {
1521                         if (m->control_pressed) { for (int j = 0; j < newLookup.size(); j++) {  delete newLookup[j];  } return 0; }
1522
1523                         //look at each sharedRabund and make sure they are not all zero
1524                         bool allZero = true;
1525                         for (int j = 0; j < thislookup.size(); j++) {
1526                                 if (thislookup[j]->getAbundance(i) != 0) { allZero = false;  break;  }
1527                         }
1528
1529                         //if they are not all zero add this bin
1530                         if (!allZero) {
1531                                 for (int j = 0; j < thislookup.size(); j++) {
1532                                         newLookup[j]->push_back(thislookup[j]->getAbundance(i), thislookup[j]->getGroup());
1533                                 }
1534                         }
1535                 }
1536
1537                 for (int j = 0; j < thislookup.size(); j++) {  delete thislookup[j];  }
1538                 thislookup.clear();
1539
1540                 thislookup = newLookup;
1541
1542                 return 0;
1543
1544         }
1545         catch(exception& e) {
1546                 m->errorOut(e, "SubSampleCommand", "eliminateZeroOTUS");
1547                 exit(1);
1548         }
1549 }
1550
1551 //**********************************************************************************************************************
1552
1553
1554