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 #include "deconvolutecommand.h"
  13
  14 //**********************************************************************************************************************
  15 vector<string> SubSampleCommand::setParameters(){
  16         try {
  17                 CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "FLSSR", "none",false,false); parameters.push_back(pfasta);
  18                 CommandParameter pname("name", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pname);
  19                 CommandParameter pgroup("group", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pgroup);
  20                 CommandParameter plist("list", "InputTypes", "", "", "none", "FLSSR", "none",false,false); parameters.push_back(plist);
  21                 CommandParameter pshared("shared", "InputTypes", "", "", "none", "FLSSR", "none",false,false); parameters.push_back(pshared);
  22                 CommandParameter prabund("rabund", "InputTypes", "", "", "none", "FLSSR", "none",false,false); parameters.push_back(prabund);
  23                 CommandParameter psabund("sabund", "InputTypes", "", "", "none", "FLSSR", "none",false,false); parameters.push_back(psabund);
  24                 CommandParameter plabel("label", "String", "", "", "", "", "",false,false); parameters.push_back(plabel);
  25                 CommandParameter pgroups("groups", "String", "", "", "", "", "",false,false); parameters.push_back(pgroups);
  26                 CommandParameter psize("size", "Number", "", "0", "", "", "",false,false); parameters.push_back(psize);
  27                 CommandParameter ppersample("persample", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(ppersample);
  28                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  29                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  30
  31                 vector<string> myArray;
  32                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  33                 return myArray;
  34         }
  35         catch(exception& e) {
  36                 m->errorOut(e, "SubSampleCommand", "setParameters");
  37                 exit(1);
  38         }
  39 }
  40 //**********************************************************************************************************************
  41 string SubSampleCommand::getHelpString(){
  42         try {
  43                 string helpString = "";
  44                 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";
  45                 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";
  46                 helpString += "The namefile is only used with the fasta file, not with the listfile, because the list file should contain all sequences.\n";
  47                 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";
  48                 helpString += "The label parameter allows you to select what distance levels you would like, and are also separated by dashes.\n";
  49                 helpString += "The size parameter allows you indicate the size of your subsample.\n";
  50                 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";
  51                 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";
  52                 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";
  53                 helpString += "The sub.sample command should be in the following format: sub.sample(list=yourListFile, group=yourGroupFile, groups=yourGroups, label=yourLabels).\n";
  54                 helpString += "Example sub.sample(list=abrecovery.fn.list, group=abrecovery.groups, groups=B-C, size=20).\n";
  55                 helpString += "The default value for groups is all the groups in your groupfile, and all labels in your inputfile will be used.\n";
  56                 helpString += "The sub.sample command outputs a .subsample file.\n";
  57                 helpString += "Note: No spaces between parameter labels (i.e. groups), '=' and parameters (i.e.yourGroups).\n";
  58                 return helpString;
  59         }
  60         catch(exception& e) {
  61                 m->errorOut(e, "SubSampleCommand", "getHelpString");
  62                 exit(1);
  63         }
  64 }
  65 //**********************************************************************************************************************
  66 SubSampleCommand::SubSampleCommand(){
  67         try {
  68                 abort = true; calledHelp = true;
  69                 setParameters();
  70                 vector<string> tempOutNames;
  71                 outputTypes["shared"] = tempOutNames;
  72                 outputTypes["list"] = tempOutNames;
  73                 outputTypes["rabund"] = tempOutNames;
  74                 outputTypes["sabund"] = tempOutNames;
  75                 outputTypes["fasta"] = tempOutNames;
  76                 outputTypes["name"] = tempOutNames;
  77                 outputTypes["group"] = tempOutNames;
  78         }
  79         catch(exception& e) {
  80                 m->errorOut(e, "SubSampleCommand", "GetRelAbundCommand");
  81                 exit(1);
  82         }
  83 }
  84 //**********************************************************************************************************************
  85 SubSampleCommand::SubSampleCommand(string option) {
  86         try {
  87                 abort = false; calledHelp = false;
  88                 allLines = 1;
  89
  90                 //allow user to run help
  91                 if(option == "help") { help(); abort = true; calledHelp = true; }
  92                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  93
  94                 else {
  95                         vector<string> myArray = setParameters();
  96
  97                         OptionParser parser(option);
  98                         map<string,string> parameters = parser.getParameters();
  99
 100                         ValidParameters validParameter;
 101
 102                         //check to make sure all parameters are valid for command
 103                         map<string,string>::iterator it;
 104                         for (it = parameters.begin(); it != parameters.end(); it++) {
 105                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 106                         }
 107
 108                         //initialize outputTypes
 109                         vector<string> tempOutNames;
 110                         outputTypes["shared"] = tempOutNames;
 111                         outputTypes["list"] = tempOutNames;
 112                         outputTypes["rabund"] = tempOutNames;
 113                         outputTypes["sabund"] = tempOutNames;
 114                         outputTypes["fasta"] = tempOutNames;
 115                         outputTypes["name"] = tempOutNames;
 116                         outputTypes["group"] = tempOutNames;
 117
 118                         //if the user changes the output directory command factory will send this info to us in the output parameter
 119                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
 120
 121                         //if the user changes the input directory command factory will send this info to us in the output parameter
 122                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 123                         if (inputDir == "not found"){   inputDir = "";          }
 124                         else {
 125                                 string path;
 126                                 it = parameters.find("list");
 127                                 //user has given a template file
 128                                 if(it != parameters.end()){
 129                                         path = m->hasPath(it->second);
 130                                         //if the user has not given a path then, add inputdir. else leave path alone.
 131                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 132                                 }
 133
 134                                 it = parameters.find("fasta");
 135                                 //user has given a template file
 136                                 if(it != parameters.end()){
 137                                         path = m->hasPath(it->second);
 138                                         //if the user has not given a path then, add inputdir. else leave path alone.
 139                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 140                                 }
 141
 142                                 it = parameters.find("shared");
 143                                 //user has given a template file
 144                                 if(it != parameters.end()){
 145                                         path = m->hasPath(it->second);
 146                                         //if the user has not given a path then, add inputdir. else leave path alone.
 147                                         if (path == "") {       parameters["shared"] = inputDir + it->second;           }
 148                                 }
 149
 150                                 it = parameters.find("group");
 151                                 //user has given a template file
 152                                 if(it != parameters.end()){
 153                                         path = m->hasPath(it->second);
 154                                         //if the user has not given a path then, add inputdir. else leave path alone.
 155                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 156                                 }
 157
 158                                 it = parameters.find("sabund");
 159                                 //user has given a template file
 160                                 if(it != parameters.end()){
 161                                         path = m->hasPath(it->second);
 162                                         //if the user has not given a path then, add inputdir. else leave path alone.
 163                                         if (path == "") {       parameters["sabund"] = inputDir + it->second;           }
 164                                 }
 165
 166                                 it = parameters.find("rabund");
 167                                 //user has given a template file
 168                                 if(it != parameters.end()){
 169                                         path = m->hasPath(it->second);
 170                                         //if the user has not given a path then, add inputdir. else leave path alone.
 171                                         if (path == "") {       parameters["rabund"] = inputDir + it->second;           }
 172                                 }
 173
 174                                 it = parameters.find("name");
 175                                 //user has given a template file
 176                                 if(it != parameters.end()){
 177                                         path = m->hasPath(it->second);
 178                                         //if the user has not given a path then, add inputdir. else leave path alone.
 179                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 180                                 }
 181                         }
 182
 183                         //check for required parameters
 184                         listfile = validParameter.validFile(parameters, "list", true);
 185                         if (listfile == "not open") { listfile = ""; abort = true; }
 186                         else if (listfile == "not found") { listfile = ""; }
 187                         else { m->setListFile(listfile); }
 188
 189                         sabundfile = validParameter.validFile(parameters, "sabund", true);
 190                         if (sabundfile == "not open") { sabundfile = ""; abort = true; }
 191                         else if (sabundfile == "not found") { sabundfile = ""; }
 192                         else { m->setSabundFile(sabundfile); }
 193
 194                         rabundfile = validParameter.validFile(parameters, "rabund", true);
 195                         if (rabundfile == "not open") { rabundfile = ""; abort = true; }
 196                         else if (rabundfile == "not found") { rabundfile = ""; }
 197                         else { m->setRabundFile(rabundfile); }
 198
 199                         fastafile = validParameter.validFile(parameters, "fasta", true);
 200                         if (fastafile == "not open") { fastafile = ""; abort = true; }
 201                         else if (fastafile == "not found") { fastafile = ""; }
 202                         else { m->setFastaFile(fastafile); }
 203
 204                         sharedfile = validParameter.validFile(parameters, "shared", true);
 205                         if (sharedfile == "not open") { sharedfile = ""; abort = true; }
 206                         else if (sharedfile == "not found") { sharedfile = ""; }
 207                         else { m->setSharedFile(sharedfile); }
 208
 209                         namefile = validParameter.validFile(parameters, "name", true);
 210                         if (namefile == "not open") { namefile = ""; abort = true; }
 211                         else if (namefile == "not found") { namefile = ""; }
 212                         else { m->setNameFile(namefile); }
 213
 214                         groupfile = validParameter.validFile(parameters, "group", true);
 215                         if (groupfile == "not open") { groupfile = ""; abort = true; }
 216                         else if (groupfile == "not found") { groupfile = ""; }
 217                         else { m->setGroupFile(groupfile); }
 218
 219                         //check for optional parameter and set defaults
 220                         // ...at some point should added some additional type checking...
 221                         label = validParameter.validFile(parameters, "label", false);
 222                         if (label == "not found") { label = ""; }
 223                         else {
 224                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 225                                 else { allLines = 1;  }
 226                         }
 227
 228                         groups = validParameter.validFile(parameters, "groups", false);
 229                         if (groups == "not found") { groups = ""; pickedGroups = false; }
 230                         else {
 231                                 pickedGroups = true;
 232                                 m->splitAtDash(groups, Groups);
 233                                 m->setGroups(Groups);
 234                         }
 235
 236                         string temp = validParameter.validFile(parameters, "size", false);              if (temp == "not found"){       temp = "0";             }
 237                         m->mothurConvert(temp, size);
 238
 239                         temp = validParameter.validFile(parameters, "persample", false);                if (temp == "not found"){       temp = "f";             }
 240                         persample = m->isTrue(temp);
 241
 242                         if (groupfile == "") { persample = false; }
 243
 244                         if ((namefile != "") && (fastafile == "")) { m->mothurOut("You may only use a namefile with a fastafile."); m->mothurOutEndLine(); abort = true; }
 245
 246                         if ((fastafile == "") && (listfile == "") && (sabundfile == "") && (rabundfile == "") && (sharedfile == "")) {
 247                                 m->mothurOut("You must provide a fasta, list, sabund, rabund or shared file as an input file."); m->mothurOutEndLine(); abort = true; }
 248
 249                         if (pickedGroups && ((groupfile == "") && (sharedfile == ""))) {
 250                                 m->mothurOut("You cannot pick groups without a valid group file or shared file."); m->mothurOutEndLine(); abort = true; }
 251
 252                         if ((groupfile != "") && ((fastafile == "") && (listfile == ""))) {
 253                                 m->mothurOut("Group file only valid with listfile or fastafile."); m->mothurOutEndLine(); abort = true; }
 254
 255                         if ((groupfile != "") && ((fastafile != "") && (listfile != ""))) {
 256                                 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; }
 257
 258                 }
 259
 260         }
 261         catch(exception& e) {
 262                 m->errorOut(e, "SubSampleCommand", "SubSampleCommand");
 263                 exit(1);
 264         }
 265 }
 266 //**********************************************************************************************************************
 267
 268 int SubSampleCommand::execute(){
 269         try {
 270
 271                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 272
 273                 if (sharedfile != "")   {   getSubSampleShared();       }
 274                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]); return 0; } }
 275
 276                 if (listfile != "")             {   getSubSampleList();         }
 277                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]); return 0; } }
 278
 279                 if (rabundfile != "")   {   getSubSampleRabund();       }
 280                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]); return 0; } }
 281
 282                 if (sabundfile != "")   {   getSubSampleSabund();       }
 283                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]); return 0; } }
 284
 285                 if (fastafile != "")    {   getSubSampleFasta();        }
 286                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]); return 0; } }
 287
 288                 //set fasta file as new current fastafile
 289                 string current = "";
 290                 itTypes = outputTypes.find("fasta");
 291                 if (itTypes != outputTypes.end()) {
 292                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 293                 }
 294
 295                 itTypes = outputTypes.find("name");
 296                 if (itTypes != outputTypes.end()) {
 297                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
 298                 }
 299
 300                 itTypes = outputTypes.find("group");
 301                 if (itTypes != outputTypes.end()) {
 302                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setGroupFile(current); }
 303                 }
 304
 305                 itTypes = outputTypes.find("list");
 306                 if (itTypes != outputTypes.end()) {
 307                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 308                 }
 309
 310                 itTypes = outputTypes.find("shared");
 311                 if (itTypes != outputTypes.end()) {
 312                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setSharedFile(current); }
 313                 }
 314
 315                 itTypes = outputTypes.find("rabund");
 316                 if (itTypes != outputTypes.end()) {
 317                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setRabundFile(current); }
 318                 }
 319
 320                 itTypes = outputTypes.find("sabund");
 321                 if (itTypes != outputTypes.end()) {
 322                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setSabundFile(current); }
 323                 }
 324
 325
 326                 m->mothurOutEndLine();
 327                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 328                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 329                 m->mothurOutEndLine();
 330
 331                 return 0;
 332         }
 333         catch(exception& e) {
 334                 m->errorOut(e, "SubSampleCommand", "execute");
 335                 exit(1);
 336         }
 337 }
 338 //**********************************************************************************************************************
 339 int SubSampleCommand::getSubSampleFasta() {
 340         try {
 341
 342                 if (namefile != "") { readNames(); }    //fills names with all names in namefile.
 343                 else { getNames(); }//no name file, so get list of names to pick from
 344
 345                 GroupMap* groupMap;
 346                 if (groupfile != "") {
 347
 348                         groupMap = new GroupMap(groupfile);
 349                         groupMap->readMap();
 350
 351                         //takes care of user setting groupNames that are invalid or setting groups=all
 352                         SharedUtil* util = new SharedUtil();
 353                         vector<string> namesGroups = groupMap->getNamesOfGroups();
 354                         util->setGroups(Groups, namesGroups);
 355                         delete util;
 356
 357                         //file mismatch quit
 358                         if (names.size() != groupMap->getNumSeqs()) {
 359                                 m->mothurOut("[ERROR]: your fasta file contains " + toString(names.size()) + " sequences, and your groupfile contains " + toString(groupMap->getNumSeqs()) + ", please correct.");
 360                                 m->mothurOutEndLine();
 361                                 delete groupMap;
 362                                 return 0;
 363                         }
 364                 }
 365
 366                 if (m->control_pressed) { return 0; }
 367
 368
 369                 //make sure that if your picked groups size is not too big
 370                 int thisSize = names.size();
 371                 if (persample) {
 372                         if (size == 0) { //user has not set size, set size = smallest samples size
 373                                 size = groupMap->getNumSeqs(Groups[0]);
 374                                 for (int i = 1; i < Groups.size(); i++) {
 375                                         int thisSize = groupMap->getNumSeqs(Groups[i]);
 376
 377                                         if (thisSize < size) {  size = thisSize;        }
 378                                 }
 379                         }else { //make sure size is not too large
 380                                 vector<string> newGroups;
 381                                 for (int i = 0; i < Groups.size(); i++) {
 382                                         int thisSize = groupMap->getNumSeqs(Groups[i]);
 383
 384                                         if (thisSize >= size) { newGroups.push_back(Groups[i]); }
 385                                         else {  m->mothurOut("You have selected a size that is larger than " + Groups[i] + " number of sequences, removing " + Groups[i] + "."); m->mothurOutEndLine(); }
 386                                 }
 387                                 Groups = newGroups;
 388                         }
 389
 390                         m->mothurOut("Sampling " + toString(size) + " from each group."); m->mothurOutEndLine();
 391                 }else {
 392                         if (pickedGroups) {
 393                                 int total = 0;
 394                                 for(int i = 0; i < Groups.size(); i++) {
 395                                         total += groupMap->getNumSeqs(Groups[i]);
 396                                 }
 397
 398                                 if (size == 0) { //user has not set size, set size = 10% samples size
 399                                         size = int (total * 0.10);
 400                                 }
 401
 402                                 if (total < size) {
 403                                         if (size != 0) {
 404                                                 m->mothurOut("Your size is too large for the number of groups you selected. Adjusting to " + toString(int (total * 0.10)) + "."); m->mothurOutEndLine();
 405                                         }
 406                                         size = int (total * 0.10);
 407                                 }
 408
 409                                 m->mothurOut("Sampling " + toString(size) + " from " + toString(total) + "."); m->mothurOutEndLine();
 410                         }
 411
 412                         if (size == 0) { //user has not set size, set size = 10% samples size
 413                                 size = int (names.size() * 0.10);
 414                         }
 415
 416                         if (size > thisSize) { m->mothurOut("Your fasta file only contains " + toString(thisSize) + " sequences. Setting size to " + toString(thisSize) + "."); m->mothurOutEndLine();
 417                                 size = thisSize;
 418                         }
 419
 420                         if (!pickedGroups) { m->mothurOut("Sampling " + toString(size) + " from " + toString(thisSize) + "."); m->mothurOutEndLine(); }
 421
 422                 }
 423                 random_shuffle(names.begin(), names.end());
 424
 425                 set<string> subset; //dont want repeat sequence names added
 426                 if (persample) {
 427                         //initialize counts
 428                         map<string, int> groupCounts;
 429                         map<string, int>::iterator itGroupCounts;
 430                         for (int i = 0; i < Groups.size(); i++) { groupCounts[Groups[i]] = 0; }
 431
 432                         for (int j = 0; j < names.size(); j++) {
 433
 434                                 if (m->control_pressed) { return 0; }
 435
 436                                 string group = groupMap->getGroup(names[j]);
 437                                 if (group == "not found") { m->mothurOut("[ERROR]: " + names[j] + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
 438                                 else{
 439                                         itGroupCounts = groupCounts.find(group);
 440                                         if (itGroupCounts != groupCounts.end()) {
 441                                                 if (groupCounts[group] < size) {        subset.insert(names[j]);        groupCounts[group]++; }
 442                                         }
 443                                 }
 444                         }
 445                 }else {
 446
 447                         //randomly select a subset of those names to include in the subsample
 448                         //since names was randomly shuffled just grab the next one
 449                         for (int j = 0; j < names.size(); j++) {
 450
 451                                 if (m->control_pressed) { return 0; }
 452
 453                                 if (groupfile != "") { //if there is a groupfile given fill in group info
 454                                         string group = groupMap->getGroup(names[j]);
 455                                         if (group == "not found") { m->mothurOut("[ERROR]: " + names[j] + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
 456
 457                                         if (pickedGroups) { //if hte user picked groups, we only want to keep the names of sequences from those groups
 458                                                 if (m->inUsersGroups(group, Groups)) {
 459                                                         subset.insert(names[j]);
 460                                                 }
 461                                         }else{
 462                                                 subset.insert(names[j]);
 463                                         }
 464                                 }else{ //save everyone, group
 465                                         subset.insert(names[j]);
 466                                 }
 467
 468                                 //do we have enough??
 469                                 if (subset.size() == size) { break; }
 470                         }
 471                 }
 472
 473                 if (subset.size() == 0) {  m->mothurOut("The size you selected is too large, skipping fasta file."); m->mothurOutEndLine();  return 0; }
 474
 475                 string thisOutputDir = outputDir;
 476                 if (outputDir == "") {  thisOutputDir += m->hasPath(fastafile);  }
 477                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(fastafile)) + "subsample" + m->getExtension(fastafile);
 478
 479                 ofstream out;
 480                 m->openOutputFile(outputFileName, out);
 481
 482                 //read through fasta file outputting only the names on the subsample list
 483                 ifstream in;
 484                 m->openInputFile(fastafile, in);
 485
 486                 string thisname;
 487                 int count = 0;
 488                 map<string, vector<string> >::iterator itNameMap;
 489
 490                 while(!in.eof()){
 491
 492                         if (m->control_pressed) { in.close(); out.close();  return 0; }
 493
 494                         Sequence currSeq(in);
 495                         thisname = currSeq.getName();
 496
 497                         if (thisname != "") {
 498
 499                                 //does the subset contain a sequence that this sequence represents
 500                                 itNameMap = nameMap.find(thisname);
 501                                 if (itNameMap != nameMap.end()) {
 502                                         vector<string> nameRepresents = itNameMap->second;
 503
 504                                         for (int i = 0; i < nameRepresents.size(); i++){
 505                                                 if (subset.count(nameRepresents[i]) != 0) {
 506                                                         out << ">" << nameRepresents[i] << endl << currSeq.getAligned() << endl;
 507                                                         count++;
 508                                                 }
 509                                         }
 510                                 }else{
 511                                         m->mothurOut("[ERROR]: " + thisname + " is not in your namefile, please correct."); m->mothurOutEndLine();
 512                                 }
 513                         }
 514                         m->gobble(in);
 515                 }
 516                 in.close();
 517                 out.close();
 518
 519                 if (count != subset.size()) {
 520                         m->mothurOut("[ERROR]: The subset selected contained " + toString(subset.size()) + " sequences, but I only found " + toString(count) + " of those in the fastafile."); m->mothurOutEndLine();
 521                 }
 522
 523                 if (namefile != "") {
 524                         m->mothurOut("Deconvoluting subsampled fasta file... "); m->mothurOutEndLine();
 525
 526                         //use unique.seqs to create new name and fastafile
 527                         string inputString = "fasta=" + outputFileName;
 528                         m->mothurOut("/******************************************/"); m->mothurOutEndLine();
 529                         m->mothurOut("Running command: unique.seqs(" + inputString + ")"); m->mothurOutEndLine();
 530
 531                         Command* uniqueCommand = new DeconvoluteCommand(inputString);
 532                         uniqueCommand->execute();
 533
 534                         map<string, vector<string> > filenames = uniqueCommand->getOutputFiles();
 535
 536                         delete uniqueCommand;
 537
 538                         outputTypes["name"].push_back(filenames["name"][0]);  outputNames.push_back(filenames["name"][0]);
 539                         m->mothurRemove(outputFileName);
 540                         outputFileName = filenames["fasta"][0];
 541
 542                         m->mothurOut("/******************************************/"); m->mothurOutEndLine();
 543
 544                         m->mothurOut("Done."); m->mothurOutEndLine();
 545                 }
 546
 547                 outputTypes["fasta"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 548
 549                 //if a groupfile is provided read through the group file only outputting the names on the subsample list
 550                 if (groupfile != "") {
 551
 552                         string groupOutputDir = outputDir;
 553                         if (outputDir == "") {  groupOutputDir += m->hasPath(groupfile);  }
 554                         string groupOutputFileName = groupOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "subsample" + m->getExtension(groupfile);
 555
 556                         ofstream outGroup;
 557                         m->openOutputFile(groupOutputFileName, outGroup);
 558                         outputTypes["group"].push_back(groupOutputFileName);  outputNames.push_back(groupOutputFileName);
 559
 560                         ifstream inGroup;
 561                         m->openInputFile(groupfile, inGroup);
 562                         string name, group;
 563
 564                         while(!inGroup.eof()){
 565
 566                                 if (m->control_pressed) { inGroup.close(); outGroup.close(); return 0; }
 567
 568                                 inGroup >> name;        m->gobble(inGroup);                     //read from first column
 569                                 inGroup >> group;                       //read from second column
 570
 571                                 //if this name is in the accnos file
 572                                 if (subset.count(name) != 0) {
 573                                         outGroup << name << '\t' << group << endl;
 574                                         subset.erase(name);
 575                                 }
 576
 577                                 m->gobble(inGroup);
 578                         }
 579                         inGroup.close();
 580                         outGroup.close();
 581
 582                         //sanity check
 583                         if (subset.size() != 0) {
 584                                 m->mothurOut("Your groupfile does not match your fasta file."); m->mothurOutEndLine();
 585                                 for (set<string>::iterator it = subset.begin(); it != subset.end(); it++) {
 586                                         m->mothurOut("[ERROR]: " + *it + " is missing from your groupfile."); m->mothurOutEndLine();
 587                                 }
 588                         }
 589                 }
 590
 591                 return 0;
 592
 593         }
 594         catch(exception& e) {
 595                 m->errorOut(e, "SubSampleCommand", "getSubSampleFasta");
 596                 exit(1);
 597         }
 598 }
 599 //**********************************************************************************************************************
 600 int SubSampleCommand::getNames() {
 601         try {
 602
 603                 ifstream in;
 604                 m->openInputFile(fastafile, in);
 605
 606                 string thisname;
 607                 while(!in.eof()){
 608
 609                         if (m->control_pressed) { in.close(); return 0; }
 610
 611                         Sequence currSeq(in);
 612                         thisname = currSeq.getName();
 613
 614                         if (thisname != "") {
 615                                 vector<string> temp; temp.push_back(thisname);
 616                                 nameMap[thisname] = temp;
 617                                 names.push_back(thisname);
 618                         }
 619                         m->gobble(in);
 620                 }
 621                 in.close();
 622
 623                 return 0;
 624
 625         }
 626         catch(exception& e) {
 627                 m->errorOut(e, "SubSampleCommand", "getNames");
 628                 exit(1);
 629         }
 630 }
 631 //**********************************************************************************************************************
 632 int SubSampleCommand::readNames() {
 633         try {
 634
 635                 ifstream in;
 636                 m->openInputFile(namefile, in);
 637
 638                 string thisname, repnames;
 639                 map<string, vector<string> >::iterator it;
 640
 641                 while(!in.eof()){
 642
 643                         if (m->control_pressed) { in.close(); return 0; }
 644
 645                         in >> thisname;         m->gobble(in);          //read from first column
 646                         in >> repnames;                 //read from second column
 647
 648                         it = nameMap.find(thisname);
 649                         if (it == nameMap.end()) {
 650
 651                                 vector<string> splitRepNames;
 652                                 m->splitAtComma(repnames, splitRepNames);
 653
 654                                 nameMap[thisname] = splitRepNames;
 655                                 for (int i = 0; i < splitRepNames.size(); i++) { names.push_back(splitRepNames[i]); }
 656
 657                         }else{  m->mothurOut(thisname + " is already in namesfile. I will use first definition."); m->mothurOutEndLine();  }
 658
 659                         m->gobble(in);
 660                 }
 661                 in.close();
 662
 663                 return 0;
 664
 665         }
 666         catch(exception& e) {
 667                 m->errorOut(e, "SubSampleCommand", "readNames");
 668                 exit(1);
 669         }
 670 }
 671 //**********************************************************************************************************************
 672 int SubSampleCommand::getSubSampleShared() {
 673         try {
 674
 675                 InputData* input = new InputData(sharedfile, "sharedfile");
 676                 vector<SharedRAbundVector*> lookup = input->getSharedRAbundVectors();
 677                 string lastLabel = lookup[0]->getLabel();
 678
 679                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 680                 set<string> processedLabels;
 681                 set<string> userLabels = labels;
 682
 683                 if (size == 0) { //user has not set size, set size = smallest samples size
 684                         size = lookup[0]->getNumSeqs();
 685                         for (int i = 1; i < lookup.size(); i++) {
 686                                 int thisSize = lookup[i]->getNumSeqs();
 687
 688                                 if (thisSize < size) {  size = thisSize;        }
 689                         }
 690                 }else {
 691                         m->clearGroups();
 692                         Groups.clear();
 693                         vector<SharedRAbundVector*> temp;
 694                         for (int i = 0; i < lookup.size(); i++) {
 695                                 if (lookup[i]->getNumSeqs() < size) {
 696                                         m->mothurOut(lookup[i]->getGroup() + " contains " + toString(lookup[i]->getNumSeqs()) + ". Eliminating."); m->mothurOutEndLine();
 697                                         delete lookup[i];
 698                                 }else {
 699                                         Groups.push_back(lookup[i]->getGroup());
 700                                         temp.push_back(lookup[i]);
 701                                 }
 702                         }
 703                         lookup = temp;
 704                         m->setGroups(Groups);
 705                 }
 706
 707                 if (lookup.size() == 0) {  m->mothurOut("The size you selected is too large, skipping shared file."); m->mothurOutEndLine(); delete input; return 0; }
 708
 709                 m->mothurOut("Sampling " + toString(size) + " from each group."); m->mothurOutEndLine();
 710
 711                 //as long as you are not at the end of the file or done wih the lines you want
 712                 while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 713                         if (m->control_pressed) {  delete input; for (int i = 0; i < lookup.size(); i++) {  delete lookup[i]; lookup[i] = NULL; }  return 0;  }
 714
 715                         if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){
 716
 717                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 718
 719                                 processShared(lookup);
 720
 721                                 processedLabels.insert(lookup[0]->getLabel());
 722                                 userLabels.erase(lookup[0]->getLabel());
 723                         }
 724
 725                         if ((m->anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 726                                 string saveLabel = lookup[0]->getLabel();
 727
 728                                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 729
 730                                 lookup = input->getSharedRAbundVectors(lastLabel);
 731                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 732
 733                                 processShared(lookup);
 734
 735                                 processedLabels.insert(lookup[0]->getLabel());
 736                                 userLabels.erase(lookup[0]->getLabel());
 737
 738                                 //restore real lastlabel to save below
 739                                 lookup[0]->setLabel(saveLabel);
 740                         }
 741
 742                         lastLabel = lookup[0]->getLabel();
 743                         //prevent memory leak
 744                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i]; lookup[i] = NULL; }
 745
 746                         //get next line to process
 747                         lookup = input->getSharedRAbundVectors();
 748                 }
 749
 750
 751                 if (m->control_pressed) {   return 0;  }
 752
 753                 //output error messages about any remaining user labels
 754                 set<string>::iterator it;
 755                 bool needToRun = false;
 756                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 757                         m->mothurOut("Your file does not include the label " + *it);
 758                         if (processedLabels.count(lastLabel) != 1) {
 759                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 760                                 needToRun = true;
 761                         }else {
 762                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 763                         }
 764                 }
 765
 766                 //run last label if you need to
 767                 if (needToRun == true)  {
 768                         for (int i = 0; i < lookup.size(); i++) { if (lookup[i] != NULL) { delete lookup[i]; } }
 769                         lookup = input->getSharedRAbundVectors(lastLabel);
 770
 771                         m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 772
 773                         processShared(lookup);
 774
 775                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 776                 }
 777
 778                 delete input;
 779
 780                 return 0;
 781
 782         }
 783         catch(exception& e) {
 784                 m->errorOut(e, "SubSampleCommand", "getSubSampleShared");
 785                 exit(1);
 786         }
 787 }
 788 //**********************************************************************************************************************
 789 int SubSampleCommand::processShared(vector<SharedRAbundVector*>& thislookup) {
 790         try {
 791
 792                 //save mothurOut's binLabels to restore for next label
 793                 vector<string> saveBinLabels = m->currentBinLabels;
 794
 795                 string thisOutputDir = outputDir;
 796                 if (outputDir == "") {  thisOutputDir += m->hasPath(sharedfile);  }
 797                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(sharedfile)) + thislookup[0]->getLabel() + ".subsample" + m->getExtension(sharedfile);
 798
 799
 800                 ofstream out;
 801                 m->openOutputFile(outputFileName, out);
 802                 outputTypes["shared"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 803
 804                 int numBins = thislookup[0]->getNumBins();
 805                 for (int i = 0; i < thislookup.size(); i++) {
 806                         int thisSize = thislookup[i]->getNumSeqs();
 807
 808                         if (thisSize != size) {
 809
 810                                 string thisgroup = thislookup[i]->getGroup();
 811
 812                                 OrderVector* order = new OrderVector();
 813                                 for(int p=0;p<numBins;p++){
 814                                         for(int j=0;j<thislookup[i]->getAbundance(p);j++){
 815                                                 order->push_back(p);
 816                                         }
 817                                 }
 818                                 random_shuffle(order->begin(), order->end());
 819
 820                                 SharedRAbundVector* temp = new SharedRAbundVector(numBins);
 821                                 temp->setLabel(thislookup[i]->getLabel());
 822                                 temp->setGroup(thislookup[i]->getGroup());
 823
 824                                 delete thislookup[i];
 825                                 thislookup[i] = temp;
 826
 827
 828                                 for (int j = 0; j < size; j++) {
 829
 830                                         if (m->control_pressed) { delete order; out.close(); return 0; }
 831
 832                                         //get random number to sample from order between 0 and thisSize-1.
 833                                         //don't need this because of the random shuffle above
 834                                         //int myrand = int((float)(thisSize) * (float)(rand()) / ((float)RAND_MAX+1.0));
 835
 836                                         int bin = order->get(j);
 837
 838                                         int abund = thislookup[i]->getAbundance(bin);
 839                                         thislookup[i]->set(bin, (abund+1), thisgroup);
 840                                 }
 841                                 delete order;
 842                         }
 843                 }
 844
 845                 //subsampling may have created some otus with no sequences in them
 846                 eliminateZeroOTUS(thislookup);
 847
 848                 if (m->control_pressed) { out.close(); return 0; }
 849
 850                 thislookup[0]->printHeaders(out);
 851
 852                 for (int i = 0; i < thislookup.size(); i++) {
 853                         out << thislookup[i]->getLabel() << '\t' << thislookup[i]->getGroup() << '\t';
 854                         thislookup[i]->print(out);
 855                 }
 856
 857                 out.close();
 858
 859                 //save mothurOut's binLabels to restore for next label
 860                 m->currentBinLabels = saveBinLabels;
 861
 862                 return 0;
 863
 864         }
 865         catch(exception& e) {
 866                 m->errorOut(e, "SubSampleCommand", "processShared");
 867                 exit(1);
 868         }
 869 }
 870 //**********************************************************************************************************************
 871 int SubSampleCommand::getSubSampleList() {
 872         try {
 873
 874                 string thisOutputDir = outputDir;
 875                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 876                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "subsample" + m->getExtension(listfile);
 877
 878                 ofstream out;
 879                 m->openOutputFile(outputFileName, out);
 880                 outputTypes["list"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 881
 882                 InputData* input = new InputData(listfile, "list");
 883                 ListVector* list = input->getListVector();
 884                 string lastLabel = list->getLabel();
 885
 886                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 887                 set<string> processedLabels;
 888                 set<string> userLabels = labels;
 889
 890                 ofstream outGroup;
 891                 GroupMap* groupMap;
 892                 if (groupfile != "") {
 893
 894                         groupMap = new GroupMap(groupfile);
 895                         groupMap->readMap();
 896
 897                         //takes care of user setting groupNames that are invalid or setting groups=all
 898                         SharedUtil* util = new SharedUtil();
 899                         vector<string> namesGroups = groupMap->getNamesOfGroups();
 900                         util->setGroups(Groups, namesGroups);
 901                         delete util;
 902
 903                         //create outputfiles
 904                         string groupOutputDir = outputDir;
 905                         if (outputDir == "") {  groupOutputDir += m->hasPath(groupfile);  }
 906                         string groupOutputFileName = groupOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "subsample" + m->getExtension(groupfile);
 907
 908                         m->openOutputFile(groupOutputFileName, outGroup);
 909                         outputTypes["group"].push_back(groupOutputFileName);  outputNames.push_back(groupOutputFileName);
 910
 911                         //file mismatch quit
 912                         if (list->getNumSeqs() != groupMap->getNumSeqs()) {
 913                                 m->mothurOut("[ERROR]: your list file contains " + toString(list->getNumSeqs()) + " sequences, and your groupfile contains " + toString(groupMap->getNumSeqs()) + ", please correct.");
 914                                 m->mothurOutEndLine();
 915                                 delete groupMap;
 916                                 delete list;
 917                                 delete input;
 918                                 out.close();
 919                                 outGroup.close();
 920                                 return 0;
 921                         }
 922                 }
 923
 924                 //make sure that if your picked groups size is not too big
 925                 if (persample) {
 926                         if (size == 0) { //user has not set size, set size = smallest samples size
 927                                 size = groupMap->getNumSeqs(Groups[0]);
 928                                 for (int i = 1; i < Groups.size(); i++) {
 929                                         int thisSize = groupMap->getNumSeqs(Groups[i]);
 930
 931                                         if (thisSize < size) {  size = thisSize;        }
 932                                 }
 933                         }else { //make sure size is not too large
 934                                 vector<string> newGroups;
 935                                 for (int i = 0; i < Groups.size(); i++) {
 936                                         int thisSize = groupMap->getNumSeqs(Groups[i]);
 937
 938                                         if (thisSize >= size) { newGroups.push_back(Groups[i]); }
 939                                         else {  m->mothurOut("You have selected a size that is larger than " + Groups[i] + " number of sequences, removing " + Groups[i] + "."); m->mothurOutEndLine(); }
 940                                 }
 941                                 Groups = newGroups;
 942                         }
 943
 944                         m->mothurOut("Sampling " + toString(size) + " from each group."); m->mothurOutEndLine();
 945                 }else{
 946                         if (pickedGroups) {
 947                                 int total = 0;
 948                                 for(int i = 0; i < Groups.size(); i++) {
 949                                         total += groupMap->getNumSeqs(Groups[i]);
 950                                 }
 951
 952                                 if (size == 0) { //user has not set size, set size = 10% samples size
 953                                         size = int (total * 0.10);
 954                                 }
 955
 956                                 if (total < size) {
 957                                         m->mothurOut("Your size is too large for the number of groups you selected. Adjusting to " + toString(int (total * 0.10)) + "."); m->mothurOutEndLine();
 958                                         size = int (total * 0.10);
 959                                 }
 960
 961                                 m->mothurOut("Sampling " + toString(size) + " from " + toString(total) + "."); m->mothurOutEndLine();
 962                         }else{
 963
 964                                 if (size == 0) { //user has not set size, set size = 10% samples size
 965                                         size = int (list->getNumSeqs() * 0.10);
 966                                 }
 967
 968                                 int thisSize = list->getNumSeqs();
 969                                 if (size > thisSize) { m->mothurOut("Your list file only contains " + toString(thisSize) + " sequences. Setting size to " + toString(thisSize) + "."); m->mothurOutEndLine();
 970                                         size = thisSize;
 971                                 }
 972
 973                                 m->mothurOut("Sampling " + toString(size) + " from " + toString(list->getNumSeqs()) + "."); m->mothurOutEndLine();
 974                         }
 975                 }
 976
 977
 978                 //fill names
 979                 for (int i = 0; i < list->getNumBins(); i++) {
 980                         string binnames = list->get(i);
 981
 982                         //parse names
 983                         string individual = "";
 984                         int length = binnames.length();
 985                         for(int j=0;j<length;j++){
 986                                 if(binnames[j] == ','){
 987
 988                                         if (groupfile != "") { //if there is a groupfile given fill in group info
 989                                                 string group = groupMap->getGroup(individual);
 990                                                 if (group == "not found") { m->mothurOut("[ERROR]: " + individual + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
 991
 992                                                 if (pickedGroups) { //if hte user picked groups, we only want to keep the names of sequences from those groups
 993                                                         if (m->inUsersGroups(group, Groups)) {
 994                                                                 names.push_back(individual);
 995                                                         }
 996                                                 }else{
 997                                                         names.push_back(individual);
 998                                                 }
 999                                         }else{ //save everyone, group
1000                                                 names.push_back(individual);
1001                                         }
1002                                         individual = "";
1003                                 }
1004                                 else{
1005                                         individual += binnames[j];
1006                                 }
1007                         }
1008                         //save last name
1009                         if (groupfile != "") { //if there is a groupfile given fill in group info
1010                                 string group = groupMap->getGroup(individual);
1011                                 if (group == "not found") { m->mothurOut("[ERROR]: " + individual + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
1012
1013                                 if (pickedGroups) { //if hte user picked groups, we only want to keep the names of sequences from those groups
1014                                         if (m->inUsersGroups(group, Groups)) {
1015                                                 names.push_back(individual);
1016                                         }
1017                                 }else{
1018                                         names.push_back(individual);
1019                                 }
1020                         }else{ //save everyone, group
1021                                 names.push_back(individual);
1022                         }
1023                 }
1024
1025                 random_shuffle(names.begin(), names.end());
1026
1027                 //randomly select a subset of those names to include in the subsample
1028                 set<string> subset; //dont want repeat sequence names added
1029                 if (persample) {
1030                         //initialize counts
1031                         map<string, int> groupCounts;
1032                         map<string, int>::iterator itGroupCounts;
1033                         for (int i = 0; i < Groups.size(); i++) { groupCounts[Groups[i]] = 0; }
1034
1035                         for (int j = 0; j < names.size(); j++) {
1036
1037                                 if (m->control_pressed) { return 0; }
1038
1039                                 string group = groupMap->getGroup(names[j]);
1040                                 if (group == "not found") { m->mothurOut("[ERROR]: " + names[j] + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
1041                                 else{
1042                                         itGroupCounts = groupCounts.find(group);
1043                                         if (itGroupCounts != groupCounts.end()) {
1044                                                 if (groupCounts[group] < size) {        subset.insert(names[j]);        groupCounts[group]++; }
1045                                         }
1046                                 }
1047                         }
1048                 }else{
1049                         for (int j = 0; j < size; j++) {
1050
1051                                 if (m->control_pressed) { break; }
1052
1053                                 subset.insert(names[j]);
1054                         }
1055                 }
1056
1057                 if (groupfile != "") {
1058                         //write out new groupfile
1059                         for (set<string>::iterator it = subset.begin(); it != subset.end(); it++) {
1060                                 string group = groupMap->getGroup(*it);
1061                                 if (group == "not found") { group = "NOTFOUND"; }
1062
1063                                 outGroup << *it << '\t' << group << endl;
1064                         }
1065                         outGroup.close(); delete groupMap;
1066                 }
1067
1068
1069                 //as long as you are not at the end of the file or done wih the lines you want
1070                 while((list != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
1071
1072                         if (m->control_pressed) {  delete list; delete input; out.close();  return 0;  }
1073
1074                         if(allLines == 1 || labels.count(list->getLabel()) == 1){
1075
1076                                 m->mothurOut(list->getLabel()); m->mothurOutEndLine();
1077
1078                                 processList(list, out, subset);
1079
1080                                 processedLabels.insert(list->getLabel());
1081                                 userLabels.erase(list->getLabel());
1082                         }
1083
1084                         if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
1085                                 string saveLabel = list->getLabel();
1086
1087                                 delete list;
1088
1089                                 list = input->getListVector(lastLabel);
1090                                 m->mothurOut(list->getLabel()); m->mothurOutEndLine();
1091
1092                                 processList(list, out, subset);
1093
1094                                 processedLabels.insert(list->getLabel());
1095                                 userLabels.erase(list->getLabel());
1096
1097                                 //restore real lastlabel to save below
1098                                 list->setLabel(saveLabel);
1099                         }
1100
1101                         lastLabel = list->getLabel();
1102
1103                         delete list; list = NULL;
1104
1105                         //get next line to process
1106                         list = input->getListVector();
1107                 }
1108
1109
1110                 if (m->control_pressed) {  if (list != NULL) { delete list; } delete input; out.close(); return 0;  }
1111
1112                 //output error messages about any remaining user labels
1113                 set<string>::iterator it;
1114                 bool needToRun = false;
1115                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
1116                         m->mothurOut("Your file does not include the label " + *it);
1117                         if (processedLabels.count(lastLabel) != 1) {
1118                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
1119                                 needToRun = true;
1120                         }else {
1121                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
1122                         }
1123                 }
1124
1125                 //run last label if you need to
1126                 if (needToRun == true)  {
1127                         if (list != NULL) { delete list; }
1128
1129                         list = input->getListVector(lastLabel);
1130
1131                         m->mothurOut(list->getLabel()); m->mothurOutEndLine();
1132
1133                         processList(list, out, subset);
1134
1135                         delete list; list = NULL;
1136                 }
1137
1138                 out.close();
1139                 if (list != NULL) { delete list; }
1140                 delete input;
1141
1142                 return 0;
1143
1144         }
1145         catch(exception& e) {
1146                 m->errorOut(e, "SubSampleCommand", "getSubSampleList");
1147                 exit(1);
1148         }
1149 }
1150 //**********************************************************************************************************************
1151 int SubSampleCommand::processList(ListVector*& list, ofstream& out, set<string>& subset) {
1152         try {
1153
1154                 int numBins = list->getNumBins();
1155
1156                 ListVector* temp = new ListVector();
1157                 temp->setLabel(list->getLabel());
1158
1159                 for (int i = 0; i < numBins; i++) {
1160
1161                         if (m->control_pressed) { break; }
1162
1163                         string binnames = list->get(i);
1164
1165                         //parse names
1166                         string individual = "";
1167                         string newNames = "";
1168                         int length = binnames.length();
1169                         for(int j=0;j<length;j++){
1170                                 if(binnames[j] == ','){
1171                                         if (subset.count(individual) != 0) {  newNames += individual + ",";  }
1172                                         individual = "";
1173                                 }else{
1174                                         individual += binnames[j];
1175                                 }
1176                         }
1177                         if (subset.count(individual) != 0) {  newNames += individual + ",";  }
1178
1179
1180                         //if there are names in this bin add to new list
1181                         if (newNames != "") {
1182                                 newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
1183                                 temp->push_back(newNames);
1184                         }
1185                 }
1186
1187                 delete list;
1188                 list = temp;
1189
1190                 if (m->control_pressed) { return 0; }
1191
1192                 list->print(out);
1193
1194                 return 0;
1195
1196         }
1197         catch(exception& e) {
1198                 m->errorOut(e, "SubSampleCommand", "processList");
1199                 exit(1);
1200         }
1201 }
1202 //**********************************************************************************************************************
1203 int SubSampleCommand::getSubSampleRabund() {
1204         try {
1205                 InputData* input = new InputData(rabundfile, "rabund");
1206                 RAbundVector* rabund = input->getRAbundVector();
1207                 string lastLabel = rabund->getLabel();
1208
1209                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
1210                 set<string> processedLabels;
1211                 set<string> userLabels = labels;
1212
1213                 if (size == 0) { //user has not set size, set size = 10%
1214                         size = int((rabund->getNumSeqs()) * 0.10);
1215                 }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; }
1216
1217                 m->mothurOut("Sampling " + toString(size) + " from " + toString(rabund->getNumSeqs()) + "."); m->mothurOutEndLine();
1218
1219                 string thisOutputDir = outputDir;
1220                 if (outputDir == "") {  thisOutputDir += m->hasPath(rabundfile);  }
1221                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(rabundfile)) + "subsample" + m->getExtension(rabundfile);
1222
1223                 ofstream out;
1224                 m->openOutputFile(outputFileName, out);
1225                 outputTypes["rabund"].push_back(outputFileName);  outputNames.push_back(outputFileName);
1226
1227                 //as long as you are not at the end of the file or done wih the lines you want
1228                 while((rabund != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
1229                         if (m->control_pressed) {  delete input; delete rabund; out.close(); return 0;  }
1230
1231                         if(allLines == 1 || labels.count(rabund->getLabel()) == 1){
1232
1233                                 m->mothurOut(rabund->getLabel()); m->mothurOutEndLine();
1234
1235                                 processRabund(rabund, out);
1236
1237                                 processedLabels.insert(rabund->getLabel());
1238                                 userLabels.erase(rabund->getLabel());
1239                         }
1240
1241                         if ((m->anyLabelsToProcess(rabund->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
1242                                 string saveLabel = rabund->getLabel();
1243
1244                                 delete rabund;
1245
1246                                 rabund = input->getRAbundVector(lastLabel);
1247                                 m->mothurOut(rabund->getLabel()); m->mothurOutEndLine();
1248
1249                                 processRabund(rabund, out);
1250
1251                                 processedLabels.insert(rabund->getLabel());
1252                                 userLabels.erase(rabund->getLabel());
1253
1254                                 //restore real lastlabel to save below
1255                                 rabund->setLabel(saveLabel);
1256                         }
1257
1258                         lastLabel = rabund->getLabel();
1259
1260                         //prevent memory leak
1261                         delete rabund; rabund = NULL;
1262
1263                         //get next line to process
1264                         rabund = input->getRAbundVector();
1265                 }
1266
1267
1268                 if (m->control_pressed) {  out.close(); return 0;  }
1269
1270                 //output error messages about any remaining user labels
1271                 set<string>::iterator it;
1272                 bool needToRun = false;
1273                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
1274                         m->mothurOut("Your file does not include the label " + *it);
1275                         if (processedLabels.count(lastLabel) != 1) {
1276                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
1277                                 needToRun = true;
1278                         }else {
1279                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
1280                         }
1281                 }
1282
1283                 //run last label if you need to
1284                 if (needToRun == true)  {
1285                         if (rabund != NULL) { delete rabund; }
1286
1287                         rabund = input->getRAbundVector(lastLabel);
1288
1289                         m->mothurOut(rabund->getLabel()); m->mothurOutEndLine();
1290
1291                         processRabund(rabund, out);
1292
1293                         delete rabund;
1294                 }
1295
1296                 delete input;
1297                 out.close();
1298
1299                 return 0;
1300
1301         }
1302         catch(exception& e) {
1303                 m->errorOut(e, "SubSampleCommand", "getSubSampleRabund");
1304                 exit(1);
1305         }
1306 }
1307 //**********************************************************************************************************************
1308 int SubSampleCommand::processRabund(RAbundVector*& rabund, ofstream& out) {
1309         try {
1310
1311                 int numBins = rabund->getNumBins();
1312                 int thisSize = rabund->getNumSeqs();
1313
1314                 if (thisSize != size) {
1315
1316                         OrderVector* order = new OrderVector();
1317                         for(int p=0;p<numBins;p++){
1318                                 for(int j=0;j<rabund->get(p);j++){
1319                                         order->push_back(p);
1320                                 }
1321                         }
1322                         random_shuffle(order->begin(), order->end());
1323
1324                         RAbundVector* temp = new RAbundVector(numBins);
1325                         temp->setLabel(rabund->getLabel());
1326
1327                         delete rabund;
1328                         rabund = temp;
1329
1330                         for (int j = 0; j < size; j++) {
1331
1332                                 if (m->control_pressed) { delete order; return 0; }
1333
1334                                 int bin = order->get(j);
1335
1336                                 int abund = rabund->get(bin);
1337                                 rabund->set(bin, (abund+1));
1338                         }
1339
1340                         delete order;
1341                 }
1342
1343                 if (m->control_pressed) { return 0; }
1344
1345                 rabund->print(out);
1346
1347                 return 0;
1348
1349         }
1350         catch(exception& e) {
1351                 m->errorOut(e, "SubSampleCommand", "processRabund");
1352                 exit(1);
1353         }
1354 }
1355 //**********************************************************************************************************************
1356 int SubSampleCommand::getSubSampleSabund() {
1357         try {
1358
1359                 InputData* input = new InputData(sabundfile, "sabund");
1360                 SAbundVector* sabund = input->getSAbundVector();
1361                 string lastLabel = sabund->getLabel();
1362
1363                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
1364                 set<string> processedLabels;
1365                 set<string> userLabels = labels;
1366
1367                 if (size == 0) { //user has not set size, set size = 10%
1368                         size = int((sabund->getNumSeqs()) * 0.10);
1369                 }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; }
1370
1371
1372                 m->mothurOut("Sampling " + toString(size) + " from " + toString(sabund->getNumSeqs()) + "."); m->mothurOutEndLine();
1373
1374                 string thisOutputDir = outputDir;
1375                 if (outputDir == "") {  thisOutputDir += m->hasPath(sabundfile);  }
1376                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(sabundfile)) + "subsample" + m->getExtension(sabundfile);
1377
1378                 ofstream out;
1379                 m->openOutputFile(outputFileName, out);
1380                 outputTypes["sabund"].push_back(outputFileName);  outputNames.push_back(outputFileName);
1381
1382
1383                 //as long as you are not at the end of the file or done wih the lines you want
1384                 while((sabund != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
1385                         if (m->control_pressed) {  delete input; delete sabund; out.close(); return 0;  }
1386
1387                         if(allLines == 1 || labels.count(sabund->getLabel()) == 1){
1388
1389                                 m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
1390
1391                                 processSabund(sabund, out);
1392
1393                                 processedLabels.insert(sabund->getLabel());
1394                                 userLabels.erase(sabund->getLabel());
1395                         }
1396
1397                         if ((m->anyLabelsToProcess(sabund->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
1398                                 string saveLabel = sabund->getLabel();
1399
1400                                 delete sabund;
1401
1402                                 sabund = input->getSAbundVector(lastLabel);
1403                                 m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
1404
1405                                 processSabund(sabund, out);
1406
1407                                 processedLabels.insert(sabund->getLabel());
1408                                 userLabels.erase(sabund->getLabel());
1409
1410                                 //restore real lastlabel to save below
1411                                 sabund->setLabel(saveLabel);
1412                         }
1413
1414                         lastLabel = sabund->getLabel();
1415
1416                         //prevent memory leak
1417                         delete sabund; sabund = NULL;
1418
1419                         //get next line to process
1420                         sabund = input->getSAbundVector();
1421                 }
1422
1423
1424                 if (m->control_pressed) {  out.close(); return 0;  }
1425
1426                 //output error messages about any remaining user labels
1427                 set<string>::iterator it;
1428                 bool needToRun = false;
1429                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
1430                         m->mothurOut("Your file does not include the label " + *it);
1431                         if (processedLabels.count(lastLabel) != 1) {
1432                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
1433                                 needToRun = true;
1434                         }else {
1435                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
1436                         }
1437                 }
1438
1439                 //run last label if you need to
1440                 if (needToRun == true)  {
1441                         if (sabund != NULL) { delete sabund; }
1442
1443                         sabund = input->getSAbundVector(lastLabel);
1444
1445                         m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
1446
1447                         processSabund(sabund, out);
1448
1449                         delete sabund;
1450                 }
1451
1452                 delete input;
1453                 out.close();
1454
1455                 return 0;
1456
1457         }
1458         catch(exception& e) {
1459                 m->errorOut(e, "SubSampleCommand", "getSubSampleSabund");
1460                 exit(1);
1461         }
1462 }
1463 //**********************************************************************************************************************
1464 int SubSampleCommand::processSabund(SAbundVector*& sabund, ofstream& out) {
1465         try {
1466
1467                 RAbundVector* rabund = new RAbundVector();
1468                 *rabund = sabund->getRAbundVector();
1469
1470                 int numBins = rabund->getNumBins();
1471                 int thisSize = rabund->getNumSeqs();
1472
1473                 if (thisSize != size) {
1474
1475                         OrderVector* order = new OrderVector();
1476                         for(int p=0;p<numBins;p++){
1477                                 for(int j=0;j<rabund->get(p);j++){
1478                                         order->push_back(p);
1479                                 }
1480                         }
1481                         random_shuffle(order->begin(), order->end());
1482
1483                         RAbundVector* temp = new RAbundVector(numBins);
1484                         temp->setLabel(rabund->getLabel());
1485
1486                         delete rabund;
1487                         rabund = temp;
1488
1489                         for (int j = 0; j < size; j++) {
1490
1491                                 if (m->control_pressed) { delete order; return 0; }
1492
1493                                 int bin = order->get(j);
1494
1495                                 int abund = rabund->get(bin);
1496                                 rabund->set(bin, (abund+1));
1497                         }
1498
1499                         delete order;
1500                 }
1501
1502                 if (m->control_pressed) { return 0; }
1503
1504                 delete sabund;
1505                 sabund = new SAbundVector();
1506                 *sabund = rabund->getSAbundVector();
1507                 delete rabund;
1508
1509                 sabund->print(out);
1510
1511                 return 0;
1512
1513         }
1514         catch(exception& e) {
1515                 m->errorOut(e, "SubSampleCommand", "processSabund");
1516                 exit(1);
1517         }
1518 }
1519 //**********************************************************************************************************************
1520 int SubSampleCommand::eliminateZeroOTUS(vector<SharedRAbundVector*>& thislookup) {
1521         try {
1522
1523                 vector<SharedRAbundVector*> newLookup;
1524                 for (int i = 0; i < thislookup.size(); i++) {
1525                         SharedRAbundVector* temp = new SharedRAbundVector();
1526                         temp->setLabel(thislookup[i]->getLabel());
1527                         temp->setGroup(thislookup[i]->getGroup());
1528                         newLookup.push_back(temp);
1529                 }
1530
1531                 //for each bin
1532                 vector<string> newBinLabels;
1533                 string snumBins = toString(thislookup[0]->getNumBins());
1534                 for (int i = 0; i < thislookup[0]->getNumBins(); i++) {
1535                         if (m->control_pressed) { for (int j = 0; j < newLookup.size(); j++) {  delete newLookup[j];  } return 0; }
1536
1537                         //look at each sharedRabund and make sure they are not all zero
1538                         bool allZero = true;
1539                         for (int j = 0; j < thislookup.size(); j++) {
1540                                 if (thislookup[j]->getAbundance(i) != 0) { allZero = false;  break;  }
1541                         }
1542
1543                         //if they are not all zero add this bin
1544                         if (!allZero) {
1545                                 for (int j = 0; j < thislookup.size(); j++) {
1546                                         newLookup[j]->push_back(thislookup[j]->getAbundance(i), thislookup[j]->getGroup());
1547                                 }
1548                                 //if there is a bin label use it otherwise make one
1549                                 string binLabel = "Otu";
1550                                 string sbinNumber = toString(i+1);
1551                                 if (sbinNumber.length() < snumBins.length()) {
1552                                         int diff = snumBins.length() - sbinNumber.length();
1553                                         for (int h = 0; h < diff; h++) { binLabel += "0"; }
1554                                 }
1555                                 binLabel += sbinNumber;
1556                                 if (i < m->currentBinLabels.size()) {  binLabel = m->currentBinLabels[i]; }
1557
1558                                 newBinLabels.push_back(binLabel);
1559                         }
1560                 }
1561
1562                 for (int j = 0; j < thislookup.size(); j++) {  delete thislookup[j];  }
1563                 thislookup.clear();
1564
1565                 thislookup = newLookup;
1566                 m->currentBinLabels = newBinLabels;
1567
1568                 return 0;
1569
1570         }
1571         catch(exception& e) {
1572                 m->errorOut(e, "SubSampleCommand", "eliminateZeroOTUS");
1573                 exit(1);
1574         }
1575 }
1576
1577 //**********************************************************************************************************************
1578
1579
1580