subsamplecommand.cpp

   1 /*
   2  *  subsamplecommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 10/27/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "subsamplecommand.h"
  11 #include "sharedutilities.h"
  12 #include "deconvolutecommand.h"
  13
  14 //**********************************************************************************************************************
  15 vector<string> SubSampleCommand::setParameters(){
  16         try {
  17                 CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "FLSSR", "none",false,false); parameters.push_back(pfasta);
  18                 CommandParameter pname("name", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pname);
  19                 CommandParameter pgroup("group", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pgroup);
  20                 CommandParameter plist("list", "InputTypes", "", "", "none", "FLSSR", "none",false,false); parameters.push_back(plist);
  21                 CommandParameter pshared("shared", "InputTypes", "", "", "none", "FLSSR", "none",false,false); parameters.push_back(pshared);
  22                 CommandParameter prabund("rabund", "InputTypes", "", "", "none", "FLSSR", "none",false,false); parameters.push_back(prabund);
  23                 CommandParameter psabund("sabund", "InputTypes", "", "", "none", "FLSSR", "none",false,false); parameters.push_back(psabund);
  24                 CommandParameter plabel("label", "String", "", "", "", "", "",false,false); parameters.push_back(plabel);
  25                 CommandParameter pgroups("groups", "String", "", "", "", "", "",false,false); parameters.push_back(pgroups);
  26                 CommandParameter psize("size", "Number", "", "0", "", "", "",false,false); parameters.push_back(psize);
  27                 CommandParameter ppersample("persample", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(ppersample);
  28                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  29                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  30
  31                 vector<string> myArray;
  32                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  33                 return myArray;
  34         }
  35         catch(exception& e) {
  36                 m->errorOut(e, "SubSampleCommand", "setParameters");
  37                 exit(1);
  38         }
  39 }
  40 //**********************************************************************************************************************
  41 string SubSampleCommand::getHelpString(){
  42         try {
  43                 string helpString = "";
  44                 helpString += "The sub.sample command is designed to be used as a way to normalize your data, or create a smaller set from your original set.\n";
  45                 helpString += "The sub.sample command parameters are fasta, name, list, group, rabund, sabund, shared, groups, size, persample and label.  You must provide a fasta, list, sabund, rabund or shared file as an input file.\n";
  46                 helpString += "The namefile is only used with the fasta file, not with the listfile, because the list file should contain all sequences.\n";
  47                 helpString += "The groups parameter allows you to specify which of the groups in your groupfile you would like included. The group names are separated by dashes.\n";
  48                 helpString += "The label parameter allows you to select what distance levels you would like, and are also separated by dashes.\n";
  49                 helpString += "The size parameter allows you indicate the size of your subsample.\n";
  50                 helpString += "The persample parameter allows you indicate you want to select subsample of the same size from each of your groups, default=false. It is only used with the list and fasta files if a groupfile is given.\n";
  51                 helpString += "persample=false will select a random set of sequences of the size you select, but the number of seqs from each group may differ.\n";
  52                 helpString += "The size parameter is not set: with shared file size=number of seqs in smallest sample, with all other files if a groupfile is given and persample=true, then size=number of seqs in smallest sample, otherwise size=10% of number of seqs.\n";
  53                 helpString += "The sub.sample command should be in the following format: sub.sample(list=yourListFile, group=yourGroupFile, groups=yourGroups, label=yourLabels).\n";
  54                 helpString += "Example sub.sample(list=abrecovery.fn.list, group=abrecovery.groups, groups=B-C, size=20).\n";
  55                 helpString += "The default value for groups is all the groups in your groupfile, and all labels in your inputfile will be used.\n";
  56                 helpString += "The sub.sample command outputs a .subsample file.\n";
  57                 helpString += "Note: No spaces between parameter labels (i.e. groups), '=' and parameters (i.e.yourGroups).\n";
  58                 return helpString;
  59         }
  60         catch(exception& e) {
  61                 m->errorOut(e, "SubSampleCommand", "getHelpString");
  62                 exit(1);
  63         }
  64 }
  65 //**********************************************************************************************************************
  66 SubSampleCommand::SubSampleCommand(){
  67         try {
  68                 abort = true; calledHelp = true;
  69                 setParameters();
  70                 vector<string> tempOutNames;
  71                 outputTypes["shared"] = tempOutNames;
  72                 outputTypes["list"] = tempOutNames;
  73                 outputTypes["rabund"] = tempOutNames;
  74                 outputTypes["sabund"] = tempOutNames;
  75                 outputTypes["fasta"] = tempOutNames;
  76                 outputTypes["name"] = tempOutNames;
  77                 outputTypes["group"] = tempOutNames;
  78         }
  79         catch(exception& e) {
  80                 m->errorOut(e, "SubSampleCommand", "GetRelAbundCommand");
  81                 exit(1);
  82         }
  83 }
  84 //**********************************************************************************************************************
  85 SubSampleCommand::SubSampleCommand(string option) {
  86         try {
  87                 abort = false; calledHelp = false;
  88                 allLines = 1;
  89
  90                 //allow user to run help
  91                 if(option == "help") { help(); abort = true; calledHelp = true; }
  92                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  93
  94                 else {
  95                         vector<string> myArray = setParameters();
  96
  97                         OptionParser parser(option);
  98                         map<string,string> parameters = parser.getParameters();
  99
 100                         ValidParameters validParameter;
 101
 102                         //check to make sure all parameters are valid for command
 103                         map<string,string>::iterator it;
 104                         for (it = parameters.begin(); it != parameters.end(); it++) {
 105                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 106                         }
 107
 108                         //initialize outputTypes
 109                         vector<string> tempOutNames;
 110                         outputTypes["shared"] = tempOutNames;
 111                         outputTypes["list"] = tempOutNames;
 112                         outputTypes["rabund"] = tempOutNames;
 113                         outputTypes["sabund"] = tempOutNames;
 114                         outputTypes["fasta"] = tempOutNames;
 115                         outputTypes["name"] = tempOutNames;
 116                         outputTypes["group"] = tempOutNames;
 117
 118                         //if the user changes the output directory command factory will send this info to us in the output parameter
 119                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
 120
 121                         //if the user changes the input directory command factory will send this info to us in the output parameter
 122                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 123                         if (inputDir == "not found"){   inputDir = "";          }
 124                         else {
 125                                 string path;
 126                                 it = parameters.find("list");
 127                                 //user has given a template file
 128                                 if(it != parameters.end()){
 129                                         path = m->hasPath(it->second);
 130                                         //if the user has not given a path then, add inputdir. else leave path alone.
 131                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 132                                 }
 133
 134                                 it = parameters.find("fasta");
 135                                 //user has given a template file
 136                                 if(it != parameters.end()){
 137                                         path = m->hasPath(it->second);
 138                                         //if the user has not given a path then, add inputdir. else leave path alone.
 139                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 140                                 }
 141
 142                                 it = parameters.find("shared");
 143                                 //user has given a template file
 144                                 if(it != parameters.end()){
 145                                         path = m->hasPath(it->second);
 146                                         //if the user has not given a path then, add inputdir. else leave path alone.
 147                                         if (path == "") {       parameters["shared"] = inputDir + it->second;           }
 148                                 }
 149
 150                                 it = parameters.find("group");
 151                                 //user has given a template file
 152                                 if(it != parameters.end()){
 153                                         path = m->hasPath(it->second);
 154                                         //if the user has not given a path then, add inputdir. else leave path alone.
 155                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 156                                 }
 157
 158                                 it = parameters.find("sabund");
 159                                 //user has given a template file
 160                                 if(it != parameters.end()){
 161                                         path = m->hasPath(it->second);
 162                                         //if the user has not given a path then, add inputdir. else leave path alone.
 163                                         if (path == "") {       parameters["sabund"] = inputDir + it->second;           }
 164                                 }
 165
 166                                 it = parameters.find("rabund");
 167                                 //user has given a template file
 168                                 if(it != parameters.end()){
 169                                         path = m->hasPath(it->second);
 170                                         //if the user has not given a path then, add inputdir. else leave path alone.
 171                                         if (path == "") {       parameters["rabund"] = inputDir + it->second;           }
 172                                 }
 173
 174                                 it = parameters.find("name");
 175                                 //user has given a template file
 176                                 if(it != parameters.end()){
 177                                         path = m->hasPath(it->second);
 178                                         //if the user has not given a path then, add inputdir. else leave path alone.
 179                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 180                                 }
 181                         }
 182
 183                         //check for required parameters
 184                         listfile = validParameter.validFile(parameters, "list", true);
 185                         if (listfile == "not open") { listfile = ""; abort = true; }
 186                         else if (listfile == "not found") { listfile = ""; }
 187                         else { m->setListFile(listfile); }
 188
 189                         sabundfile = validParameter.validFile(parameters, "sabund", true);
 190                         if (sabundfile == "not open") { sabundfile = ""; abort = true; }
 191                         else if (sabundfile == "not found") { sabundfile = ""; }
 192                         else { m->setSabundFile(sabundfile); }
 193
 194                         rabundfile = validParameter.validFile(parameters, "rabund", true);
 195                         if (rabundfile == "not open") { rabundfile = ""; abort = true; }
 196                         else if (rabundfile == "not found") { rabundfile = ""; }
 197                         else { m->setRabundFile(rabundfile); }
 198
 199                         fastafile = validParameter.validFile(parameters, "fasta", true);
 200                         if (fastafile == "not open") { fastafile = ""; abort = true; }
 201                         else if (fastafile == "not found") { fastafile = ""; }
 202                         else { m->setFastaFile(fastafile); }
 203
 204                         sharedfile = validParameter.validFile(parameters, "shared", true);
 205                         if (sharedfile == "not open") { sharedfile = ""; abort = true; }
 206                         else if (sharedfile == "not found") { sharedfile = ""; }
 207                         else { m->setSharedFile(sharedfile); }
 208
 209                         namefile = validParameter.validFile(parameters, "name", true);
 210                         if (namefile == "not open") { namefile = ""; abort = true; }
 211                         else if (namefile == "not found") { namefile = ""; }
 212                         else { m->setNameFile(namefile); }
 213
 214                         groupfile = validParameter.validFile(parameters, "group", true);
 215                         if (groupfile == "not open") { groupfile = ""; abort = true; }
 216                         else if (groupfile == "not found") { groupfile = ""; }
 217                         else { m->setGroupFile(groupfile); }
 218
 219                         //check for optional parameter and set defaults
 220                         // ...at some point should added some additional type checking...
 221                         label = validParameter.validFile(parameters, "label", false);
 222                         if (label == "not found") { label = ""; }
 223                         else {
 224                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 225                                 else { allLines = 1;  }
 226                         }
 227
 228                         groups = validParameter.validFile(parameters, "groups", false);
 229                         if (groups == "not found") { groups = ""; pickedGroups = false; }
 230                         else {
 231                                 pickedGroups = true;
 232                                 m->splitAtDash(groups, Groups);
 233                                 m->setGroups(Groups);
 234                         }
 235
 236                         string temp = validParameter.validFile(parameters, "size", false);              if (temp == "not found"){       temp = "0";             }
 237                         m->mothurConvert(temp, size);
 238
 239                         temp = validParameter.validFile(parameters, "persample", false);                if (temp == "not found"){       temp = "f";             }
 240                         persample = m->isTrue(temp);
 241
 242                         if (groupfile == "") { persample = false; }
 243
 244                         if ((namefile != "") && (fastafile == "")) { m->mothurOut("You may only use a namefile with a fastafile."); m->mothurOutEndLine(); abort = true; }
 245
 246                         if ((fastafile == "") && (listfile == "") && (sabundfile == "") && (rabundfile == "") && (sharedfile == "")) {
 247                                 m->mothurOut("You must provide a fasta, list, sabund, rabund or shared file as an input file."); m->mothurOutEndLine(); abort = true; }
 248
 249                         if (pickedGroups && ((groupfile == "") && (sharedfile == ""))) {
 250                                 m->mothurOut("You cannot pick groups without a valid group file or shared file."); m->mothurOutEndLine(); abort = true; }
 251
 252                         if ((groupfile != "") && ((fastafile == "") && (listfile == ""))) {
 253                                 m->mothurOut("Group file only valid with listfile or fastafile."); m->mothurOutEndLine(); abort = true; }
 254
 255                         if ((groupfile != "") && ((fastafile != "") && (listfile != ""))) {
 256                                 m->mothurOut("A new group file can only be made from the subsample of a listfile or fastafile, not both. Please correct."); m->mothurOutEndLine(); abort = true; }
 257
 258                         if ((fastafile != "") && (namefile == "")) {
 259                                 vector<string> files; files.push_back(fastafile);
 260                                 parser.getNameFile(files);
 261                         }
 262                 }
 263
 264         }
 265         catch(exception& e) {
 266                 m->errorOut(e, "SubSampleCommand", "SubSampleCommand");
 267                 exit(1);
 268         }
 269 }
 270 //**********************************************************************************************************************
 271
 272 int SubSampleCommand::execute(){
 273         try {
 274
 275                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 276
 277                 if (sharedfile != "")   {   getSubSampleShared();       }
 278                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]); return 0; } }
 279
 280                 if (listfile != "")             {   getSubSampleList();         }
 281                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]); return 0; } }
 282
 283                 if (rabundfile != "")   {   getSubSampleRabund();       }
 284                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]); return 0; } }
 285
 286                 if (sabundfile != "")   {   getSubSampleSabund();       }
 287                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]); return 0; } }
 288
 289                 if (fastafile != "")    {   getSubSampleFasta();        }
 290                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]); return 0; } }
 291
 292                 //set fasta file as new current fastafile
 293                 string current = "";
 294                 itTypes = outputTypes.find("fasta");
 295                 if (itTypes != outputTypes.end()) {
 296                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 297                 }
 298
 299                 itTypes = outputTypes.find("name");
 300                 if (itTypes != outputTypes.end()) {
 301                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
 302                 }
 303
 304                 itTypes = outputTypes.find("group");
 305                 if (itTypes != outputTypes.end()) {
 306                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setGroupFile(current); }
 307                 }
 308
 309                 itTypes = outputTypes.find("list");
 310                 if (itTypes != outputTypes.end()) {
 311                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 312                 }
 313
 314                 itTypes = outputTypes.find("shared");
 315                 if (itTypes != outputTypes.end()) {
 316                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setSharedFile(current); }
 317                 }
 318
 319                 itTypes = outputTypes.find("rabund");
 320                 if (itTypes != outputTypes.end()) {
 321                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setRabundFile(current); }
 322                 }
 323
 324                 itTypes = outputTypes.find("sabund");
 325                 if (itTypes != outputTypes.end()) {
 326                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setSabundFile(current); }
 327                 }
 328
 329
 330                 m->mothurOutEndLine();
 331                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 332                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 333                 m->mothurOutEndLine();
 334
 335                 return 0;
 336         }
 337         catch(exception& e) {
 338                 m->errorOut(e, "SubSampleCommand", "execute");
 339                 exit(1);
 340         }
 341 }
 342 //**********************************************************************************************************************
 343 int SubSampleCommand::getSubSampleFasta() {
 344         try {
 345
 346                 if (namefile != "") { readNames(); }    //fills names with all names in namefile.
 347                 else { getNames(); }//no name file, so get list of names to pick from
 348
 349                 GroupMap* groupMap;
 350                 if (groupfile != "") {
 351
 352                         groupMap = new GroupMap(groupfile);
 353                         groupMap->readMap();
 354
 355                         //takes care of user setting groupNames that are invalid or setting groups=all
 356                         SharedUtil* util = new SharedUtil();
 357                         vector<string> namesGroups = groupMap->getNamesOfGroups();
 358                         util->setGroups(Groups, namesGroups);
 359                         delete util;
 360
 361                         //file mismatch quit
 362                         if (names.size() != groupMap->getNumSeqs()) {
 363                                 m->mothurOut("[ERROR]: your fasta file contains " + toString(names.size()) + " sequences, and your groupfile contains " + toString(groupMap->getNumSeqs()) + ", please correct.");
 364                                 m->mothurOutEndLine();
 365                                 delete groupMap;
 366                                 return 0;
 367                         }
 368                 }
 369
 370                 if (m->control_pressed) { return 0; }
 371
 372
 373                 //make sure that if your picked groups size is not too big
 374                 int thisSize = names.size();
 375                 if (persample) {
 376                         if (size == 0) { //user has not set size, set size = smallest samples size
 377                                 size = groupMap->getNumSeqs(Groups[0]);
 378                                 for (int i = 1; i < Groups.size(); i++) {
 379                                         int thisSize = groupMap->getNumSeqs(Groups[i]);
 380
 381                                         if (thisSize < size) {  size = thisSize;        }
 382                                 }
 383                         }else { //make sure size is not too large
 384                                 vector<string> newGroups;
 385                                 for (int i = 0; i < Groups.size(); i++) {
 386                                         int thisSize = groupMap->getNumSeqs(Groups[i]);
 387
 388                                         if (thisSize >= size) { newGroups.push_back(Groups[i]); }
 389                                         else {  m->mothurOut("You have selected a size that is larger than " + Groups[i] + " number of sequences, removing " + Groups[i] + "."); m->mothurOutEndLine(); }
 390                                 }
 391                                 Groups = newGroups;
 392                         }
 393
 394                         m->mothurOut("Sampling " + toString(size) + " from each group."); m->mothurOutEndLine();
 395                 }else {
 396                         if (pickedGroups) {
 397                                 int total = 0;
 398                                 for(int i = 0; i < Groups.size(); i++) {
 399                                         total += groupMap->getNumSeqs(Groups[i]);
 400                                 }
 401
 402                                 if (size == 0) { //user has not set size, set size = 10% samples size
 403                                         size = int (total * 0.10);
 404                                 }
 405
 406                                 if (total < size) {
 407                                         if (size != 0) {
 408                                                 m->mothurOut("Your size is too large for the number of groups you selected. Adjusting to " + toString(int (total * 0.10)) + "."); m->mothurOutEndLine();
 409                                         }
 410                                         size = int (total * 0.10);
 411                                 }
 412
 413                                 m->mothurOut("Sampling " + toString(size) + " from " + toString(total) + "."); m->mothurOutEndLine();
 414                         }
 415
 416                         if (size == 0) { //user has not set size, set size = 10% samples size
 417                                 size = int (names.size() * 0.10);
 418                         }
 419
 420                         if (size > thisSize) { m->mothurOut("Your fasta file only contains " + toString(thisSize) + " sequences. Setting size to " + toString(thisSize) + "."); m->mothurOutEndLine();
 421                                 size = thisSize;
 422                         }
 423
 424                         if (!pickedGroups) { m->mothurOut("Sampling " + toString(size) + " from " + toString(thisSize) + "."); m->mothurOutEndLine(); }
 425
 426                 }
 427                 random_shuffle(names.begin(), names.end());
 428
 429                 set<string> subset; //dont want repeat sequence names added
 430                 if (persample) {
 431                         //initialize counts
 432                         map<string, int> groupCounts;
 433                         map<string, int>::iterator itGroupCounts;
 434                         for (int i = 0; i < Groups.size(); i++) { groupCounts[Groups[i]] = 0; }
 435
 436                         for (int j = 0; j < names.size(); j++) {
 437
 438                                 if (m->control_pressed) { return 0; }
 439
 440                                 string group = groupMap->getGroup(names[j]);
 441                                 if (group == "not found") { m->mothurOut("[ERROR]: " + names[j] + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
 442                                 else{
 443                                         itGroupCounts = groupCounts.find(group);
 444                                         if (itGroupCounts != groupCounts.end()) {
 445                                                 if (groupCounts[group] < size) {        subset.insert(names[j]);        groupCounts[group]++; }
 446                                         }
 447                                 }
 448                         }
 449                 }else {
 450
 451                         //randomly select a subset of those names to include in the subsample
 452                         //since names was randomly shuffled just grab the next one
 453                         for (int j = 0; j < names.size(); j++) {
 454
 455                                 if (m->control_pressed) { return 0; }
 456
 457                                 if (groupfile != "") { //if there is a groupfile given fill in group info
 458                                         string group = groupMap->getGroup(names[j]);
 459                                         if (group == "not found") { m->mothurOut("[ERROR]: " + names[j] + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
 460
 461                                         if (pickedGroups) { //if hte user picked groups, we only want to keep the names of sequences from those groups
 462                                                 if (m->inUsersGroups(group, Groups)) {
 463                                                         subset.insert(names[j]);
 464                                                 }
 465                                         }else{
 466                                                 subset.insert(names[j]);
 467                                         }
 468                                 }else{ //save everyone, group
 469                                         subset.insert(names[j]);
 470                                 }
 471
 472                                 //do we have enough??
 473                                 if (subset.size() == size) { break; }
 474                         }
 475                 }
 476
 477                 if (subset.size() == 0) {  m->mothurOut("The size you selected is too large, skipping fasta file."); m->mothurOutEndLine();  return 0; }
 478
 479                 string thisOutputDir = outputDir;
 480                 if (outputDir == "") {  thisOutputDir += m->hasPath(fastafile);  }
 481                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(fastafile)) + "subsample" + m->getExtension(fastafile);
 482
 483                 ofstream out;
 484                 m->openOutputFile(outputFileName, out);
 485
 486                 //read through fasta file outputting only the names on the subsample list
 487                 ifstream in;
 488                 m->openInputFile(fastafile, in);
 489
 490                 string thisname;
 491                 int count = 0;
 492                 map<string, vector<string> >::iterator itNameMap;
 493
 494                 while(!in.eof()){
 495
 496                         if (m->control_pressed) { in.close(); out.close();  return 0; }
 497
 498                         Sequence currSeq(in);
 499                         thisname = currSeq.getName();
 500
 501                         if (thisname != "") {
 502
 503                                 //does the subset contain a sequence that this sequence represents
 504                                 itNameMap = nameMap.find(thisname);
 505                                 if (itNameMap != nameMap.end()) {
 506                                         vector<string> nameRepresents = itNameMap->second;
 507
 508                                         for (int i = 0; i < nameRepresents.size(); i++){
 509                                                 if (subset.count(nameRepresents[i]) != 0) {
 510                                                         out << ">" << nameRepresents[i] << endl << currSeq.getAligned() << endl;
 511                                                         count++;
 512                                                 }
 513                                         }
 514                                 }else{
 515                                         m->mothurOut("[ERROR]: " + thisname + " is not in your namefile, please correct."); m->mothurOutEndLine();
 516                                 }
 517                         }
 518                         m->gobble(in);
 519                 }
 520                 in.close();
 521                 out.close();
 522
 523                 if (count != subset.size()) {
 524                         m->mothurOut("[ERROR]: The subset selected contained " + toString(subset.size()) + " sequences, but I only found " + toString(count) + " of those in the fastafile."); m->mothurOutEndLine();
 525                 }
 526
 527                 if (namefile != "") {
 528                         m->mothurOut("Deconvoluting subsampled fasta file... "); m->mothurOutEndLine();
 529
 530                         //use unique.seqs to create new name and fastafile
 531                         string inputString = "fasta=" + outputFileName;
 532                         m->mothurOut("/******************************************/"); m->mothurOutEndLine();
 533                         m->mothurOut("Running command: unique.seqs(" + inputString + ")"); m->mothurOutEndLine();
 534
 535                         Command* uniqueCommand = new DeconvoluteCommand(inputString);
 536                         uniqueCommand->execute();
 537
 538                         map<string, vector<string> > filenames = uniqueCommand->getOutputFiles();
 539
 540                         delete uniqueCommand;
 541
 542                         outputTypes["name"].push_back(filenames["name"][0]);  outputNames.push_back(filenames["name"][0]);
 543                         m->mothurRemove(outputFileName);
 544                         outputFileName = filenames["fasta"][0];
 545
 546                         m->mothurOut("/******************************************/"); m->mothurOutEndLine();
 547
 548                         m->mothurOut("Done."); m->mothurOutEndLine();
 549                 }
 550
 551                 outputTypes["fasta"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 552
 553                 //if a groupfile is provided read through the group file only outputting the names on the subsample list
 554                 if (groupfile != "") {
 555
 556                         string groupOutputDir = outputDir;
 557                         if (outputDir == "") {  groupOutputDir += m->hasPath(groupfile);  }
 558                         string groupOutputFileName = groupOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "subsample" + m->getExtension(groupfile);
 559
 560                         ofstream outGroup;
 561                         m->openOutputFile(groupOutputFileName, outGroup);
 562                         outputTypes["group"].push_back(groupOutputFileName);  outputNames.push_back(groupOutputFileName);
 563
 564                         ifstream inGroup;
 565                         m->openInputFile(groupfile, inGroup);
 566                         string name, group;
 567
 568                         while(!inGroup.eof()){
 569
 570                                 if (m->control_pressed) { inGroup.close(); outGroup.close(); return 0; }
 571
 572                                 inGroup >> name;        m->gobble(inGroup);                     //read from first column
 573                                 inGroup >> group;                       //read from second column
 574
 575                                 //if this name is in the accnos file
 576                                 if (subset.count(name) != 0) {
 577                                         outGroup << name << '\t' << group << endl;
 578                                         subset.erase(name);
 579                                 }
 580
 581                                 m->gobble(inGroup);
 582                         }
 583                         inGroup.close();
 584                         outGroup.close();
 585
 586                         //sanity check
 587                         if (subset.size() != 0) {
 588                                 m->mothurOut("Your groupfile does not match your fasta file."); m->mothurOutEndLine();
 589                                 for (set<string>::iterator it = subset.begin(); it != subset.end(); it++) {
 590                                         m->mothurOut("[ERROR]: " + *it + " is missing from your groupfile."); m->mothurOutEndLine();
 591                                 }
 592                         }
 593                 }
 594
 595                 return 0;
 596
 597         }
 598         catch(exception& e) {
 599                 m->errorOut(e, "SubSampleCommand", "getSubSampleFasta");
 600                 exit(1);
 601         }
 602 }
 603 //**********************************************************************************************************************
 604 int SubSampleCommand::getNames() {
 605         try {
 606
 607                 ifstream in;
 608                 m->openInputFile(fastafile, in);
 609
 610                 string thisname;
 611                 while(!in.eof()){
 612
 613                         if (m->control_pressed) { in.close(); return 0; }
 614
 615                         Sequence currSeq(in);
 616                         thisname = currSeq.getName();
 617
 618                         if (thisname != "") {
 619                                 vector<string> temp; temp.push_back(thisname);
 620                                 nameMap[thisname] = temp;
 621                                 names.push_back(thisname);
 622                         }
 623                         m->gobble(in);
 624                 }
 625                 in.close();
 626
 627                 return 0;
 628
 629         }
 630         catch(exception& e) {
 631                 m->errorOut(e, "SubSampleCommand", "getNames");
 632                 exit(1);
 633         }
 634 }
 635 //**********************************************************************************************************************
 636 int SubSampleCommand::readNames() {
 637         try {
 638
 639                 ifstream in;
 640                 m->openInputFile(namefile, in);
 641
 642                 string thisname, repnames;
 643                 map<string, vector<string> >::iterator it;
 644
 645                 while(!in.eof()){
 646
 647                         if (m->control_pressed) { in.close(); return 0; }
 648
 649                         in >> thisname;         m->gobble(in);          //read from first column
 650                         in >> repnames;                 //read from second column
 651
 652                         it = nameMap.find(thisname);
 653                         if (it == nameMap.end()) {
 654
 655                                 vector<string> splitRepNames;
 656                                 m->splitAtComma(repnames, splitRepNames);
 657
 658                                 nameMap[thisname] = splitRepNames;
 659                                 for (int i = 0; i < splitRepNames.size(); i++) { names.push_back(splitRepNames[i]); }
 660
 661                         }else{  m->mothurOut(thisname + " is already in namesfile. I will use first definition."); m->mothurOutEndLine();  }
 662
 663                         m->gobble(in);
 664                 }
 665                 in.close();
 666
 667                 return 0;
 668
 669         }
 670         catch(exception& e) {
 671                 m->errorOut(e, "SubSampleCommand", "readNames");
 672                 exit(1);
 673         }
 674 }
 675 //**********************************************************************************************************************
 676 int SubSampleCommand::getSubSampleShared() {
 677         try {
 678
 679                 InputData* input = new InputData(sharedfile, "sharedfile");
 680                 vector<SharedRAbundVector*> lookup = input->getSharedRAbundVectors();
 681                 string lastLabel = lookup[0]->getLabel();
 682
 683                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 684                 set<string> processedLabels;
 685                 set<string> userLabels = labels;
 686
 687                 if (size == 0) { //user has not set size, set size = smallest samples size
 688                         size = lookup[0]->getNumSeqs();
 689                         for (int i = 1; i < lookup.size(); i++) {
 690                                 int thisSize = lookup[i]->getNumSeqs();
 691
 692                                 if (thisSize < size) {  size = thisSize;        }
 693                         }
 694                 }else {
 695                         m->clearGroups();
 696                         Groups.clear();
 697                         vector<SharedRAbundVector*> temp;
 698                         for (int i = 0; i < lookup.size(); i++) {
 699                                 if (lookup[i]->getNumSeqs() < size) {
 700                                         m->mothurOut(lookup[i]->getGroup() + " contains " + toString(lookup[i]->getNumSeqs()) + ". Eliminating."); m->mothurOutEndLine();
 701                                         delete lookup[i];
 702                                 }else {
 703                                         Groups.push_back(lookup[i]->getGroup());
 704                                         temp.push_back(lookup[i]);
 705                                 }
 706                         }
 707                         lookup = temp;
 708                         m->setGroups(Groups);
 709                 }
 710
 711                 if (lookup.size() == 0) {  m->mothurOut("The size you selected is too large, skipping shared file."); m->mothurOutEndLine(); delete input; return 0; }
 712
 713                 m->mothurOut("Sampling " + toString(size) + " from each group."); m->mothurOutEndLine();
 714
 715                 //as long as you are not at the end of the file or done wih the lines you want
 716                 while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 717                         if (m->control_pressed) {  delete input; for (int i = 0; i < lookup.size(); i++) {  delete lookup[i]; lookup[i] = NULL; }  return 0;  }
 718
 719                         if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){
 720
 721                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 722
 723                                 processShared(lookup);
 724
 725                                 processedLabels.insert(lookup[0]->getLabel());
 726                                 userLabels.erase(lookup[0]->getLabel());
 727                         }
 728
 729                         if ((m->anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 730                                 string saveLabel = lookup[0]->getLabel();
 731
 732                                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 733
 734                                 lookup = input->getSharedRAbundVectors(lastLabel);
 735                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 736
 737                                 processShared(lookup);
 738
 739                                 processedLabels.insert(lookup[0]->getLabel());
 740                                 userLabels.erase(lookup[0]->getLabel());
 741
 742                                 //restore real lastlabel to save below
 743                                 lookup[0]->setLabel(saveLabel);
 744                         }
 745
 746                         lastLabel = lookup[0]->getLabel();
 747                         //prevent memory leak
 748                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i]; lookup[i] = NULL; }
 749
 750                         //get next line to process
 751                         lookup = input->getSharedRAbundVectors();
 752                 }
 753
 754
 755                 if (m->control_pressed) {   return 0;  }
 756
 757                 //output error messages about any remaining user labels
 758                 set<string>::iterator it;
 759                 bool needToRun = false;
 760                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 761                         m->mothurOut("Your file does not include the label " + *it);
 762                         if (processedLabels.count(lastLabel) != 1) {
 763                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 764                                 needToRun = true;
 765                         }else {
 766                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 767                         }
 768                 }
 769
 770                 //run last label if you need to
 771                 if (needToRun == true)  {
 772                         for (int i = 0; i < lookup.size(); i++) { if (lookup[i] != NULL) { delete lookup[i]; } }
 773                         lookup = input->getSharedRAbundVectors(lastLabel);
 774
 775                         m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 776
 777                         processShared(lookup);
 778
 779                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 780                 }
 781
 782                 delete input;
 783
 784                 return 0;
 785
 786         }
 787         catch(exception& e) {
 788                 m->errorOut(e, "SubSampleCommand", "getSubSampleShared");
 789                 exit(1);
 790         }
 791 }
 792 //**********************************************************************************************************************
 793 int SubSampleCommand::processShared(vector<SharedRAbundVector*>& thislookup) {
 794         try {
 795
 796                 //save mothurOut's binLabels to restore for next label
 797                 vector<string> saveBinLabels = m->currentBinLabels;
 798
 799                 string thisOutputDir = outputDir;
 800                 if (outputDir == "") {  thisOutputDir += m->hasPath(sharedfile);  }
 801                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(sharedfile)) + thislookup[0]->getLabel() + ".subsample" + m->getExtension(sharedfile);
 802
 803
 804                 ofstream out;
 805                 m->openOutputFile(outputFileName, out);
 806                 outputTypes["shared"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 807
 808                 int numBins = thislookup[0]->getNumBins();
 809                 for (int i = 0; i < thislookup.size(); i++) {
 810                         int thisSize = thislookup[i]->getNumSeqs();
 811
 812                         if (thisSize != size) {
 813
 814                                 string thisgroup = thislookup[i]->getGroup();
 815
 816                                 OrderVector* order = new OrderVector();
 817                                 for(int p=0;p<numBins;p++){
 818                                         for(int j=0;j<thislookup[i]->getAbundance(p);j++){
 819                                                 order->push_back(p);
 820                                         }
 821                                 }
 822                                 random_shuffle(order->begin(), order->end());
 823
 824                                 SharedRAbundVector* temp = new SharedRAbundVector(numBins);
 825                                 temp->setLabel(thislookup[i]->getLabel());
 826                                 temp->setGroup(thislookup[i]->getGroup());
 827
 828                                 delete thislookup[i];
 829                                 thislookup[i] = temp;
 830
 831
 832                                 for (int j = 0; j < size; j++) {
 833
 834                                         if (m->control_pressed) { delete order; out.close(); return 0; }
 835
 836                                         //get random number to sample from order between 0 and thisSize-1.
 837                                         //don't need this because of the random shuffle above
 838                                         //int myrand = int((float)(thisSize) * (float)(rand()) / ((float)RAND_MAX+1.0));
 839
 840                                         int bin = order->get(j);
 841
 842                                         int abund = thislookup[i]->getAbundance(bin);
 843                                         thislookup[i]->set(bin, (abund+1), thisgroup);
 844                                 }
 845                                 delete order;
 846                         }
 847                 }
 848
 849                 //subsampling may have created some otus with no sequences in them
 850                 eliminateZeroOTUS(thislookup);
 851
 852                 if (m->control_pressed) { out.close(); return 0; }
 853
 854                 thislookup[0]->printHeaders(out);
 855
 856                 for (int i = 0; i < thislookup.size(); i++) {
 857                         out << thislookup[i]->getLabel() << '\t' << thislookup[i]->getGroup() << '\t';
 858                         thislookup[i]->print(out);
 859                 }
 860
 861                 out.close();
 862
 863                 //save mothurOut's binLabels to restore for next label
 864                 m->currentBinLabels = saveBinLabels;
 865
 866                 return 0;
 867
 868         }
 869         catch(exception& e) {
 870                 m->errorOut(e, "SubSampleCommand", "processShared");
 871                 exit(1);
 872         }
 873 }
 874 //**********************************************************************************************************************
 875 int SubSampleCommand::getSubSampleList() {
 876         try {
 877
 878                 string thisOutputDir = outputDir;
 879                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 880                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "subsample" + m->getExtension(listfile);
 881
 882                 ofstream out;
 883                 m->openOutputFile(outputFileName, out);
 884                 outputTypes["list"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 885
 886                 InputData* input = new InputData(listfile, "list");
 887                 ListVector* list = input->getListVector();
 888                 string lastLabel = list->getLabel();
 889
 890                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 891                 set<string> processedLabels;
 892                 set<string> userLabels = labels;
 893
 894                 ofstream outGroup;
 895                 GroupMap* groupMap;
 896                 if (groupfile != "") {
 897
 898                         groupMap = new GroupMap(groupfile);
 899                         groupMap->readMap();
 900
 901                         //takes care of user setting groupNames that are invalid or setting groups=all
 902                         SharedUtil* util = new SharedUtil();
 903                         vector<string> namesGroups = groupMap->getNamesOfGroups();
 904                         util->setGroups(Groups, namesGroups);
 905                         delete util;
 906
 907                         //create outputfiles
 908                         string groupOutputDir = outputDir;
 909                         if (outputDir == "") {  groupOutputDir += m->hasPath(groupfile);  }
 910                         string groupOutputFileName = groupOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "subsample" + m->getExtension(groupfile);
 911
 912                         m->openOutputFile(groupOutputFileName, outGroup);
 913                         outputTypes["group"].push_back(groupOutputFileName);  outputNames.push_back(groupOutputFileName);
 914
 915                         //file mismatch quit
 916                         if (list->getNumSeqs() != groupMap->getNumSeqs()) {
 917                                 m->mothurOut("[ERROR]: your list file contains " + toString(list->getNumSeqs()) + " sequences, and your groupfile contains " + toString(groupMap->getNumSeqs()) + ", please correct.");
 918                                 m->mothurOutEndLine();
 919                                 delete groupMap;
 920                                 delete list;
 921                                 delete input;
 922                                 out.close();
 923                                 outGroup.close();
 924                                 return 0;
 925                         }
 926                 }
 927
 928                 //make sure that if your picked groups size is not too big
 929                 if (persample) {
 930                         if (size == 0) { //user has not set size, set size = smallest samples size
 931                                 size = groupMap->getNumSeqs(Groups[0]);
 932                                 for (int i = 1; i < Groups.size(); i++) {
 933                                         int thisSize = groupMap->getNumSeqs(Groups[i]);
 934
 935                                         if (thisSize < size) {  size = thisSize;        }
 936                                 }
 937                         }else { //make sure size is not too large
 938                                 vector<string> newGroups;
 939                                 for (int i = 0; i < Groups.size(); i++) {
 940                                         int thisSize = groupMap->getNumSeqs(Groups[i]);
 941
 942                                         if (thisSize >= size) { newGroups.push_back(Groups[i]); }
 943                                         else {  m->mothurOut("You have selected a size that is larger than " + Groups[i] + " number of sequences, removing " + Groups[i] + "."); m->mothurOutEndLine(); }
 944                                 }
 945                                 Groups = newGroups;
 946                         }
 947
 948                         m->mothurOut("Sampling " + toString(size) + " from each group."); m->mothurOutEndLine();
 949                 }else{
 950                         if (pickedGroups) {
 951                                 int total = 0;
 952                                 for(int i = 0; i < Groups.size(); i++) {
 953                                         total += groupMap->getNumSeqs(Groups[i]);
 954                                 }
 955
 956                                 if (size == 0) { //user has not set size, set size = 10% samples size
 957                                         size = int (total * 0.10);
 958                                 }
 959
 960                                 if (total < size) {
 961                                         m->mothurOut("Your size is too large for the number of groups you selected. Adjusting to " + toString(int (total * 0.10)) + "."); m->mothurOutEndLine();
 962                                         size = int (total * 0.10);
 963                                 }
 964
 965                                 m->mothurOut("Sampling " + toString(size) + " from " + toString(total) + "."); m->mothurOutEndLine();
 966                         }else{
 967
 968                                 if (size == 0) { //user has not set size, set size = 10% samples size
 969                                         size = int (list->getNumSeqs() * 0.10);
 970                                 }
 971
 972                                 int thisSize = list->getNumSeqs();
 973                                 if (size > thisSize) { m->mothurOut("Your list file only contains " + toString(thisSize) + " sequences. Setting size to " + toString(thisSize) + "."); m->mothurOutEndLine();
 974                                         size = thisSize;
 975                                 }
 976
 977                                 m->mothurOut("Sampling " + toString(size) + " from " + toString(list->getNumSeqs()) + "."); m->mothurOutEndLine();
 978                         }
 979                 }
 980
 981
 982                 //fill names
 983                 for (int i = 0; i < list->getNumBins(); i++) {
 984                         string binnames = list->get(i);
 985
 986                         //parse names
 987                         string individual = "";
 988                         int length = binnames.length();
 989                         for(int j=0;j<length;j++){
 990                                 if(binnames[j] == ','){
 991
 992                                         if (groupfile != "") { //if there is a groupfile given fill in group info
 993                                                 string group = groupMap->getGroup(individual);
 994                                                 if (group == "not found") { m->mothurOut("[ERROR]: " + individual + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
 995
 996                                                 if (pickedGroups) { //if hte user picked groups, we only want to keep the names of sequences from those groups
 997                                                         if (m->inUsersGroups(group, Groups)) {
 998                                                                 names.push_back(individual);
 999                                                         }
1000                                                 }else{
1001                                                         names.push_back(individual);
1002                                                 }
1003                                         }else{ //save everyone, group
1004                                                 names.push_back(individual);
1005                                         }
1006                                         individual = "";
1007                                 }
1008                                 else{
1009                                         individual += binnames[j];
1010                                 }
1011                         }
1012                         //save last name
1013                         if (groupfile != "") { //if there is a groupfile given fill in group info
1014                                 string group = groupMap->getGroup(individual);
1015                                 if (group == "not found") { m->mothurOut("[ERROR]: " + individual + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
1016
1017                                 if (pickedGroups) { //if hte user picked groups, we only want to keep the names of sequences from those groups
1018                                         if (m->inUsersGroups(group, Groups)) {
1019                                                 names.push_back(individual);
1020                                         }
1021                                 }else{
1022                                         names.push_back(individual);
1023                                 }
1024                         }else{ //save everyone, group
1025                                 names.push_back(individual);
1026                         }
1027                 }
1028
1029                 random_shuffle(names.begin(), names.end());
1030
1031                 //randomly select a subset of those names to include in the subsample
1032                 set<string> subset; //dont want repeat sequence names added
1033                 if (persample) {
1034                         //initialize counts
1035                         map<string, int> groupCounts;
1036                         map<string, int>::iterator itGroupCounts;
1037                         for (int i = 0; i < Groups.size(); i++) { groupCounts[Groups[i]] = 0; }
1038
1039                         for (int j = 0; j < names.size(); j++) {
1040
1041                                 if (m->control_pressed) { return 0; }
1042
1043                                 string group = groupMap->getGroup(names[j]);
1044                                 if (group == "not found") { m->mothurOut("[ERROR]: " + names[j] + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
1045                                 else{
1046                                         itGroupCounts = groupCounts.find(group);
1047                                         if (itGroupCounts != groupCounts.end()) {
1048                                                 if (groupCounts[group] < size) {        subset.insert(names[j]);        groupCounts[group]++; }
1049                                         }
1050                                 }
1051                         }
1052                 }else{
1053                         for (int j = 0; j < size; j++) {
1054
1055                                 if (m->control_pressed) { break; }
1056
1057                                 subset.insert(names[j]);
1058                         }
1059                 }
1060
1061                 if (groupfile != "") {
1062                         //write out new groupfile
1063                         for (set<string>::iterator it = subset.begin(); it != subset.end(); it++) {
1064                                 string group = groupMap->getGroup(*it);
1065                                 if (group == "not found") { group = "NOTFOUND"; }
1066
1067                                 outGroup << *it << '\t' << group << endl;
1068                         }
1069                         outGroup.close(); delete groupMap;
1070                 }
1071
1072
1073                 //as long as you are not at the end of the file or done wih the lines you want
1074                 while((list != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
1075
1076                         if (m->control_pressed) {  delete list; delete input; out.close();  return 0;  }
1077
1078                         if(allLines == 1 || labels.count(list->getLabel()) == 1){
1079
1080                                 m->mothurOut(list->getLabel()); m->mothurOutEndLine();
1081
1082                                 processList(list, out, subset);
1083
1084                                 processedLabels.insert(list->getLabel());
1085                                 userLabels.erase(list->getLabel());
1086                         }
1087
1088                         if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
1089                                 string saveLabel = list->getLabel();
1090
1091                                 delete list;
1092
1093                                 list = input->getListVector(lastLabel);
1094                                 m->mothurOut(list->getLabel()); m->mothurOutEndLine();
1095
1096                                 processList(list, out, subset);
1097
1098                                 processedLabels.insert(list->getLabel());
1099                                 userLabels.erase(list->getLabel());
1100
1101                                 //restore real lastlabel to save below
1102                                 list->setLabel(saveLabel);
1103                         }
1104
1105                         lastLabel = list->getLabel();
1106
1107                         delete list; list = NULL;
1108
1109                         //get next line to process
1110                         list = input->getListVector();
1111                 }
1112
1113
1114                 if (m->control_pressed) {  if (list != NULL) { delete list; } delete input; out.close(); return 0;  }
1115
1116                 //output error messages about any remaining user labels
1117                 set<string>::iterator it;
1118                 bool needToRun = false;
1119                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
1120                         m->mothurOut("Your file does not include the label " + *it);
1121                         if (processedLabels.count(lastLabel) != 1) {
1122                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
1123                                 needToRun = true;
1124                         }else {
1125                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
1126                         }
1127                 }
1128
1129                 //run last label if you need to
1130                 if (needToRun == true)  {
1131                         if (list != NULL) { delete list; }
1132
1133                         list = input->getListVector(lastLabel);
1134
1135                         m->mothurOut(list->getLabel()); m->mothurOutEndLine();
1136
1137                         processList(list, out, subset);
1138
1139                         delete list; list = NULL;
1140                 }
1141
1142                 out.close();
1143                 if (list != NULL) { delete list; }
1144                 delete input;
1145
1146                 return 0;
1147
1148         }
1149         catch(exception& e) {
1150                 m->errorOut(e, "SubSampleCommand", "getSubSampleList");
1151                 exit(1);
1152         }
1153 }
1154 //**********************************************************************************************************************
1155 int SubSampleCommand::processList(ListVector*& list, ofstream& out, set<string>& subset) {
1156         try {
1157
1158                 int numBins = list->getNumBins();
1159
1160                 ListVector* temp = new ListVector();
1161                 temp->setLabel(list->getLabel());
1162
1163                 for (int i = 0; i < numBins; i++) {
1164
1165                         if (m->control_pressed) { break; }
1166
1167                         string binnames = list->get(i);
1168
1169                         //parse names
1170                         string individual = "";
1171                         string newNames = "";
1172                         int length = binnames.length();
1173                         for(int j=0;j<length;j++){
1174                                 if(binnames[j] == ','){
1175                                         if (subset.count(individual) != 0) {  newNames += individual + ",";  }
1176                                         individual = "";
1177                                 }else{
1178                                         individual += binnames[j];
1179                                 }
1180                         }
1181                         if (subset.count(individual) != 0) {  newNames += individual + ",";  }
1182
1183
1184                         //if there are names in this bin add to new list
1185                         if (newNames != "") {
1186                                 newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
1187                                 temp->push_back(newNames);
1188                         }
1189                 }
1190
1191                 delete list;
1192                 list = temp;
1193
1194                 if (m->control_pressed) { return 0; }
1195
1196                 list->print(out);
1197
1198                 return 0;
1199
1200         }
1201         catch(exception& e) {
1202                 m->errorOut(e, "SubSampleCommand", "processList");
1203                 exit(1);
1204         }
1205 }
1206 //**********************************************************************************************************************
1207 int SubSampleCommand::getSubSampleRabund() {
1208         try {
1209                 InputData* input = new InputData(rabundfile, "rabund");
1210                 RAbundVector* rabund = input->getRAbundVector();
1211                 string lastLabel = rabund->getLabel();
1212
1213                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
1214                 set<string> processedLabels;
1215                 set<string> userLabels = labels;
1216
1217                 if (size == 0) { //user has not set size, set size = 10%
1218                         size = int((rabund->getNumSeqs()) * 0.10);
1219                 }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; }
1220
1221                 m->mothurOut("Sampling " + toString(size) + " from " + toString(rabund->getNumSeqs()) + "."); m->mothurOutEndLine();
1222
1223                 string thisOutputDir = outputDir;
1224                 if (outputDir == "") {  thisOutputDir += m->hasPath(rabundfile);  }
1225                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(rabundfile)) + "subsample" + m->getExtension(rabundfile);
1226
1227                 ofstream out;
1228                 m->openOutputFile(outputFileName, out);
1229                 outputTypes["rabund"].push_back(outputFileName);  outputNames.push_back(outputFileName);
1230
1231                 //as long as you are not at the end of the file or done wih the lines you want
1232                 while((rabund != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
1233                         if (m->control_pressed) {  delete input; delete rabund; out.close(); return 0;  }
1234
1235                         if(allLines == 1 || labels.count(rabund->getLabel()) == 1){
1236
1237                                 m->mothurOut(rabund->getLabel()); m->mothurOutEndLine();
1238
1239                                 processRabund(rabund, out);
1240
1241                                 processedLabels.insert(rabund->getLabel());
1242                                 userLabels.erase(rabund->getLabel());
1243                         }
1244
1245                         if ((m->anyLabelsToProcess(rabund->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
1246                                 string saveLabel = rabund->getLabel();
1247
1248                                 delete rabund;
1249
1250                                 rabund = input->getRAbundVector(lastLabel);
1251                                 m->mothurOut(rabund->getLabel()); m->mothurOutEndLine();
1252
1253                                 processRabund(rabund, out);
1254
1255                                 processedLabels.insert(rabund->getLabel());
1256                                 userLabels.erase(rabund->getLabel());
1257
1258                                 //restore real lastlabel to save below
1259                                 rabund->setLabel(saveLabel);
1260                         }
1261
1262                         lastLabel = rabund->getLabel();
1263
1264                         //prevent memory leak
1265                         delete rabund; rabund = NULL;
1266
1267                         //get next line to process
1268                         rabund = input->getRAbundVector();
1269                 }
1270
1271
1272                 if (m->control_pressed) {  out.close(); return 0;  }
1273
1274                 //output error messages about any remaining user labels
1275                 set<string>::iterator it;
1276                 bool needToRun = false;
1277                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
1278                         m->mothurOut("Your file does not include the label " + *it);
1279                         if (processedLabels.count(lastLabel) != 1) {
1280                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
1281                                 needToRun = true;
1282                         }else {
1283                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
1284                         }
1285                 }
1286
1287                 //run last label if you need to
1288                 if (needToRun == true)  {
1289                         if (rabund != NULL) { delete rabund; }
1290
1291                         rabund = input->getRAbundVector(lastLabel);
1292
1293                         m->mothurOut(rabund->getLabel()); m->mothurOutEndLine();
1294
1295                         processRabund(rabund, out);
1296
1297                         delete rabund;
1298                 }
1299
1300                 delete input;
1301                 out.close();
1302
1303                 return 0;
1304
1305         }
1306         catch(exception& e) {
1307                 m->errorOut(e, "SubSampleCommand", "getSubSampleRabund");
1308                 exit(1);
1309         }
1310 }
1311 //**********************************************************************************************************************
1312 int SubSampleCommand::processRabund(RAbundVector*& rabund, ofstream& out) {
1313         try {
1314
1315                 int numBins = rabund->getNumBins();
1316                 int thisSize = rabund->getNumSeqs();
1317
1318                 if (thisSize != size) {
1319
1320                         OrderVector* order = new OrderVector();
1321                         for(int p=0;p<numBins;p++){
1322                                 for(int j=0;j<rabund->get(p);j++){
1323                                         order->push_back(p);
1324                                 }
1325                         }
1326                         random_shuffle(order->begin(), order->end());
1327
1328                         RAbundVector* temp = new RAbundVector(numBins);
1329                         temp->setLabel(rabund->getLabel());
1330
1331                         delete rabund;
1332                         rabund = temp;
1333
1334                         for (int j = 0; j < size; j++) {
1335
1336                                 if (m->control_pressed) { delete order; return 0; }
1337
1338                                 int bin = order->get(j);
1339
1340                                 int abund = rabund->get(bin);
1341                                 rabund->set(bin, (abund+1));
1342                         }
1343
1344                         delete order;
1345                 }
1346
1347                 if (m->control_pressed) { return 0; }
1348
1349                 rabund->print(out);
1350
1351                 return 0;
1352
1353         }
1354         catch(exception& e) {
1355                 m->errorOut(e, "SubSampleCommand", "processRabund");
1356                 exit(1);
1357         }
1358 }
1359 //**********************************************************************************************************************
1360 int SubSampleCommand::getSubSampleSabund() {
1361         try {
1362
1363                 InputData* input = new InputData(sabundfile, "sabund");
1364                 SAbundVector* sabund = input->getSAbundVector();
1365                 string lastLabel = sabund->getLabel();
1366
1367                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
1368                 set<string> processedLabels;
1369                 set<string> userLabels = labels;
1370
1371                 if (size == 0) { //user has not set size, set size = 10%
1372                         size = int((sabund->getNumSeqs()) * 0.10);
1373                 }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; }
1374
1375
1376                 m->mothurOut("Sampling " + toString(size) + " from " + toString(sabund->getNumSeqs()) + "."); m->mothurOutEndLine();
1377
1378                 string thisOutputDir = outputDir;
1379                 if (outputDir == "") {  thisOutputDir += m->hasPath(sabundfile);  }
1380                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(sabundfile)) + "subsample" + m->getExtension(sabundfile);
1381
1382                 ofstream out;
1383                 m->openOutputFile(outputFileName, out);
1384                 outputTypes["sabund"].push_back(outputFileName);  outputNames.push_back(outputFileName);
1385
1386
1387                 //as long as you are not at the end of the file or done wih the lines you want
1388                 while((sabund != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
1389                         if (m->control_pressed) {  delete input; delete sabund; out.close(); return 0;  }
1390
1391                         if(allLines == 1 || labels.count(sabund->getLabel()) == 1){
1392
1393                                 m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
1394
1395                                 processSabund(sabund, out);
1396
1397                                 processedLabels.insert(sabund->getLabel());
1398                                 userLabels.erase(sabund->getLabel());
1399                         }
1400
1401                         if ((m->anyLabelsToProcess(sabund->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
1402                                 string saveLabel = sabund->getLabel();
1403
1404                                 delete sabund;
1405
1406                                 sabund = input->getSAbundVector(lastLabel);
1407                                 m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
1408
1409                                 processSabund(sabund, out);
1410
1411                                 processedLabels.insert(sabund->getLabel());
1412                                 userLabels.erase(sabund->getLabel());
1413
1414                                 //restore real lastlabel to save below
1415                                 sabund->setLabel(saveLabel);
1416                         }
1417
1418                         lastLabel = sabund->getLabel();
1419
1420                         //prevent memory leak
1421                         delete sabund; sabund = NULL;
1422
1423                         //get next line to process
1424                         sabund = input->getSAbundVector();
1425                 }
1426
1427
1428                 if (m->control_pressed) {  out.close(); return 0;  }
1429
1430                 //output error messages about any remaining user labels
1431                 set<string>::iterator it;
1432                 bool needToRun = false;
1433                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
1434                         m->mothurOut("Your file does not include the label " + *it);
1435                         if (processedLabels.count(lastLabel) != 1) {
1436                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
1437                                 needToRun = true;
1438                         }else {
1439                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
1440                         }
1441                 }
1442
1443                 //run last label if you need to
1444                 if (needToRun == true)  {
1445                         if (sabund != NULL) { delete sabund; }
1446
1447                         sabund = input->getSAbundVector(lastLabel);
1448
1449                         m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
1450
1451                         processSabund(sabund, out);
1452
1453                         delete sabund;
1454                 }
1455
1456                 delete input;
1457                 out.close();
1458
1459                 return 0;
1460
1461         }
1462         catch(exception& e) {
1463                 m->errorOut(e, "SubSampleCommand", "getSubSampleSabund");
1464                 exit(1);
1465         }
1466 }
1467 //**********************************************************************************************************************
1468 int SubSampleCommand::processSabund(SAbundVector*& sabund, ofstream& out) {
1469         try {
1470
1471                 RAbundVector* rabund = new RAbundVector();
1472                 *rabund = sabund->getRAbundVector();
1473
1474                 int numBins = rabund->getNumBins();
1475                 int thisSize = rabund->getNumSeqs();
1476
1477                 if (thisSize != size) {
1478
1479                         OrderVector* order = new OrderVector();
1480                         for(int p=0;p<numBins;p++){
1481                                 for(int j=0;j<rabund->get(p);j++){
1482                                         order->push_back(p);
1483                                 }
1484                         }
1485                         random_shuffle(order->begin(), order->end());
1486
1487                         RAbundVector* temp = new RAbundVector(numBins);
1488                         temp->setLabel(rabund->getLabel());
1489
1490                         delete rabund;
1491                         rabund = temp;
1492
1493                         for (int j = 0; j < size; j++) {
1494
1495                                 if (m->control_pressed) { delete order; return 0; }
1496
1497                                 int bin = order->get(j);
1498
1499                                 int abund = rabund->get(bin);
1500                                 rabund->set(bin, (abund+1));
1501                         }
1502
1503                         delete order;
1504                 }
1505
1506                 if (m->control_pressed) { return 0; }
1507
1508                 delete sabund;
1509                 sabund = new SAbundVector();
1510                 *sabund = rabund->getSAbundVector();
1511                 delete rabund;
1512
1513                 sabund->print(out);
1514
1515                 return 0;
1516
1517         }
1518         catch(exception& e) {
1519                 m->errorOut(e, "SubSampleCommand", "processSabund");
1520                 exit(1);
1521         }
1522 }
1523 //**********************************************************************************************************************
1524 int SubSampleCommand::eliminateZeroOTUS(vector<SharedRAbundVector*>& thislookup) {
1525         try {
1526
1527                 vector<SharedRAbundVector*> newLookup;
1528                 for (int i = 0; i < thislookup.size(); i++) {
1529                         SharedRAbundVector* temp = new SharedRAbundVector();
1530                         temp->setLabel(thislookup[i]->getLabel());
1531                         temp->setGroup(thislookup[i]->getGroup());
1532                         newLookup.push_back(temp);
1533                 }
1534
1535                 //for each bin
1536                 vector<string> newBinLabels;
1537                 string snumBins = toString(thislookup[0]->getNumBins());
1538                 for (int i = 0; i < thislookup[0]->getNumBins(); i++) {
1539                         if (m->control_pressed) { for (int j = 0; j < newLookup.size(); j++) {  delete newLookup[j];  } return 0; }
1540
1541                         //look at each sharedRabund and make sure they are not all zero
1542                         bool allZero = true;
1543                         for (int j = 0; j < thislookup.size(); j++) {
1544                                 if (thislookup[j]->getAbundance(i) != 0) { allZero = false;  break;  }
1545                         }
1546
1547                         //if they are not all zero add this bin
1548                         if (!allZero) {
1549                                 for (int j = 0; j < thislookup.size(); j++) {
1550                                         newLookup[j]->push_back(thislookup[j]->getAbundance(i), thislookup[j]->getGroup());
1551                                 }
1552                                 //if there is a bin label use it otherwise make one
1553                                 string binLabel = "Otu";
1554                                 string sbinNumber = toString(i+1);
1555                                 if (sbinNumber.length() < snumBins.length()) {
1556                                         int diff = snumBins.length() - sbinNumber.length();
1557                                         for (int h = 0; h < diff; h++) { binLabel += "0"; }
1558                                 }
1559                                 binLabel += sbinNumber;
1560                                 if (i < m->currentBinLabels.size()) {  binLabel = m->currentBinLabels[i]; }
1561
1562                                 newBinLabels.push_back(binLabel);
1563                         }
1564                 }
1565
1566                 for (int j = 0; j < thislookup.size(); j++) {  delete thislookup[j];  }
1567                 thislookup.clear();
1568
1569                 thislookup = newLookup;
1570                 m->currentBinLabels = newBinLabels;
1571
1572                 return 0;
1573
1574         }
1575         catch(exception& e) {
1576                 m->errorOut(e, "SubSampleCommand", "eliminateZeroOTUS");
1577                 exit(1);
1578         }
1579 }
1580
1581 //**********************************************************************************************************************
1582
1583
1584