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