subsamplecommand.cpp

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