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","fasta",false,false,true); parameters.push_back(pfasta);
  19         CommandParameter pname("name", "InputTypes", "", "", "NameCount", "none", "none","name",false,false,true); parameters.push_back(pname);
  20         CommandParameter pcount("count", "InputTypes", "", "", "NameCount-CountGroup", "none", "none","count",false,false,true); parameters.push_back(pcount);
  21                 CommandParameter pgroup("group", "InputTypes", "", "", "CountGroup", "none", "none","group",false,false,true); parameters.push_back(pgroup);
  22                 CommandParameter plist("list", "InputTypes", "", "", "none", "FLSSR", "none","list",false,false,true); parameters.push_back(plist);
  23                 CommandParameter pshared("shared", "InputTypes", "", "", "none", "FLSSR", "none","shared",false,false,true); parameters.push_back(pshared);
  24                 CommandParameter prabund("rabund", "InputTypes", "", "", "none", "FLSSR", "none","rabund",false,false); parameters.push_back(prabund);
  25                 CommandParameter psabund("sabund", "InputTypes", "", "", "none", "FLSSR", "none","sabund",false,false); parameters.push_back(psabund);
  26                 CommandParameter plabel("label", "String", "", "", "", "", "","",false,false); parameters.push_back(plabel);
  27                 CommandParameter pgroups("groups", "String", "", "", "", "", "","",false,false); parameters.push_back(pgroups);
  28                 CommandParameter psize("size", "Number", "", "0", "", "", "","",false,false,true); parameters.push_back(psize);
  29                 CommandParameter ppersample("persample", "Boolean", "", "F", "", "", "","",false,false,true); parameters.push_back(ppersample);
  30                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(pinputdir);
  31                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(poutputdir);
  32
  33                 vector<string> myArray;
  34                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  35                 return myArray;
  36         }
  37         catch(exception& e) {
  38                 m->errorOut(e, "SubSampleCommand", "setParameters");
  39                 exit(1);
  40         }
  41 }
  42 //**********************************************************************************************************************
  43 string SubSampleCommand::getHelpString(){
  44         try {
  45                 string helpString = "";
  46                 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";
  47                 helpString += "The sub.sample command parameters are fasta, name, list, group, count, rabund, sabund, shared, groups, size, persample and label.  You must provide a fasta, list, sabund, rabund or shared file as an input file.\n";
  48                 helpString += "The namefile is only used with the fasta file, not with the listfile, because the list file should contain all sequences.\n";
  49                 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";
  50                 helpString += "The label parameter allows you to select what distance levels you would like, and are also separated by dashes.\n";
  51                 helpString += "The size parameter allows you indicate the size of your subsample.\n";
  52                 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";
  53                 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";
  54                 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";
  55                 helpString += "The sub.sample command should be in the following format: sub.sample(list=yourListFile, group=yourGroupFile, groups=yourGroups, label=yourLabels).\n";
  56                 helpString += "Example sub.sample(list=abrecovery.fn.list, group=abrecovery.groups, groups=B-C, size=20).\n";
  57                 helpString += "The default value for groups is all the groups in your groupfile, and all labels in your inputfile will be used.\n";
  58                 helpString += "The sub.sample command outputs a .subsample file.\n";
  59                 helpString += "Note: No spaces between parameter labels (i.e. groups), '=' and parameters (i.e.yourGroups).\n";
  60                 return helpString;
  61         }
  62         catch(exception& e) {
  63                 m->errorOut(e, "SubSampleCommand", "getHelpString");
  64                 exit(1);
  65         }
  66 }
  67 //**********************************************************************************************************************
  68 string SubSampleCommand::getOutputPattern(string type) {
  69     try {
  70         string pattern = "";
  71
  72         if (type == "fasta")            {   pattern = "[filename],subsample,[extension]";    }
  73         else if (type == "sabund")    {   pattern = "[filename],subsample,[extension]";    }
  74         else if (type == "name")        {   pattern = "[filename],subsample,[extension]";    }
  75         else if (type == "group")       {   pattern = "[filename],subsample,[extension]";    }
  76         else if (type == "count")       {   pattern = "[filename],subsample,[extension]";    }
  77         else if (type == "list")        {   pattern = "[filename],subsample,[extension]";    }
  78         else if (type == "shared")      {   pattern = "[filename],[distance],subsample,[extension]";    }
  79         else if (type == "rabund")      {   pattern = "[filename],subsample,[extension]";    }
  80         else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true;  }
  81
  82         return pattern;
  83     }
  84     catch(exception& e) {
  85         m->errorOut(e, "SubSampleCommand", "getOutputPattern");
  86         exit(1);
  87     }
  88 }
  89 //**********************************************************************************************************************
  90 SubSampleCommand::SubSampleCommand(){
  91         try {
  92                 abort = true; calledHelp = true;
  93                 setParameters();
  94                 vector<string> tempOutNames;
  95                 outputTypes["shared"] = tempOutNames;
  96                 outputTypes["list"] = tempOutNames;
  97                 outputTypes["rabund"] = tempOutNames;
  98                 outputTypes["sabund"] = tempOutNames;
  99                 outputTypes["fasta"] = tempOutNames;
 100                 outputTypes["name"] = tempOutNames;
 101                 outputTypes["group"] = tempOutNames;
 102         outputTypes["count"] = tempOutNames;
 103         }
 104         catch(exception& e) {
 105                 m->errorOut(e, "SubSampleCommand", "GetRelAbundCommand");
 106                 exit(1);
 107         }
 108 }
 109 //**********************************************************************************************************************
 110 SubSampleCommand::SubSampleCommand(string option) {
 111         try {
 112                 abort = false; calledHelp = false;
 113                 allLines = 1;
 114
 115                 //allow user to run help
 116                 if(option == "help") { help(); abort = true; calledHelp = true; }
 117                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 118
 119                 else {
 120                         vector<string> myArray = setParameters();
 121
 122                         OptionParser parser(option);
 123                         map<string,string> parameters = parser.getParameters();
 124
 125                         ValidParameters validParameter;
 126
 127                         //check to make sure all parameters are valid for command
 128                         map<string,string>::iterator it;
 129                         for (it = parameters.begin(); it != parameters.end(); it++) {
 130                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 131                         }
 132
 133                         //initialize outputTypes
 134                         vector<string> tempOutNames;
 135                         outputTypes["shared"] = tempOutNames;
 136                         outputTypes["list"] = tempOutNames;
 137                         outputTypes["rabund"] = tempOutNames;
 138                         outputTypes["sabund"] = tempOutNames;
 139                         outputTypes["fasta"] = tempOutNames;
 140                         outputTypes["name"] = tempOutNames;
 141                         outputTypes["group"] = tempOutNames;
 142             outputTypes["count"] = tempOutNames;
 143
 144                         //if the user changes the output directory command factory will send this info to us in the output parameter
 145                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
 146
 147                         //if the user changes the input directory command factory will send this info to us in the output parameter
 148                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 149                         if (inputDir == "not found"){   inputDir = "";          }
 150                         else {
 151                                 string path;
 152                                 it = parameters.find("list");
 153                                 //user has given a template file
 154                                 if(it != parameters.end()){
 155                                         path = m->hasPath(it->second);
 156                                         //if the user has not given a path then, add inputdir. else leave path alone.
 157                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 158                                 }
 159
 160                                 it = parameters.find("fasta");
 161                                 //user has given a template file
 162                                 if(it != parameters.end()){
 163                                         path = m->hasPath(it->second);
 164                                         //if the user has not given a path then, add inputdir. else leave path alone.
 165                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 166                                 }
 167
 168                                 it = parameters.find("shared");
 169                                 //user has given a template file
 170                                 if(it != parameters.end()){
 171                                         path = m->hasPath(it->second);
 172                                         //if the user has not given a path then, add inputdir. else leave path alone.
 173                                         if (path == "") {       parameters["shared"] = inputDir + it->second;           }
 174                                 }
 175
 176                                 it = parameters.find("group");
 177                                 //user has given a template file
 178                                 if(it != parameters.end()){
 179                                         path = m->hasPath(it->second);
 180                                         //if the user has not given a path then, add inputdir. else leave path alone.
 181                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 182                                 }
 183
 184                                 it = parameters.find("sabund");
 185                                 //user has given a template file
 186                                 if(it != parameters.end()){
 187                                         path = m->hasPath(it->second);
 188                                         //if the user has not given a path then, add inputdir. else leave path alone.
 189                                         if (path == "") {       parameters["sabund"] = inputDir + it->second;           }
 190                                 }
 191
 192                                 it = parameters.find("rabund");
 193                                 //user has given a template file
 194                                 if(it != parameters.end()){
 195                                         path = m->hasPath(it->second);
 196                                         //if the user has not given a path then, add inputdir. else leave path alone.
 197                                         if (path == "") {       parameters["rabund"] = inputDir + it->second;           }
 198                                 }
 199
 200                                 it = parameters.find("name");
 201                                 //user has given a template file
 202                                 if(it != parameters.end()){
 203                                         path = m->hasPath(it->second);
 204                                         //if the user has not given a path then, add inputdir. else leave path alone.
 205                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 206                                 }
 207
 208                 it = parameters.find("count");
 209                                 //user has given a template file
 210                                 if(it != parameters.end()){
 211                                         path = m->hasPath(it->second);
 212                                         //if the user has not given a path then, add inputdir. else leave path alone.
 213                                         if (path == "") {       parameters["count"] = inputDir + it->second;            }
 214                                 }
 215                         }
 216
 217                         //check for required parameters
 218                         listfile = validParameter.validFile(parameters, "list", true);
 219                         if (listfile == "not open") { listfile = ""; abort = true; }
 220                         else if (listfile == "not found") { listfile = ""; }
 221                         else { m->setListFile(listfile); }
 222
 223                         sabundfile = validParameter.validFile(parameters, "sabund", true);
 224                         if (sabundfile == "not open") { sabundfile = ""; abort = true; }
 225                         else if (sabundfile == "not found") { sabundfile = ""; }
 226                         else { m->setSabundFile(sabundfile); }
 227
 228                         rabundfile = validParameter.validFile(parameters, "rabund", true);
 229                         if (rabundfile == "not open") { rabundfile = ""; abort = true; }
 230                         else if (rabundfile == "not found") { rabundfile = ""; }
 231                         else { m->setRabundFile(rabundfile); }
 232
 233                         fastafile = validParameter.validFile(parameters, "fasta", true);
 234                         if (fastafile == "not open") { fastafile = ""; abort = true; }
 235                         else if (fastafile == "not found") { fastafile = ""; }
 236                         else { m->setFastaFile(fastafile); }
 237
 238                         sharedfile = validParameter.validFile(parameters, "shared", true);
 239                         if (sharedfile == "not open") { sharedfile = ""; abort = true; }
 240                         else if (sharedfile == "not found") { sharedfile = ""; }
 241                         else { m->setSharedFile(sharedfile); }
 242
 243                         namefile = validParameter.validFile(parameters, "name", true);
 244                         if (namefile == "not open") { namefile = ""; abort = true; }
 245                         else if (namefile == "not found") { namefile = ""; }
 246                         else { m->setNameFile(namefile); }
 247
 248                         groupfile = validParameter.validFile(parameters, "group", true);
 249                         if (groupfile == "not open") { groupfile = ""; abort = true; }
 250                         else if (groupfile == "not found") { groupfile = ""; }
 251                         else { m->setGroupFile(groupfile); }
 252
 253             countfile = validParameter.validFile(parameters, "count", true);
 254                         if (countfile == "not open") { countfile = ""; abort = true; }
 255                         else if (countfile == "not found") { countfile = "";  }
 256                         else {
 257                 m->setCountTableFile(countfile);
 258                 ct.readTable(countfile);
 259             }
 260
 261             if ((namefile != "") && (countfile != "")) {
 262                 m->mothurOut("[ERROR]: you may only use one of the following: name or count."); m->mothurOutEndLine(); abort = true;
 263             }
 264
 265             if ((groupfile != "") && (countfile != "")) {
 266                 m->mothurOut("[ERROR]: you may only use one of the following: group or count."); m->mothurOutEndLine(); abort=true;
 267             }
 268
 269                         //check for optional parameter and set defaults
 270                         // ...at some point should added some additional type checking...
 271                         label = validParameter.validFile(parameters, "label", false);
 272                         if (label == "not found") { label = ""; }
 273                         else {
 274                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 275                                 else { allLines = 1;  }
 276                         }
 277
 278                         groups = validParameter.validFile(parameters, "groups", false);
 279                         if (groups == "not found") { groups = ""; pickedGroups = false; }
 280                         else {
 281                                 pickedGroups = true;
 282                                 m->splitAtDash(groups, Groups);
 283                                 m->setGroups(Groups);
 284                         }
 285
 286                         string temp = validParameter.validFile(parameters, "size", false);              if (temp == "not found"){       temp = "0";             }
 287                         m->mothurConvert(temp, size);
 288
 289                         temp = validParameter.validFile(parameters, "persample", false);                if (temp == "not found"){       temp = "f";             }
 290                         persample = m->isTrue(temp);
 291
 292                         if ((groupfile == "") && (countfile == "")) { persample = false; }
 293             if (countfile != "") {
 294                 if (!ct.hasGroupInfo()) {
 295                     persample = false;
 296                     if (pickedGroups) { m->mothurOut("You cannot pick groups without group info in your count file."); m->mothurOutEndLine(); abort = true; }
 297                 }
 298             }
 299
 300                         if ((namefile != "") && (fastafile == "")) { m->mothurOut("You may only use a namefile with a fastafile."); m->mothurOutEndLine(); abort = true; }
 301
 302                         if ((fastafile == "") && (listfile == "") && (sabundfile == "") && (rabundfile == "") && (sharedfile == "")) {
 303                                 m->mothurOut("You must provide a fasta, list, sabund, rabund or shared file as an input file."); m->mothurOutEndLine(); abort = true; }
 304
 305                         if (pickedGroups && ((groupfile == "") && (sharedfile == "") && (countfile == ""))) {
 306                                 m->mothurOut("You cannot pick groups without a valid group, count or shared file."); m->mothurOutEndLine(); abort = true; }
 307
 308                         if (((groupfile != "") || (countfile != "")) && ((fastafile == "") && (listfile == ""))) {
 309                                 m->mothurOut("Group or count files are only valid with listfile or fastafile."); m->mothurOutEndLine(); abort = true; }
 310
 311                         if (((groupfile != "") || (countfile != "")) && ((fastafile != "") && (listfile != ""))) {
 312                                 m->mothurOut("A new group or count file can only be made from the subsample of a listfile or fastafile, not both. Please correct."); m->mothurOutEndLine(); abort = true; }
 313
 314             if (countfile == "") {
 315                 if ((fastafile != "") && (namefile == "")) {
 316                     vector<string> files; files.push_back(fastafile);
 317                     parser.getNameFile(files);
 318                 }
 319             }
 320                 }
 321
 322         }
 323         catch(exception& e) {
 324                 m->errorOut(e, "SubSampleCommand", "SubSampleCommand");
 325                 exit(1);
 326         }
 327 }
 328 //**********************************************************************************************************************
 329
 330 int SubSampleCommand::execute(){
 331         try {
 332
 333                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 334
 335                 if (sharedfile != "")   {   getSubSampleShared();       }
 336                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]); return 0; } }
 337
 338                 if (listfile != "")             {   getSubSampleList();         }
 339                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]); return 0; } }
 340
 341                 if (rabundfile != "")   {   getSubSampleRabund();       }
 342                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]); return 0; } }
 343
 344                 if (sabundfile != "")   {   getSubSampleSabund();       }
 345                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]); return 0; } }
 346
 347                 if (fastafile != "")    {   getSubSampleFasta();        }
 348                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]); return 0; } }
 349
 350                 //set fasta file as new current fastafile
 351                 string current = "";
 352                 itTypes = outputTypes.find("fasta");
 353                 if (itTypes != outputTypes.end()) {
 354                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 355                 }
 356
 357                 itTypes = outputTypes.find("name");
 358                 if (itTypes != outputTypes.end()) {
 359                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
 360                 }
 361
 362                 itTypes = outputTypes.find("group");
 363                 if (itTypes != outputTypes.end()) {
 364                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setGroupFile(current); }
 365                 }
 366
 367                 itTypes = outputTypes.find("list");
 368                 if (itTypes != outputTypes.end()) {
 369                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 370                 }
 371
 372                 itTypes = outputTypes.find("shared");
 373                 if (itTypes != outputTypes.end()) {
 374                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setSharedFile(current); }
 375                 }
 376
 377                 itTypes = outputTypes.find("rabund");
 378                 if (itTypes != outputTypes.end()) {
 379                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setRabundFile(current); }
 380                 }
 381
 382                 itTypes = outputTypes.find("sabund");
 383                 if (itTypes != outputTypes.end()) {
 384                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setSabundFile(current); }
 385                 }
 386
 387         itTypes = outputTypes.find("count");
 388                 if (itTypes != outputTypes.end()) {
 389                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setCountTableFile(current); }
 390                 }
 391
 392
 393                 m->mothurOutEndLine();
 394                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 395                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 396                 m->mothurOutEndLine();
 397
 398                 return 0;
 399         }
 400         catch(exception& e) {
 401                 m->errorOut(e, "SubSampleCommand", "execute");
 402                 exit(1);
 403         }
 404 }
 405 //**********************************************************************************************************************
 406 int SubSampleCommand::getSubSampleFasta() {
 407         try {
 408
 409                 if (namefile != "") { readNames(); }    //fills names with all names in namefile.
 410                 else { getNames(); }//no name file, so get list of names to pick from
 411
 412                 GroupMap groupMap;
 413                 if (groupfile != "") {
 414                         groupMap.readMap(groupfile);
 415
 416                         //takes care of user setting groupNames that are invalid or setting groups=all
 417                         SharedUtil util;
 418                         vector<string> namesGroups = groupMap.getNamesOfGroups();
 419                         util.setGroups(Groups, namesGroups);
 420
 421                         //file mismatch quit
 422                         if (names.size() != groupMap.getNumSeqs()) {
 423                                 m->mothurOut("[ERROR]: your fasta file contains " + toString(names.size()) + " sequences, and your groupfile contains " + toString(groupMap.getNumSeqs()) + ", please correct.");
 424                                 m->mothurOutEndLine();
 425                                 return 0;
 426                         }
 427                 }else if (countfile != "") {
 428             if (ct.hasGroupInfo()) {
 429                 SharedUtil util;
 430                 vector<string> namesGroups = ct.getNamesOfGroups();
 431                 util.setGroups(Groups, namesGroups);
 432             }
 433
 434             //file mismatch quit
 435                         if (names.size() != ct.getNumUniqueSeqs()) {
 436                                 m->mothurOut("[ERROR]: your fasta file contains " + toString(names.size()) + " sequences, and your count file contains " + toString(ct.getNumUniqueSeqs()) + " unique sequences, please correct.");
 437                                 m->mothurOutEndLine();
 438                                 return 0;
 439                         }
 440         }
 441
 442                 if (m->control_pressed) { return 0; }
 443
 444                 //make sure that if your picked groups size is not too big
 445                 int thisSize = 0;
 446         if (countfile == "") { thisSize = names.size();  }
 447         else {  thisSize = ct. getNumSeqs();  }  //all seqs not just unique
 448
 449                 if (persample) {
 450                         if (size == 0) { //user has not set size, set size = smallest samples size
 451                                 if (countfile == "") { size = groupMap.getNumSeqs(Groups[0]); }
 452                 else {  size = ct.getGroupCount(Groups[0]);  }
 453
 454                                 for (int i = 1; i < Groups.size(); i++) {
 455                                         int thisSize = 0;
 456                     if (countfile == "") { thisSize = groupMap.getNumSeqs(Groups[i]); }
 457                     else {  thisSize = ct.getGroupCount(Groups[i]);  }
 458
 459                                         if (thisSize < size) {  size = thisSize;        }
 460                                 }
 461                         }else { //make sure size is not too large
 462                                 vector<string> newGroups;
 463                                 for (int i = 0; i < Groups.size(); i++) {
 464                                         int thisSize = 0;
 465                     if (countfile == "") { thisSize = groupMap.getNumSeqs(Groups[i]); }
 466                     else {  thisSize = ct.getGroupCount(Groups[i]);  }
 467
 468                                         if (thisSize >= size) { newGroups.push_back(Groups[i]); }
 469                                         else {  m->mothurOut("You have selected a size that is larger than " + Groups[i] + " number of sequences, removing " + Groups[i] + "."); m->mothurOutEndLine(); }
 470                                 }
 471                                 Groups = newGroups;
 472                 if (newGroups.size() == 0) {  m->mothurOut("[ERROR]: all groups removed."); m->mothurOutEndLine(); m->control_pressed = true; }
 473                         }
 474
 475                         m->mothurOut("Sampling " + toString(size) + " from each group."); m->mothurOutEndLine();
 476                 }else {
 477                         if (pickedGroups) {
 478                                 int total = 0;
 479                                 for(int i = 0; i < Groups.size(); i++) {
 480                     if (countfile == "") { total += groupMap.getNumSeqs(Groups[i]); }
 481                     else {  total += ct.getGroupCount(Groups[i]);  }
 482                                 }
 483
 484                                 if (size == 0) { //user has not set size, set size = 10% samples size
 485                                         size = int (total * 0.10);
 486                                 }
 487
 488                                 if (total < size) {
 489                                         if (size != 0) {
 490                                                 m->mothurOut("Your size is too large for the number of groups you selected. Adjusting to " + toString(int (total * 0.10)) + "."); m->mothurOutEndLine();
 491                                         }
 492                                         size = int (total * 0.10);
 493                                 }
 494
 495                                 m->mothurOut("Sampling " + toString(size) + " from " + toString(total) + "."); m->mothurOutEndLine();
 496                         }
 497
 498                         if (size == 0) { //user has not set size, set size = 10% samples size
 499                                 if (countfile == "") {  size = int (names.size() * 0.10); }
 500                 else {  size = int (ct.getNumSeqs() * 0.10); }
 501                         }
 502
 503
 504             if (size > thisSize) { m->mothurOut("Your fasta file only contains " + toString(thisSize) + " sequences. Setting size to " + toString(thisSize) + "."); m->mothurOutEndLine();
 505                     size = thisSize;
 506             }
 507
 508             if (!pickedGroups) { m->mothurOut("Sampling " + toString(size) + " from " + toString(thisSize) + "."); m->mothurOutEndLine(); }
 509
 510                 }
 511                 random_shuffle(names.begin(), names.end());
 512
 513                 set<string> subset; //dont want repeat sequence names added
 514                 if (persample) {
 515             if (countfile == "") {
 516                 //initialize counts
 517                 map<string, int> groupCounts;
 518                 map<string, int>::iterator itGroupCounts;
 519                 for (int i = 0; i < Groups.size(); i++) { groupCounts[Groups[i]] = 0; }
 520
 521                 for (int j = 0; j < names.size(); j++) {
 522
 523                     if (m->control_pressed) { return 0; }
 524
 525                     string group = groupMap.getGroup(names[j]);
 526                     if (group == "not found") { m->mothurOut("[ERROR]: " + names[j] + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
 527                     else{
 528                         itGroupCounts = groupCounts.find(group);
 529                         if (itGroupCounts != groupCounts.end()) {
 530                             if (groupCounts[group] < size) {    subset.insert(names[j]);        groupCounts[group]++; }
 531                         }
 532                     }
 533                 }
 534             }else {
 535                 SubSample sample;
 536                 CountTable sampledCt = sample.getSample(ct, size, Groups);
 537                 vector<string> sampledSeqs = sampledCt.getNamesOfSeqs();
 538                 for (int i = 0; i < sampledSeqs.size(); i++) { subset.insert(sampledSeqs[i]); }
 539
 540                 string countOutputDir = outputDir;
 541                 if (outputDir == "") {  countOutputDir += m->hasPath(countfile);  }
 542                 map<string, string> variables;
 543                 variables["[filename]"] = countOutputDir + m->getRootName(m->getSimpleName(countfile));
 544                 variables["[extension]"] = m->getExtension(countfile);
 545                 string countOutputFileName = getOutputFileName("count", variables);
 546                 outputTypes["count"].push_back(countOutputFileName);  outputNames.push_back(countOutputFileName);
 547                 sampledCt.printTable(countOutputFileName);
 548             }
 549                 }else {
 550                         if (countfile == "") {
 551                 //randomly select a subset of those names to include in the subsample
 552                 //since names was randomly shuffled just grab the next one
 553                 for (int j = 0; j < names.size(); j++) {
 554
 555                     if (m->control_pressed) { return 0; }
 556
 557                     if (groupfile != "") { //if there is a groupfile given fill in group info
 558                         string group = groupMap.getGroup(names[j]);
 559                         if (group == "not found") { m->mothurOut("[ERROR]: " + names[j] + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
 560
 561                         if (pickedGroups) { //if hte user picked groups, we only want to keep the names of sequences from those groups
 562                             if (m->inUsersGroups(group, Groups)) {  subset.insert(names[j]); }
 563                         }else{  subset.insert(names[j]); }
 564                     }else{ //save everyone, group
 565                         subset.insert(names[j]);
 566                     }
 567
 568                     //do we have enough??
 569                     if (subset.size() == size) { break; }
 570                 }
 571             }else {
 572                 SubSample sample;
 573                 CountTable sampledCt = sample.getSample(ct, size, Groups, pickedGroups);
 574                 vector<string> sampledSeqs = sampledCt.getNamesOfSeqs();
 575                 for (int i = 0; i < sampledSeqs.size(); i++) { subset.insert(sampledSeqs[i]); }
 576
 577                 string countOutputDir = outputDir;
 578                 if (outputDir == "") {  countOutputDir += m->hasPath(countfile);  }
 579                 map<string, string> variables;
 580                 variables["[filename]"] = countOutputDir + m->getRootName(m->getSimpleName(countfile));
 581                 variables["[extension]"] = m->getExtension(countfile);
 582                 string countOutputFileName = getOutputFileName("count", variables);
 583                 outputTypes["count"].push_back(countOutputFileName);  outputNames.push_back(countOutputFileName);
 584                 sampledCt.printTable(countOutputFileName);
 585             }
 586                 }
 587
 588                 if (subset.size() == 0) {  m->mothurOut("The size you selected is too large, skipping fasta file."); m->mothurOutEndLine();  return 0; }
 589
 590                 string thisOutputDir = outputDir;
 591                 if (outputDir == "") {  thisOutputDir += m->hasPath(fastafile);  }
 592         map<string, string> variables;
 593         variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(fastafile));
 594         variables["[extension]"] = m->getExtension(fastafile);
 595                 string outputFileName = getOutputFileName("fasta", variables);
 596                 ofstream out;
 597                 m->openOutputFile(outputFileName, out);
 598
 599                 //read through fasta file outputting only the names on the subsample list
 600                 ifstream in;
 601                 m->openInputFile(fastafile, in);
 602
 603                 string thisname;
 604                 int count = 0;
 605                 map<string, vector<string> >::iterator itNameMap;
 606
 607                 while(!in.eof()){
 608
 609                         if (m->control_pressed) { in.close(); out.close();  return 0; }
 610
 611                         Sequence currSeq(in);
 612                         thisname = currSeq.getName();
 613
 614                         if (thisname != "") {
 615
 616                                 //does the subset contain a sequence that this sequence represents
 617                                 itNameMap = nameMap.find(thisname);
 618                                 if (itNameMap != nameMap.end()) {
 619                                         vector<string> nameRepresents = itNameMap->second;
 620
 621                                         for (int i = 0; i < nameRepresents.size(); i++){
 622                                                 if (subset.count(nameRepresents[i]) != 0) {
 623                                                         out << ">" << nameRepresents[i] << endl << currSeq.getAligned() << endl;
 624                                                         count++;
 625                                                 }
 626                                         }
 627                                 }else{
 628                                         m->mothurOut("[ERROR]: " + thisname + " is not in your namefile, please correct."); m->mothurOutEndLine();
 629                                 }
 630                         }
 631                         m->gobble(in);
 632                 }
 633                 in.close();
 634                 out.close();
 635
 636                 if (count != subset.size()) {
 637                         m->mothurOut("[ERROR]: The subset selected contained " + toString(subset.size()) + " sequences, but I only found " + toString(count) + " of those in the fastafile."); m->mothurOutEndLine();
 638                 }
 639
 640                 if (namefile != "") {
 641                         m->mothurOut("Deconvoluting subsampled fasta file... "); m->mothurOutEndLine();
 642                         map<string, string> variables;
 643             variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(namefile));
 644             variables["[extension]"] = m->getExtension(namefile);
 645             string outputNameFileName = getOutputFileName("name", variables);
 646                         //use unique.seqs to create new name and fastafile
 647                         string inputString = "fasta=" + outputFileName;
 648                         m->mothurOut("/******************************************/"); m->mothurOutEndLine();
 649                         m->mothurOut("Running command: unique.seqs(" + inputString + ")"); m->mothurOutEndLine();
 650                         m->mothurCalling = true;
 651
 652                         Command* uniqueCommand = new DeconvoluteCommand(inputString);
 653                         uniqueCommand->execute();
 654
 655                         map<string, vector<string> > filenames = uniqueCommand->getOutputFiles();
 656
 657                         delete uniqueCommand;
 658                         m->mothurCalling = false;
 659
 660             m->renameFile(filenames["name"][0], outputNameFileName);
 661             m->renameFile(filenames["fasta"][0], outputFileName);
 662
 663                         outputTypes["name"].push_back(outputNameFileName);  outputNames.push_back(outputNameFileName);
 664
 665                         m->mothurOut("/******************************************/"); m->mothurOutEndLine();
 666
 667                         m->mothurOut("Done."); m->mothurOutEndLine();
 668                 }
 669
 670                 outputTypes["fasta"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 671
 672                 //if a groupfile is provided read through the group file only outputting the names on the subsample list
 673                 if (groupfile != "") {
 674
 675                         string groupOutputDir = outputDir;
 676                         if (outputDir == "") {  groupOutputDir += m->hasPath(groupfile);  }
 677             map<string, string> variables;
 678             variables["[filename]"] = groupOutputDir + m->getRootName(m->getSimpleName(groupfile));
 679             variables["[extension]"] = m->getExtension(groupfile);
 680                         string groupOutputFileName = getOutputFileName("group", variables);
 681
 682                         ofstream outGroup;
 683                         m->openOutputFile(groupOutputFileName, outGroup);
 684                         outputTypes["group"].push_back(groupOutputFileName);  outputNames.push_back(groupOutputFileName);
 685
 686                         ifstream inGroup;
 687                         m->openInputFile(groupfile, inGroup);
 688                         string name, group;
 689
 690                         while(!inGroup.eof()){
 691
 692                                 if (m->control_pressed) { inGroup.close(); outGroup.close(); return 0; }
 693
 694                                 inGroup >> name;        m->gobble(inGroup);                     //read from first column
 695                                 inGroup >> group;                       //read from second column
 696
 697                                 //if this name is in the accnos file
 698                                 if (subset.count(name) != 0) {
 699                                         outGroup << name << '\t' << group << endl;
 700                                         subset.erase(name);
 701                                 }
 702
 703                                 m->gobble(inGroup);
 704                         }
 705                         inGroup.close();
 706                         outGroup.close();
 707
 708                         //sanity check
 709                         if (subset.size() != 0) {
 710                                 m->mothurOut("Your groupfile does not match your fasta file."); m->mothurOutEndLine();
 711                                 for (set<string>::iterator it = subset.begin(); it != subset.end(); it++) {
 712                                         m->mothurOut("[ERROR]: " + *it + " is missing from your groupfile."); m->mothurOutEndLine();
 713                                 }
 714                         }
 715                 }
 716
 717                 return 0;
 718
 719         }
 720         catch(exception& e) {
 721                 m->errorOut(e, "SubSampleCommand", "getSubSampleFasta");
 722                 exit(1);
 723         }
 724 }
 725 //**********************************************************************************************************************
 726 int SubSampleCommand::getNames() {
 727         try {
 728
 729                 ifstream in;
 730                 m->openInputFile(fastafile, in);
 731
 732                 string thisname;
 733                 while(!in.eof()){
 734
 735                         if (m->control_pressed) { in.close(); return 0; }
 736
 737                         Sequence currSeq(in);
 738                         thisname = currSeq.getName();
 739
 740                         if (thisname != "") {
 741                                 vector<string> temp; temp.push_back(thisname);
 742                                 nameMap[thisname] = temp;
 743                                 names.push_back(thisname);
 744                         }
 745                         m->gobble(in);
 746                 }
 747                 in.close();
 748
 749                 return 0;
 750
 751         }
 752         catch(exception& e) {
 753                 m->errorOut(e, "SubSampleCommand", "getNames");
 754                 exit(1);
 755         }
 756 }
 757 //**********************************************************************************************************************
 758 int SubSampleCommand::readNames() {
 759         try {
 760
 761         nameMap.clear();
 762         m->readNames(namefile, nameMap);
 763
 764         //save names of all sequences
 765         map<string, vector<string> >::iterator it;
 766         for (it = nameMap.begin(); it != nameMap.end(); it++) { for (int i = 0; i < (it->second).size(); i++) { names.push_back((it->second)[i]); } }
 767
 768                 return 0;
 769
 770         }
 771         catch(exception& e) {
 772                 m->errorOut(e, "SubSampleCommand", "readNames");
 773                 exit(1);
 774         }
 775 }
 776 //**********************************************************************************************************************
 777 int SubSampleCommand::getSubSampleShared() {
 778         try {
 779
 780                 InputData* input = new InputData(sharedfile, "sharedfile");
 781                 vector<SharedRAbundVector*> lookup = input->getSharedRAbundVectors();
 782                 string lastLabel = lookup[0]->getLabel();
 783
 784                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 785                 set<string> processedLabels;
 786                 set<string> userLabels = labels;
 787
 788                 if (size == 0) { //user has not set size, set size = smallest samples size
 789                         size = lookup[0]->getNumSeqs();
 790                         for (int i = 1; i < lookup.size(); i++) {
 791                                 int thisSize = lookup[i]->getNumSeqs();
 792
 793                                 if (thisSize < size) {  size = thisSize;        }
 794                         }
 795                 }else {
 796                         m->clearGroups();
 797                         Groups.clear();
 798                         vector<SharedRAbundVector*> temp;
 799                         for (int i = 0; i < lookup.size(); i++) {
 800                                 if (lookup[i]->getNumSeqs() < size) {
 801                                         m->mothurOut(lookup[i]->getGroup() + " contains " + toString(lookup[i]->getNumSeqs()) + ". Eliminating."); m->mothurOutEndLine();
 802                                         delete lookup[i];
 803                                 }else {
 804                                         Groups.push_back(lookup[i]->getGroup());
 805                                         temp.push_back(lookup[i]);
 806                                 }
 807                         }
 808                         lookup = temp;
 809                         m->setGroups(Groups);
 810                 }
 811
 812                 if (lookup.size() == 0) {  m->mothurOut("The size you selected is too large, skipping shared file."); m->mothurOutEndLine(); delete input; return 0; }
 813
 814                 m->mothurOut("Sampling " + toString(size) + " from each group."); m->mothurOutEndLine();
 815
 816                 //as long as you are not at the end of the file or done wih the lines you want
 817                 while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 818                         if (m->control_pressed) {  delete input; for (int i = 0; i < lookup.size(); i++) {  delete lookup[i]; lookup[i] = NULL; }  return 0;  }
 819
 820                         if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){
 821
 822                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 823
 824                                 processShared(lookup);
 825
 826                                 processedLabels.insert(lookup[0]->getLabel());
 827                                 userLabels.erase(lookup[0]->getLabel());
 828                         }
 829
 830                         if ((m->anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 831                                 string saveLabel = lookup[0]->getLabel();
 832
 833                                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 834
 835                                 lookup = input->getSharedRAbundVectors(lastLabel);
 836                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 837
 838                                 processShared(lookup);
 839
 840                                 processedLabels.insert(lookup[0]->getLabel());
 841                                 userLabels.erase(lookup[0]->getLabel());
 842
 843                                 //restore real lastlabel to save below
 844                                 lookup[0]->setLabel(saveLabel);
 845                         }
 846
 847                         lastLabel = lookup[0]->getLabel();
 848                         //prevent memory leak
 849                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i]; lookup[i] = NULL; }
 850
 851                         //get next line to process
 852                         lookup = input->getSharedRAbundVectors();
 853                 }
 854
 855
 856                 if (m->control_pressed) {   return 0;  }
 857
 858                 //output error messages about any remaining user labels
 859                 set<string>::iterator it;
 860                 bool needToRun = false;
 861                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 862                         m->mothurOut("Your file does not include the label " + *it);
 863                         if (processedLabels.count(lastLabel) != 1) {
 864                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 865                                 needToRun = true;
 866                         }else {
 867                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 868                         }
 869                 }
 870
 871                 //run last label if you need to
 872                 if (needToRun == true)  {
 873                         for (int i = 0; i < lookup.size(); i++) { if (lookup[i] != NULL) { delete lookup[i]; } }
 874                         lookup = input->getSharedRAbundVectors(lastLabel);
 875
 876                         m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 877
 878                         processShared(lookup);
 879
 880                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 881                 }
 882
 883                 delete input;
 884
 885                 return 0;
 886
 887         }
 888         catch(exception& e) {
 889                 m->errorOut(e, "SubSampleCommand", "getSubSampleShared");
 890                 exit(1);
 891         }
 892 }
 893 //**********************************************************************************************************************
 894 int SubSampleCommand::processShared(vector<SharedRAbundVector*>& thislookup) {
 895         try {
 896
 897                 //save mothurOut's binLabels to restore for next label
 898                 vector<string> saveBinLabels = m->currentBinLabels;
 899
 900                 string thisOutputDir = outputDir;
 901                 if (outputDir == "") {  thisOutputDir += m->hasPath(sharedfile);  }
 902         map<string, string> variables;
 903         variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(sharedfile));
 904         variables["[extension]"] = m->getExtension(sharedfile);
 905         variables["[distance]"] = thislookup[0]->getLabel();
 906                 string outputFileName = getOutputFileName("shared", variables);
 907         SubSample sample;
 908         vector<string> subsampledLabels = sample.getSample(thislookup, size);
 909
 910         if (m->control_pressed) {  return 0; }
 911
 912         ofstream out;
 913                 m->openOutputFile(outputFileName, out);
 914                 outputTypes["shared"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 915
 916         m->currentBinLabels = subsampledLabels;
 917
 918                 thislookup[0]->printHeaders(out);
 919
 920                 for (int i = 0; i < thislookup.size(); i++) {
 921                         out << thislookup[i]->getLabel() << '\t' << thislookup[i]->getGroup() << '\t';
 922                         thislookup[i]->print(out);
 923                 }
 924                 out.close();
 925
 926
 927         //save mothurOut's binLabels to restore for next label
 928                 m->currentBinLabels = saveBinLabels;
 929
 930                 return 0;
 931
 932         }
 933         catch(exception& e) {
 934                 m->errorOut(e, "SubSampleCommand", "processShared");
 935                 exit(1);
 936         }
 937 }
 938 //**********************************************************************************************************************
 939 int SubSampleCommand::getSubSampleList() {
 940         try {
 941
 942                 string thisOutputDir = outputDir;
 943                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 944                 map<string, string> variables;
 945         variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(listfile));
 946         variables["[extension]"] = m->getExtension(listfile);
 947                 string outputFileName = getOutputFileName("list", variables);
 948                 ofstream out;
 949                 m->openOutputFile(outputFileName, out);
 950                 outputTypes["list"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 951
 952                 InputData* input = new InputData(listfile, "list");
 953                 ListVector* list = input->getListVector();
 954                 string lastLabel = list->getLabel();
 955
 956                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 957                 set<string> processedLabels;
 958                 set<string> userLabels = labels;
 959
 960                 ofstream outGroup;
 961                 GroupMap groupMap;
 962                 if (groupfile != "") {
 963                         groupMap.readMap(groupfile);
 964
 965                         //takes care of user setting groupNames that are invalid or setting groups=all
 966                         SharedUtil util; vector<string> namesGroups = groupMap.getNamesOfGroups(); util.setGroups(Groups, namesGroups);
 967
 968                         //create outputfiles
 969                         string groupOutputDir = outputDir;
 970                         if (outputDir == "") {  groupOutputDir += m->hasPath(groupfile);  }
 971                         string groupOutputFileName = groupOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "subsample" + m->getExtension(groupfile);
 972                         m->openOutputFile(groupOutputFileName, outGroup);
 973                         outputTypes["group"].push_back(groupOutputFileName);  outputNames.push_back(groupOutputFileName);
 974
 975                         //file mismatch quit
 976                         if (list->getNumSeqs() != groupMap.getNumSeqs()) {
 977                                 m->mothurOut("[ERROR]: your list file contains " + toString(list->getNumSeqs()) + " sequences, and your groupfile contains " + toString(groupMap.getNumSeqs()) + ", please correct.");
 978                                 m->mothurOutEndLine(); delete list; delete input; out.close(); outGroup.close(); return 0;
 979                         }
 980                 }else if (countfile != "") {
 981             if (ct.hasGroupInfo()) {
 982                 SharedUtil util;
 983                 vector<string> namesGroups = ct.getNamesOfGroups();
 984                 util.setGroups(Groups, namesGroups);
 985             }
 986
 987             //file mismatch quit
 988                         if (list->getNumSeqs() != ct.getNumUniqueSeqs()) {
 989                                 m->mothurOut("[ERROR]: your list file contains " + toString(list->getNumSeqs()) + " sequences, and your count file contains " + toString(ct.getNumUniqueSeqs()) + " unique sequences, please correct.");
 990                                 m->mothurOutEndLine();
 991                                 return 0;
 992                         }
 993         }
 994
 995                 //make sure that if your picked groups size is not too big
 996                 if (persample) {
 997                         if (size == 0) { //user has not set size, set size = smallest samples size
 998                                 if (countfile == "") { size = groupMap.getNumSeqs(Groups[0]); }
 999                 else {  size = ct.getGroupCount(Groups[0]);  }
1000
1001                                 for (int i = 1; i < Groups.size(); i++) {
1002                                         int thisSize = 0;
1003                     if (countfile == "") { thisSize = groupMap.getNumSeqs(Groups[i]); }
1004                     else {  thisSize = ct.getGroupCount(Groups[i]);  }
1005
1006                                         if (thisSize < size) {  size = thisSize;        }
1007                                 }
1008                         }else { //make sure size is not too large
1009                                 vector<string> newGroups;
1010                                 for (int i = 0; i < Groups.size(); i++) {
1011                                         int thisSize = 0;
1012                     if (countfile == "") { thisSize = groupMap.getNumSeqs(Groups[i]); }
1013                     else {  thisSize = ct.getGroupCount(Groups[i]);  }
1014
1015                                         if (thisSize >= size) { newGroups.push_back(Groups[i]); }
1016                                         else {  m->mothurOut("You have selected a size that is larger than " + Groups[i] + " number of sequences, removing " + Groups[i] + "."); m->mothurOutEndLine(); }
1017                                 }
1018                                 Groups = newGroups;
1019                 if (newGroups.size() == 0) {  m->mothurOut("[ERROR]: all groups removed."); m->mothurOutEndLine(); m->control_pressed = true; }
1020                         }
1021
1022                         m->mothurOut("Sampling " + toString(size) + " from each group."); m->mothurOutEndLine();
1023                 }else{
1024             if (pickedGroups) {
1025                                 int total = 0;
1026                                 for(int i = 0; i < Groups.size(); i++) {
1027                     if (countfile == "") { total += groupMap.getNumSeqs(Groups[i]); }
1028                     else {  total += ct.getGroupCount(Groups[i]);  }
1029                                 }
1030
1031                                 if (size == 0) { //user has not set size, set size = 10% samples size
1032                                         size = int (total * 0.10);
1033                                 }
1034
1035                                 if (total < size) {
1036                                         if (size != 0) {
1037                                                 m->mothurOut("Your size is too large for the number of groups you selected. Adjusting to " + toString(int (total * 0.10)) + "."); m->mothurOutEndLine();
1038                                         }
1039                                         size = int (total * 0.10);
1040                                 }
1041
1042                                 m->mothurOut("Sampling " + toString(size) + " from " + toString(total) + "."); m->mothurOutEndLine();
1043                         }else {
1044                 if (size == 0) { //user has not set size, set size = 10% samples size
1045                                         if (countfile == "") {  size = int (list->getNumSeqs() * 0.10);  }
1046                     else { size = int (ct.getNumSeqs() * 0.10);  }
1047                                 }
1048
1049                                 int thisSize = 0;
1050                 if (countfile == "") { thisSize = list->getNumSeqs();  }
1051                 else { thisSize = ct.getNumSeqs(); }
1052
1053                                 if (size > thisSize) { m->mothurOut("Your list file only contains " + toString(thisSize) + " sequences. Setting size to " + toString(thisSize) + "."); m->mothurOutEndLine();
1054                                         size = thisSize;
1055                                 }
1056
1057                                 m->mothurOut("Sampling " + toString(size) + " from " + toString(thisSize) + "."); m->mothurOutEndLine();
1058             }
1059         }
1060
1061         set<string> subset; //dont want repeat sequence names added
1062                 if (countfile == "") {
1063             //fill names
1064             for (int i = 0; i < list->getNumBins(); i++) {
1065                 string binnames = list->get(i);
1066                 vector<string> thisBin;
1067                 m->splitAtComma(binnames, thisBin);
1068
1069                 for(int j=0;j<thisBin.size();j++){
1070                     if (groupfile != "") { //if there is a groupfile given fill in group info
1071                         string group = groupMap.getGroup(thisBin[j]);
1072                         if (group == "not found") { m->mothurOut("[ERROR]: " + thisBin[j] + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
1073
1074                                                 //if hte user picked groups, we only want to keep the names of sequences from those groups
1075                                                 if (pickedGroups) { if (m->inUsersGroups(group, Groups)) { names.push_back(thisBin[j]); }  }
1076                                                 else{ names.push_back(thisBin[j]); }
1077                     }//save everyone, group
1078                     else{ names.push_back(thisBin[j]); }
1079                 }
1080             }
1081
1082             random_shuffle(names.begin(), names.end());
1083
1084             //randomly select a subset of those names to include in the subsample
1085             if (persample) {
1086                 //initialize counts
1087                 map<string, int> groupCounts;
1088                 map<string, int>::iterator itGroupCounts;
1089                 for (int i = 0; i < Groups.size(); i++) { groupCounts[Groups[i]] = 0; }
1090
1091                 for (int j = 0; j < names.size(); j++) {
1092
1093                     if (m->control_pressed) { delete list; delete input;  return 0; }
1094
1095                     string group = groupMap.getGroup(names[j]);
1096                     if (group == "not found") { m->mothurOut("[ERROR]: " + names[j] + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
1097                     else{
1098                         itGroupCounts = groupCounts.find(group);
1099                         if (itGroupCounts != groupCounts.end()) {
1100                             if (groupCounts[group] < size) {    subset.insert(names[j]);        groupCounts[group]++; }
1101                         }
1102                     }
1103                 }
1104             }else{
1105                 for (int j = 0; j < size; j++) {
1106                     if (m->control_pressed) { break; }
1107                     subset.insert(names[j]);
1108                 }
1109             }
1110
1111             if (groupfile != "") {
1112                 //write out new groupfile
1113                 for (set<string>::iterator it = subset.begin(); it != subset.end(); it++) {
1114                     string group = groupMap.getGroup(*it);
1115                     if (group == "not found") { group = "NOTFOUND"; }
1116                     outGroup << *it << '\t' << group << endl;
1117                 }
1118                 outGroup.close();
1119             }
1120                 }else {
1121             SubSample sample; CountTable sampledCt;
1122
1123             if (persample)  { sampledCt = sample.getSample(ct, size, Groups);               }
1124             else            { sampledCt = sample.getSample(ct, size, Groups, pickedGroups); }
1125
1126             vector<string> sampledSeqs = sampledCt.getNamesOfSeqs();
1127             for (int i = 0; i < sampledSeqs.size(); i++) { subset.insert(sampledSeqs[i]); }
1128
1129             string countOutputDir = outputDir;
1130             if (outputDir == "") {  countOutputDir += m->hasPath(countfile);  }
1131             map<string, string> variables;
1132             variables["[filename]"] = countOutputDir + m->getRootName(m->getSimpleName(countfile));
1133             variables["[extension]"] = m->getExtension(countfile);
1134             string countOutputFileName = getOutputFileName("count", variables);
1135             outputTypes["count"].push_back(countOutputFileName);  outputNames.push_back(countOutputFileName);
1136             sampledCt.printTable(countOutputFileName);
1137         }
1138
1139                 //as long as you are not at the end of the file or done wih the lines you want
1140                 while((list != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
1141
1142                         if (m->control_pressed) {  delete list; delete input;  out.close();  return 0;  }
1143
1144                         if(allLines == 1 || labels.count(list->getLabel()) == 1){
1145
1146                                 m->mothurOut(list->getLabel()); m->mothurOutEndLine();
1147
1148                                 processList(list, out, subset);
1149
1150                                 processedLabels.insert(list->getLabel());
1151                                 userLabels.erase(list->getLabel());
1152                         }
1153
1154                         if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
1155                                 string saveLabel = list->getLabel();
1156
1157                                 delete list;
1158
1159                                 list = input->getListVector(lastLabel);
1160                                 m->mothurOut(list->getLabel()); m->mothurOutEndLine();
1161
1162                                 processList(list, out, subset);
1163
1164                                 processedLabels.insert(list->getLabel());
1165                                 userLabels.erase(list->getLabel());
1166
1167                                 //restore real lastlabel to save below
1168                                 list->setLabel(saveLabel);
1169                         }
1170
1171                         lastLabel = list->getLabel();
1172
1173                         delete list; list = NULL;
1174
1175                         //get next line to process
1176                         list = input->getListVector();
1177                 }
1178
1179
1180                 if (m->control_pressed) {  if (list != NULL) { delete list; } delete input; out.close(); return 0;  }
1181
1182                 //output error messages about any remaining user labels
1183                 set<string>::iterator it;
1184                 bool needToRun = false;
1185                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
1186                         m->mothurOut("Your file does not include the label " + *it);
1187                         if (processedLabels.count(lastLabel) != 1) {
1188                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
1189                                 needToRun = true;
1190                         }else {
1191                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
1192                         }
1193                 }
1194
1195                 //run last label if you need to
1196                 if (needToRun == true)  {
1197                         if (list != NULL) { delete list; }
1198
1199                         list = input->getListVector(lastLabel);
1200
1201                         m->mothurOut(list->getLabel()); m->mothurOutEndLine();
1202
1203                         processList(list, out, subset);
1204
1205                         delete list; list = NULL;
1206                 }
1207
1208                 out.close();
1209                 if (list != NULL) { delete list; }
1210                 delete input;
1211
1212                 return 0;
1213
1214         }
1215         catch(exception& e) {
1216                 m->errorOut(e, "SubSampleCommand", "getSubSampleList");
1217                 exit(1);
1218         }
1219 }
1220 //**********************************************************************************************************************
1221 int SubSampleCommand::processList(ListVector*& list, ofstream& out, set<string>& subset) {
1222         try {
1223
1224                 int numBins = list->getNumBins();
1225
1226                 ListVector* temp = new ListVector();
1227                 temp->setLabel(list->getLabel());
1228
1229                 for (int i = 0; i < numBins; i++) {
1230
1231                         if (m->control_pressed) { break; }
1232
1233                         string bin = list->get(i);
1234             vector<string> binnames;
1235             m->splitAtComma(bin, binnames);
1236
1237             string newNames = "";
1238                         for(int j=0;j<binnames.size();j++){ if (subset.count(binnames[j]) != 0) {  newNames += binnames[j] + ",";  } }
1239
1240                         //if there are names in this bin add to new list
1241                         if (newNames != "") {
1242                                 newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
1243                                 temp->push_back(newNames);
1244                         }
1245                 }
1246
1247                 delete list;
1248                 list = temp;
1249
1250                 if (m->control_pressed) { return 0; }
1251
1252                 list->print(out);
1253
1254                 return 0;
1255
1256         }
1257         catch(exception& e) {
1258                 m->errorOut(e, "SubSampleCommand", "processList");
1259                 exit(1);
1260         }
1261 }
1262 //**********************************************************************************************************************
1263 int SubSampleCommand::getSubSampleRabund() {
1264         try {
1265                 InputData* input = new InputData(rabundfile, "rabund");
1266                 RAbundVector* rabund = input->getRAbundVector();
1267                 string lastLabel = rabund->getLabel();
1268
1269                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
1270                 set<string> processedLabels;
1271                 set<string> userLabels = labels;
1272
1273                 if (size == 0) { //user has not set size, set size = 10%
1274                         size = int((rabund->getNumSeqs()) * 0.10);
1275                 }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; }
1276
1277                 m->mothurOut("Sampling " + toString(size) + " from " + toString(rabund->getNumSeqs()) + "."); m->mothurOutEndLine();
1278
1279                 string thisOutputDir = outputDir;
1280                 if (outputDir == "") {  thisOutputDir += m->hasPath(rabundfile);  }
1281         map<string, string> variables;
1282                 variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(rabundfile));
1283         variables["[extension]"] = m->getExtension(rabundfile);
1284                 string outputFileName = getOutputFileName("rabund", variables);
1285                 ofstream out;
1286                 m->openOutputFile(outputFileName, out);
1287                 outputTypes["rabund"].push_back(outputFileName);  outputNames.push_back(outputFileName);
1288
1289                 //as long as you are not at the end of the file or done wih the lines you want
1290                 while((rabund != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
1291                         if (m->control_pressed) {  delete input; delete rabund; out.close(); return 0;  }
1292
1293                         if(allLines == 1 || labels.count(rabund->getLabel()) == 1){
1294
1295                                 m->mothurOut(rabund->getLabel()); m->mothurOutEndLine();
1296
1297                                 processRabund(rabund, out);
1298
1299                                 processedLabels.insert(rabund->getLabel());
1300                                 userLabels.erase(rabund->getLabel());
1301                         }
1302
1303                         if ((m->anyLabelsToProcess(rabund->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
1304                                 string saveLabel = rabund->getLabel();
1305
1306                                 delete rabund;
1307
1308                                 rabund = input->getRAbundVector(lastLabel);
1309                                 m->mothurOut(rabund->getLabel()); m->mothurOutEndLine();
1310
1311                                 processRabund(rabund, out);
1312
1313                                 processedLabels.insert(rabund->getLabel());
1314                                 userLabels.erase(rabund->getLabel());
1315
1316                                 //restore real lastlabel to save below
1317                                 rabund->setLabel(saveLabel);
1318                         }
1319
1320                         lastLabel = rabund->getLabel();
1321
1322                         //prevent memory leak
1323                         delete rabund; rabund = NULL;
1324
1325                         //get next line to process
1326                         rabund = input->getRAbundVector();
1327                 }
1328
1329
1330                 if (m->control_pressed) {  out.close(); return 0;  }
1331
1332                 //output error messages about any remaining user labels
1333                 set<string>::iterator it;
1334                 bool needToRun = false;
1335                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
1336                         m->mothurOut("Your file does not include the label " + *it);
1337                         if (processedLabels.count(lastLabel) != 1) {
1338                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
1339                                 needToRun = true;
1340                         }else {
1341                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
1342                         }
1343                 }
1344
1345                 //run last label if you need to
1346                 if (needToRun == true)  {
1347                         if (rabund != NULL) { delete rabund; }
1348
1349                         rabund = input->getRAbundVector(lastLabel);
1350
1351                         m->mothurOut(rabund->getLabel()); m->mothurOutEndLine();
1352
1353                         processRabund(rabund, out);
1354
1355                         delete rabund;
1356                 }
1357
1358                 delete input;
1359                 out.close();
1360
1361                 return 0;
1362
1363         }
1364         catch(exception& e) {
1365                 m->errorOut(e, "SubSampleCommand", "getSubSampleRabund");
1366                 exit(1);
1367         }
1368 }
1369 //**********************************************************************************************************************
1370 int SubSampleCommand::processRabund(RAbundVector*& rabund, ofstream& out) {
1371         try {
1372
1373                 int numBins = rabund->getNumBins();
1374                 int thisSize = rabund->getNumSeqs();
1375
1376                 if (thisSize != size) {
1377
1378                         OrderVector* order = new OrderVector();
1379                         for(int p=0;p<numBins;p++){
1380                                 for(int j=0;j<rabund->get(p);j++){
1381                                         order->push_back(p);
1382                                 }
1383                         }
1384                         random_shuffle(order->begin(), order->end());
1385
1386                         RAbundVector* temp = new RAbundVector(numBins);
1387                         temp->setLabel(rabund->getLabel());
1388
1389                         delete rabund;
1390                         rabund = temp;
1391
1392                         for (int j = 0; j < size; j++) {
1393
1394                                 if (m->control_pressed) { delete order; return 0; }
1395
1396                                 int bin = order->get(j);
1397
1398                                 int abund = rabund->get(bin);
1399                                 rabund->set(bin, (abund+1));
1400                         }
1401
1402                         delete order;
1403                 }
1404
1405                 if (m->control_pressed) { return 0; }
1406
1407                 rabund->print(out);
1408
1409                 return 0;
1410
1411         }
1412         catch(exception& e) {
1413                 m->errorOut(e, "SubSampleCommand", "processRabund");
1414                 exit(1);
1415         }
1416 }
1417 //**********************************************************************************************************************
1418 int SubSampleCommand::getSubSampleSabund() {
1419         try {
1420
1421                 InputData* input = new InputData(sabundfile, "sabund");
1422                 SAbundVector* sabund = input->getSAbundVector();
1423                 string lastLabel = sabund->getLabel();
1424
1425                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
1426                 set<string> processedLabels;
1427                 set<string> userLabels = labels;
1428
1429                 if (size == 0) { //user has not set size, set size = 10%
1430                         size = int((sabund->getNumSeqs()) * 0.10);
1431                 }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; }
1432
1433
1434                 m->mothurOut("Sampling " + toString(size) + " from " + toString(sabund->getNumSeqs()) + "."); m->mothurOutEndLine();
1435
1436                 string thisOutputDir = outputDir;
1437                 if (outputDir == "") {  thisOutputDir += m->hasPath(sabundfile);  }
1438         map<string, string> variables;
1439                 variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(sabundfile));
1440         variables["[extension]"] = m->getExtension(sabundfile);
1441                 string outputFileName = getOutputFileName("sabund", variables);
1442                 ofstream out;
1443                 m->openOutputFile(outputFileName, out);
1444                 outputTypes["sabund"].push_back(outputFileName);  outputNames.push_back(outputFileName);
1445
1446
1447                 //as long as you are not at the end of the file or done wih the lines you want
1448                 while((sabund != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
1449                         if (m->control_pressed) {  delete input; delete sabund; out.close(); return 0;  }
1450
1451                         if(allLines == 1 || labels.count(sabund->getLabel()) == 1){
1452
1453                                 m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
1454
1455                                 processSabund(sabund, out);
1456
1457                                 processedLabels.insert(sabund->getLabel());
1458                                 userLabels.erase(sabund->getLabel());
1459                         }
1460
1461                         if ((m->anyLabelsToProcess(sabund->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
1462                                 string saveLabel = sabund->getLabel();
1463
1464                                 delete sabund;
1465
1466                                 sabund = input->getSAbundVector(lastLabel);
1467                                 m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
1468
1469                                 processSabund(sabund, out);
1470
1471                                 processedLabels.insert(sabund->getLabel());
1472                                 userLabels.erase(sabund->getLabel());
1473
1474                                 //restore real lastlabel to save below
1475                                 sabund->setLabel(saveLabel);
1476                         }
1477
1478                         lastLabel = sabund->getLabel();
1479
1480                         //prevent memory leak
1481                         delete sabund; sabund = NULL;
1482
1483                         //get next line to process
1484                         sabund = input->getSAbundVector();
1485                 }
1486
1487
1488                 if (m->control_pressed) {  out.close(); return 0;  }
1489
1490                 //output error messages about any remaining user labels
1491                 set<string>::iterator it;
1492                 bool needToRun = false;
1493                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
1494                         m->mothurOut("Your file does not include the label " + *it);
1495                         if (processedLabels.count(lastLabel) != 1) {
1496                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
1497                                 needToRun = true;
1498                         }else {
1499                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
1500                         }
1501                 }
1502
1503                 //run last label if you need to
1504                 if (needToRun == true)  {
1505                         if (sabund != NULL) { delete sabund; }
1506
1507                         sabund = input->getSAbundVector(lastLabel);
1508
1509                         m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
1510
1511                         processSabund(sabund, out);
1512
1513                         delete sabund;
1514                 }
1515
1516                 delete input;
1517                 out.close();
1518
1519                 return 0;
1520
1521         }
1522         catch(exception& e) {
1523                 m->errorOut(e, "SubSampleCommand", "getSubSampleSabund");
1524                 exit(1);
1525         }
1526 }
1527 //**********************************************************************************************************************
1528 int SubSampleCommand::processSabund(SAbundVector*& sabund, ofstream& out) {
1529         try {
1530
1531                 RAbundVector* rabund = new RAbundVector();
1532                 *rabund = sabund->getRAbundVector();
1533
1534                 int numBins = rabund->getNumBins();
1535                 int thisSize = rabund->getNumSeqs();
1536
1537                 if (thisSize != size) {
1538
1539                         OrderVector* order = new OrderVector();
1540                         for(int p=0;p<numBins;p++){
1541                                 for(int j=0;j<rabund->get(p);j++){
1542                                         order->push_back(p);
1543                                 }
1544                         }
1545                         random_shuffle(order->begin(), order->end());
1546
1547                         RAbundVector* temp = new RAbundVector(numBins);
1548                         temp->setLabel(rabund->getLabel());
1549
1550                         delete rabund;
1551                         rabund = temp;
1552
1553                         for (int j = 0; j < size; j++) {
1554
1555                                 if (m->control_pressed) { delete order; return 0; }
1556
1557                                 int bin = order->get(j);
1558
1559                                 int abund = rabund->get(bin);
1560                                 rabund->set(bin, (abund+1));
1561                         }
1562
1563                         delete order;
1564                 }
1565
1566                 if (m->control_pressed) { return 0; }
1567
1568                 delete sabund;
1569                 sabund = new SAbundVector();
1570                 *sabund = rabund->getSAbundVector();
1571                 delete rabund;
1572
1573                 sabund->print(out);
1574
1575                 return 0;
1576
1577         }
1578         catch(exception& e) {
1579                 m->errorOut(e, "SubSampleCommand", "processSabund");
1580                 exit(1);
1581         }
1582 }
1583 //**********************************************************************************************************************
1584
1585
1586