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