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                         m->mothurCalling = true;
 535
 536                         Command* uniqueCommand = new DeconvoluteCommand(inputString);
 537                         uniqueCommand->execute();
 538
 539                         map<string, vector<string> > filenames = uniqueCommand->getOutputFiles();
 540
 541                         delete uniqueCommand;
 542                         m->mothurCalling = false;
 543
 544                         outputTypes["name"].push_back(filenames["name"][0]);  outputNames.push_back(filenames["name"][0]);
 545                         m->mothurRemove(outputFileName);
 546                         outputFileName = filenames["fasta"][0];
 547
 548                         m->mothurOut("/******************************************/"); m->mothurOutEndLine();
 549
 550                         m->mothurOut("Done."); m->mothurOutEndLine();
 551                 }
 552
 553                 outputTypes["fasta"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 554
 555                 //if a groupfile is provided read through the group file only outputting the names on the subsample list
 556                 if (groupfile != "") {
 557
 558                         string groupOutputDir = outputDir;
 559                         if (outputDir == "") {  groupOutputDir += m->hasPath(groupfile);  }
 560                         string groupOutputFileName = groupOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "subsample" + m->getExtension(groupfile);
 561
 562                         ofstream outGroup;
 563                         m->openOutputFile(groupOutputFileName, outGroup);
 564                         outputTypes["group"].push_back(groupOutputFileName);  outputNames.push_back(groupOutputFileName);
 565
 566                         ifstream inGroup;
 567                         m->openInputFile(groupfile, inGroup);
 568                         string name, group;
 569
 570                         while(!inGroup.eof()){
 571
 572                                 if (m->control_pressed) { inGroup.close(); outGroup.close(); return 0; }
 573
 574                                 inGroup >> name;        m->gobble(inGroup);                     //read from first column
 575                                 inGroup >> group;                       //read from second column
 576
 577                                 //if this name is in the accnos file
 578                                 if (subset.count(name) != 0) {
 579                                         outGroup << name << '\t' << group << endl;
 580                                         subset.erase(name);
 581                                 }
 582
 583                                 m->gobble(inGroup);
 584                         }
 585                         inGroup.close();
 586                         outGroup.close();
 587
 588                         //sanity check
 589                         if (subset.size() != 0) {
 590                                 m->mothurOut("Your groupfile does not match your fasta file."); m->mothurOutEndLine();
 591                                 for (set<string>::iterator it = subset.begin(); it != subset.end(); it++) {
 592                                         m->mothurOut("[ERROR]: " + *it + " is missing from your groupfile."); m->mothurOutEndLine();
 593                                 }
 594                         }
 595                 }
 596
 597                 return 0;
 598
 599         }
 600         catch(exception& e) {
 601                 m->errorOut(e, "SubSampleCommand", "getSubSampleFasta");
 602                 exit(1);
 603         }
 604 }
 605 //**********************************************************************************************************************
 606 int SubSampleCommand::getNames() {
 607         try {
 608
 609                 ifstream in;
 610                 m->openInputFile(fastafile, in);
 611
 612                 string thisname;
 613                 while(!in.eof()){
 614
 615                         if (m->control_pressed) { in.close(); return 0; }
 616
 617                         Sequence currSeq(in);
 618                         thisname = currSeq.getName();
 619
 620                         if (thisname != "") {
 621                                 vector<string> temp; temp.push_back(thisname);
 622                                 nameMap[thisname] = temp;
 623                                 names.push_back(thisname);
 624                         }
 625                         m->gobble(in);
 626                 }
 627                 in.close();
 628
 629                 return 0;
 630
 631         }
 632         catch(exception& e) {
 633                 m->errorOut(e, "SubSampleCommand", "getNames");
 634                 exit(1);
 635         }
 636 }
 637 //**********************************************************************************************************************
 638 int SubSampleCommand::readNames() {
 639         try {
 640
 641                 ifstream in;
 642                 m->openInputFile(namefile, in);
 643
 644                 string thisname, repnames;
 645                 map<string, vector<string> >::iterator it;
 646
 647                 while(!in.eof()){
 648
 649                         if (m->control_pressed) { in.close(); return 0; }
 650
 651                         in >> thisname;         m->gobble(in);          //read from first column
 652                         in >> repnames;                 //read from second column
 653
 654                         it = nameMap.find(thisname);
 655                         if (it == nameMap.end()) {
 656
 657                                 vector<string> splitRepNames;
 658                                 m->splitAtComma(repnames, splitRepNames);
 659
 660                                 nameMap[thisname] = splitRepNames;
 661                                 for (int i = 0; i < splitRepNames.size(); i++) { names.push_back(splitRepNames[i]); }
 662
 663                         }else{  m->mothurOut(thisname + " is already in namesfile. I will use first definition."); m->mothurOutEndLine();  }
 664
 665                         m->gobble(in);
 666                 }
 667                 in.close();
 668
 669                 return 0;
 670
 671         }
 672         catch(exception& e) {
 673                 m->errorOut(e, "SubSampleCommand", "readNames");
 674                 exit(1);
 675         }
 676 }
 677 //**********************************************************************************************************************
 678 int SubSampleCommand::getSubSampleShared() {
 679         try {
 680
 681                 InputData* input = new InputData(sharedfile, "sharedfile");
 682                 vector<SharedRAbundVector*> lookup = input->getSharedRAbundVectors();
 683                 string lastLabel = lookup[0]->getLabel();
 684
 685                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 686                 set<string> processedLabels;
 687                 set<string> userLabels = labels;
 688
 689                 if (size == 0) { //user has not set size, set size = smallest samples size
 690                         size = lookup[0]->getNumSeqs();
 691                         for (int i = 1; i < lookup.size(); i++) {
 692                                 int thisSize = lookup[i]->getNumSeqs();
 693
 694                                 if (thisSize < size) {  size = thisSize;        }
 695                         }
 696                 }else {
 697                         m->clearGroups();
 698                         Groups.clear();
 699                         vector<SharedRAbundVector*> temp;
 700                         for (int i = 0; i < lookup.size(); i++) {
 701                                 if (lookup[i]->getNumSeqs() < size) {
 702                                         m->mothurOut(lookup[i]->getGroup() + " contains " + toString(lookup[i]->getNumSeqs()) + ". Eliminating."); m->mothurOutEndLine();
 703                                         delete lookup[i];
 704                                 }else {
 705                                         Groups.push_back(lookup[i]->getGroup());
 706                                         temp.push_back(lookup[i]);
 707                                 }
 708                         }
 709                         lookup = temp;
 710                         m->setGroups(Groups);
 711                 }
 712
 713                 if (lookup.size() == 0) {  m->mothurOut("The size you selected is too large, skipping shared file."); m->mothurOutEndLine(); delete input; return 0; }
 714
 715                 m->mothurOut("Sampling " + toString(size) + " from each group."); m->mothurOutEndLine();
 716
 717                 //as long as you are not at the end of the file or done wih the lines you want
 718                 while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 719                         if (m->control_pressed) {  delete input; for (int i = 0; i < lookup.size(); i++) {  delete lookup[i]; lookup[i] = NULL; }  return 0;  }
 720
 721                         if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){
 722
 723                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 724
 725                                 processShared(lookup);
 726
 727                                 processedLabels.insert(lookup[0]->getLabel());
 728                                 userLabels.erase(lookup[0]->getLabel());
 729                         }
 730
 731                         if ((m->anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 732                                 string saveLabel = lookup[0]->getLabel();
 733
 734                                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 735
 736                                 lookup = input->getSharedRAbundVectors(lastLabel);
 737                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 738
 739                                 processShared(lookup);
 740
 741                                 processedLabels.insert(lookup[0]->getLabel());
 742                                 userLabels.erase(lookup[0]->getLabel());
 743
 744                                 //restore real lastlabel to save below
 745                                 lookup[0]->setLabel(saveLabel);
 746                         }
 747
 748                         lastLabel = lookup[0]->getLabel();
 749                         //prevent memory leak
 750                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i]; lookup[i] = NULL; }
 751
 752                         //get next line to process
 753                         lookup = input->getSharedRAbundVectors();
 754                 }
 755
 756
 757                 if (m->control_pressed) {   return 0;  }
 758
 759                 //output error messages about any remaining user labels
 760                 set<string>::iterator it;
 761                 bool needToRun = false;
 762                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 763                         m->mothurOut("Your file does not include the label " + *it);
 764                         if (processedLabels.count(lastLabel) != 1) {
 765                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 766                                 needToRun = true;
 767                         }else {
 768                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 769                         }
 770                 }
 771
 772                 //run last label if you need to
 773                 if (needToRun == true)  {
 774                         for (int i = 0; i < lookup.size(); i++) { if (lookup[i] != NULL) { delete lookup[i]; } }
 775                         lookup = input->getSharedRAbundVectors(lastLabel);
 776
 777                         m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 778
 779                         processShared(lookup);
 780
 781                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 782                 }
 783
 784                 delete input;
 785
 786                 return 0;
 787
 788         }
 789         catch(exception& e) {
 790                 m->errorOut(e, "SubSampleCommand", "getSubSampleShared");
 791                 exit(1);
 792         }
 793 }
 794 //**********************************************************************************************************************
 795 int SubSampleCommand::processShared(vector<SharedRAbundVector*>& thislookup) {
 796         try {
 797
 798                 //save mothurOut's binLabels to restore for next label
 799                 vector<string> saveBinLabels = m->currentBinLabels;
 800
 801                 string thisOutputDir = outputDir;
 802                 if (outputDir == "") {  thisOutputDir += m->hasPath(sharedfile);  }
 803                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(sharedfile)) + thislookup[0]->getLabel() + ".subsample" + m->getExtension(sharedfile);
 804
 805
 806                 ofstream out;
 807                 m->openOutputFile(outputFileName, out);
 808                 outputTypes["shared"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 809
 810                 int numBins = thislookup[0]->getNumBins();
 811                 for (int i = 0; i < thislookup.size(); i++) {
 812                         int thisSize = thislookup[i]->getNumSeqs();
 813
 814                         if (thisSize != size) {
 815
 816                                 string thisgroup = thislookup[i]->getGroup();
 817
 818                                 OrderVector* order = new OrderVector();
 819                                 for(int p=0;p<numBins;p++){
 820                                         for(int j=0;j<thislookup[i]->getAbundance(p);j++){
 821                                                 order->push_back(p);
 822                                         }
 823                                 }
 824                                 random_shuffle(order->begin(), order->end());
 825
 826                                 SharedRAbundVector* temp = new SharedRAbundVector(numBins);
 827                                 temp->setLabel(thislookup[i]->getLabel());
 828                                 temp->setGroup(thislookup[i]->getGroup());
 829
 830                                 delete thislookup[i];
 831                                 thislookup[i] = temp;
 832
 833
 834                                 for (int j = 0; j < size; j++) {
 835
 836                                         if (m->control_pressed) { delete order; out.close(); return 0; }
 837
 838                                         //get random number to sample from order between 0 and thisSize-1.
 839                                         //don't need this because of the random shuffle above
 840                                         //int myrand = int((float)(thisSize) * (float)(rand()) / ((float)RAND_MAX+1.0));
 841
 842                                         int bin = order->get(j);
 843
 844                                         int abund = thislookup[i]->getAbundance(bin);
 845                                         thislookup[i]->set(bin, (abund+1), thisgroup);
 846                                 }
 847                                 delete order;
 848                         }
 849                 }
 850
 851                 //subsampling may have created some otus with no sequences in them
 852                 eliminateZeroOTUS(thislookup);
 853
 854                 if (m->control_pressed) { out.close(); return 0; }
 855
 856                 thislookup[0]->printHeaders(out);
 857
 858                 for (int i = 0; i < thislookup.size(); i++) {
 859                         out << thislookup[i]->getLabel() << '\t' << thislookup[i]->getGroup() << '\t';
 860                         thislookup[i]->print(out);
 861                 }
 862
 863                 out.close();
 864
 865                 //save mothurOut's binLabels to restore for next label
 866                 m->currentBinLabels = saveBinLabels;
 867
 868                 return 0;
 869
 870         }
 871         catch(exception& e) {
 872                 m->errorOut(e, "SubSampleCommand", "processShared");
 873                 exit(1);
 874         }
 875 }
 876 //**********************************************************************************************************************
 877 int SubSampleCommand::getSubSampleList() {
 878         try {
 879
 880                 string thisOutputDir = outputDir;
 881                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 882                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "subsample" + m->getExtension(listfile);
 883
 884                 ofstream out;
 885                 m->openOutputFile(outputFileName, out);
 886                 outputTypes["list"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 887
 888                 InputData* input = new InputData(listfile, "list");
 889                 ListVector* list = input->getListVector();
 890                 string lastLabel = list->getLabel();
 891
 892                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 893                 set<string> processedLabels;
 894                 set<string> userLabels = labels;
 895
 896                 ofstream outGroup;
 897                 GroupMap* groupMap;
 898                 if (groupfile != "") {
 899
 900                         groupMap = new GroupMap(groupfile);
 901                         groupMap->readMap();
 902
 903                         //takes care of user setting groupNames that are invalid or setting groups=all
 904                         SharedUtil* util = new SharedUtil();
 905                         vector<string> namesGroups = groupMap->getNamesOfGroups();
 906                         util->setGroups(Groups, namesGroups);
 907                         delete util;
 908
 909                         //create outputfiles
 910                         string groupOutputDir = outputDir;
 911                         if (outputDir == "") {  groupOutputDir += m->hasPath(groupfile);  }
 912                         string groupOutputFileName = groupOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "subsample" + m->getExtension(groupfile);
 913
 914                         m->openOutputFile(groupOutputFileName, outGroup);
 915                         outputTypes["group"].push_back(groupOutputFileName);  outputNames.push_back(groupOutputFileName);
 916
 917                         //file mismatch quit
 918                         if (list->getNumSeqs() != groupMap->getNumSeqs()) {
 919                                 m->mothurOut("[ERROR]: your list file contains " + toString(list->getNumSeqs()) + " sequences, and your groupfile contains " + toString(groupMap->getNumSeqs()) + ", please correct.");
 920                                 m->mothurOutEndLine();
 921                                 delete groupMap;
 922                                 delete list;
 923                                 delete input;
 924                                 out.close();
 925                                 outGroup.close();
 926                                 return 0;
 927                         }
 928                 }
 929
 930                 //make sure that if your picked groups size is not too big
 931                 if (persample) {
 932                         if (size == 0) { //user has not set size, set size = smallest samples size
 933                                 size = groupMap->getNumSeqs(Groups[0]);
 934                                 for (int i = 1; i < Groups.size(); i++) {
 935                                         int thisSize = groupMap->getNumSeqs(Groups[i]);
 936
 937                                         if (thisSize < size) {  size = thisSize;        }
 938                                 }
 939                         }else { //make sure size is not too large
 940                                 vector<string> newGroups;
 941                                 for (int i = 0; i < Groups.size(); i++) {
 942                                         int thisSize = groupMap->getNumSeqs(Groups[i]);
 943
 944                                         if (thisSize >= size) { newGroups.push_back(Groups[i]); }
 945                                         else {  m->mothurOut("You have selected a size that is larger than " + Groups[i] + " number of sequences, removing " + Groups[i] + "."); m->mothurOutEndLine(); }
 946                                 }
 947                                 Groups = newGroups;
 948                         }
 949
 950                         m->mothurOut("Sampling " + toString(size) + " from each group."); m->mothurOutEndLine();
 951                 }else{
 952                         if (pickedGroups) {
 953                                 int total = 0;
 954                                 for(int i = 0; i < Groups.size(); i++) {
 955                                         total += groupMap->getNumSeqs(Groups[i]);
 956                                 }
 957
 958                                 if (size == 0) { //user has not set size, set size = 10% samples size
 959                                         size = int (total * 0.10);
 960                                 }
 961
 962                                 if (total < size) {
 963                                         m->mothurOut("Your size is too large for the number of groups you selected. Adjusting to " + toString(int (total * 0.10)) + "."); m->mothurOutEndLine();
 964                                         size = int (total * 0.10);
 965                                 }
 966
 967                                 m->mothurOut("Sampling " + toString(size) + " from " + toString(total) + "."); m->mothurOutEndLine();
 968                         }else{
 969
 970                                 if (size == 0) { //user has not set size, set size = 10% samples size
 971                                         size = int (list->getNumSeqs() * 0.10);
 972                                 }
 973
 974                                 int thisSize = list->getNumSeqs();
 975                                 if (size > thisSize) { m->mothurOut("Your list file only contains " + toString(thisSize) + " sequences. Setting size to " + toString(thisSize) + "."); m->mothurOutEndLine();
 976                                         size = thisSize;
 977                                 }
 978
 979                                 m->mothurOut("Sampling " + toString(size) + " from " + toString(list->getNumSeqs()) + "."); m->mothurOutEndLine();
 980                         }
 981                 }
 982
 983
 984                 //fill names
 985                 for (int i = 0; i < list->getNumBins(); i++) {
 986                         string binnames = list->get(i);
 987
 988                         //parse names
 989                         string individual = "";
 990                         int length = binnames.length();
 991                         for(int j=0;j<length;j++){
 992                                 if(binnames[j] == ','){
 993
 994                                         if (groupfile != "") { //if there is a groupfile given fill in group info
 995                                                 string group = groupMap->getGroup(individual);
 996                                                 if (group == "not found") { m->mothurOut("[ERROR]: " + individual + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
 997
 998                                                 if (pickedGroups) { //if hte user picked groups, we only want to keep the names of sequences from those groups
 999                                                         if (m->inUsersGroups(group, Groups)) {
1000                                                                 names.push_back(individual);
1001                                                         }
1002                                                 }else{
1003                                                         names.push_back(individual);
1004                                                 }
1005                                         }else{ //save everyone, group
1006                                                 names.push_back(individual);
1007                                         }
1008                                         individual = "";
1009                                 }
1010                                 else{
1011                                         individual += binnames[j];
1012                                 }
1013                         }
1014                         //save last name
1015                         if (groupfile != "") { //if there is a groupfile given fill in group info
1016                                 string group = groupMap->getGroup(individual);
1017                                 if (group == "not found") { m->mothurOut("[ERROR]: " + individual + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
1018
1019                                 if (pickedGroups) { //if hte user picked groups, we only want to keep the names of sequences from those groups
1020                                         if (m->inUsersGroups(group, Groups)) {
1021                                                 names.push_back(individual);
1022                                         }
1023                                 }else{
1024                                         names.push_back(individual);
1025                                 }
1026                         }else{ //save everyone, group
1027                                 names.push_back(individual);
1028                         }
1029                 }
1030
1031                 random_shuffle(names.begin(), names.end());
1032
1033                 //randomly select a subset of those names to include in the subsample
1034                 set<string> subset; //dont want repeat sequence names added
1035                 if (persample) {
1036                         //initialize counts
1037                         map<string, int> groupCounts;
1038                         map<string, int>::iterator itGroupCounts;
1039                         for (int i = 0; i < Groups.size(); i++) { groupCounts[Groups[i]] = 0; }
1040
1041                         for (int j = 0; j < names.size(); j++) {
1042
1043                                 if (m->control_pressed) { return 0; }
1044
1045                                 string group = groupMap->getGroup(names[j]);
1046                                 if (group == "not found") { m->mothurOut("[ERROR]: " + names[j] + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
1047                                 else{
1048                                         itGroupCounts = groupCounts.find(group);
1049                                         if (itGroupCounts != groupCounts.end()) {
1050                                                 if (groupCounts[group] < size) {        subset.insert(names[j]);        groupCounts[group]++; }
1051                                         }
1052                                 }
1053                         }
1054                 }else{
1055                         for (int j = 0; j < size; j++) {
1056
1057                                 if (m->control_pressed) { break; }
1058
1059                                 subset.insert(names[j]);
1060                         }
1061                 }
1062
1063                 if (groupfile != "") {
1064                         //write out new groupfile
1065                         for (set<string>::iterator it = subset.begin(); it != subset.end(); it++) {
1066                                 string group = groupMap->getGroup(*it);
1067                                 if (group == "not found") { group = "NOTFOUND"; }
1068
1069                                 outGroup << *it << '\t' << group << endl;
1070                         }
1071                         outGroup.close(); delete groupMap;
1072                 }
1073
1074
1075                 //as long as you are not at the end of the file or done wih the lines you want
1076                 while((list != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
1077
1078                         if (m->control_pressed) {  delete list; delete input; out.close();  return 0;  }
1079
1080                         if(allLines == 1 || labels.count(list->getLabel()) == 1){
1081
1082                                 m->mothurOut(list->getLabel()); m->mothurOutEndLine();
1083
1084                                 processList(list, out, subset);
1085
1086                                 processedLabels.insert(list->getLabel());
1087                                 userLabels.erase(list->getLabel());
1088                         }
1089
1090                         if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
1091                                 string saveLabel = list->getLabel();
1092
1093                                 delete list;
1094
1095                                 list = input->getListVector(lastLabel);
1096                                 m->mothurOut(list->getLabel()); m->mothurOutEndLine();
1097
1098                                 processList(list, out, subset);
1099
1100                                 processedLabels.insert(list->getLabel());
1101                                 userLabels.erase(list->getLabel());
1102
1103                                 //restore real lastlabel to save below
1104                                 list->setLabel(saveLabel);
1105                         }
1106
1107                         lastLabel = list->getLabel();
1108
1109                         delete list; list = NULL;
1110
1111                         //get next line to process
1112                         list = input->getListVector();
1113                 }
1114
1115
1116                 if (m->control_pressed) {  if (list != NULL) { delete list; } delete input; out.close(); return 0;  }
1117
1118                 //output error messages about any remaining user labels
1119                 set<string>::iterator it;
1120                 bool needToRun = false;
1121                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
1122                         m->mothurOut("Your file does not include the label " + *it);
1123                         if (processedLabels.count(lastLabel) != 1) {
1124                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
1125                                 needToRun = true;
1126                         }else {
1127                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
1128                         }
1129                 }
1130
1131                 //run last label if you need to
1132                 if (needToRun == true)  {
1133                         if (list != NULL) { delete list; }
1134
1135                         list = input->getListVector(lastLabel);
1136
1137                         m->mothurOut(list->getLabel()); m->mothurOutEndLine();
1138
1139                         processList(list, out, subset);
1140
1141                         delete list; list = NULL;
1142                 }
1143
1144                 out.close();
1145                 if (list != NULL) { delete list; }
1146                 delete input;
1147
1148                 return 0;
1149
1150         }
1151         catch(exception& e) {
1152                 m->errorOut(e, "SubSampleCommand", "getSubSampleList");
1153                 exit(1);
1154         }
1155 }
1156 //**********************************************************************************************************************
1157 int SubSampleCommand::processList(ListVector*& list, ofstream& out, set<string>& subset) {
1158         try {
1159
1160                 int numBins = list->getNumBins();
1161
1162                 ListVector* temp = new ListVector();
1163                 temp->setLabel(list->getLabel());
1164
1165                 for (int i = 0; i < numBins; i++) {
1166
1167                         if (m->control_pressed) { break; }
1168
1169                         string binnames = list->get(i);
1170
1171                         //parse names
1172                         string individual = "";
1173                         string newNames = "";
1174                         int length = binnames.length();
1175                         for(int j=0;j<length;j++){
1176                                 if(binnames[j] == ','){
1177                                         if (subset.count(individual) != 0) {  newNames += individual + ",";  }
1178                                         individual = "";
1179                                 }else{
1180                                         individual += binnames[j];
1181                                 }
1182                         }
1183                         if (subset.count(individual) != 0) {  newNames += individual + ",";  }
1184
1185
1186                         //if there are names in this bin add to new list
1187                         if (newNames != "") {
1188                                 newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
1189                                 temp->push_back(newNames);
1190                         }
1191                 }
1192
1193                 delete list;
1194                 list = temp;
1195
1196                 if (m->control_pressed) { return 0; }
1197
1198                 list->print(out);
1199
1200                 return 0;
1201
1202         }
1203         catch(exception& e) {
1204                 m->errorOut(e, "SubSampleCommand", "processList");
1205                 exit(1);
1206         }
1207 }
1208 //**********************************************************************************************************************
1209 int SubSampleCommand::getSubSampleRabund() {
1210         try {
1211                 InputData* input = new InputData(rabundfile, "rabund");
1212                 RAbundVector* rabund = input->getRAbundVector();
1213                 string lastLabel = rabund->getLabel();
1214
1215                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
1216                 set<string> processedLabels;
1217                 set<string> userLabels = labels;
1218
1219                 if (size == 0) { //user has not set size, set size = 10%
1220                         size = int((rabund->getNumSeqs()) * 0.10);
1221                 }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; }
1222
1223                 m->mothurOut("Sampling " + toString(size) + " from " + toString(rabund->getNumSeqs()) + "."); m->mothurOutEndLine();
1224
1225                 string thisOutputDir = outputDir;
1226                 if (outputDir == "") {  thisOutputDir += m->hasPath(rabundfile);  }
1227                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(rabundfile)) + "subsample" + m->getExtension(rabundfile);
1228
1229                 ofstream out;
1230                 m->openOutputFile(outputFileName, out);
1231                 outputTypes["rabund"].push_back(outputFileName);  outputNames.push_back(outputFileName);
1232
1233                 //as long as you are not at the end of the file or done wih the lines you want
1234                 while((rabund != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
1235                         if (m->control_pressed) {  delete input; delete rabund; out.close(); return 0;  }
1236
1237                         if(allLines == 1 || labels.count(rabund->getLabel()) == 1){
1238
1239                                 m->mothurOut(rabund->getLabel()); m->mothurOutEndLine();
1240
1241                                 processRabund(rabund, out);
1242
1243                                 processedLabels.insert(rabund->getLabel());
1244                                 userLabels.erase(rabund->getLabel());
1245                         }
1246
1247                         if ((m->anyLabelsToProcess(rabund->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
1248                                 string saveLabel = rabund->getLabel();
1249
1250                                 delete rabund;
1251
1252                                 rabund = input->getRAbundVector(lastLabel);
1253                                 m->mothurOut(rabund->getLabel()); m->mothurOutEndLine();
1254
1255                                 processRabund(rabund, out);
1256
1257                                 processedLabels.insert(rabund->getLabel());
1258                                 userLabels.erase(rabund->getLabel());
1259
1260                                 //restore real lastlabel to save below
1261                                 rabund->setLabel(saveLabel);
1262                         }
1263
1264                         lastLabel = rabund->getLabel();
1265
1266                         //prevent memory leak
1267                         delete rabund; rabund = NULL;
1268
1269                         //get next line to process
1270                         rabund = input->getRAbundVector();
1271                 }
1272
1273
1274                 if (m->control_pressed) {  out.close(); return 0;  }
1275
1276                 //output error messages about any remaining user labels
1277                 set<string>::iterator it;
1278                 bool needToRun = false;
1279                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
1280                         m->mothurOut("Your file does not include the label " + *it);
1281                         if (processedLabels.count(lastLabel) != 1) {
1282                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
1283                                 needToRun = true;
1284                         }else {
1285                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
1286                         }
1287                 }
1288
1289                 //run last label if you need to
1290                 if (needToRun == true)  {
1291                         if (rabund != NULL) { delete rabund; }
1292
1293                         rabund = input->getRAbundVector(lastLabel);
1294
1295                         m->mothurOut(rabund->getLabel()); m->mothurOutEndLine();
1296
1297                         processRabund(rabund, out);
1298
1299                         delete rabund;
1300                 }
1301
1302                 delete input;
1303                 out.close();
1304
1305                 return 0;
1306
1307         }
1308         catch(exception& e) {
1309                 m->errorOut(e, "SubSampleCommand", "getSubSampleRabund");
1310                 exit(1);
1311         }
1312 }
1313 //**********************************************************************************************************************
1314 int SubSampleCommand::processRabund(RAbundVector*& rabund, ofstream& out) {
1315         try {
1316
1317                 int numBins = rabund->getNumBins();
1318                 int thisSize = rabund->getNumSeqs();
1319
1320                 if (thisSize != size) {
1321
1322                         OrderVector* order = new OrderVector();
1323                         for(int p=0;p<numBins;p++){
1324                                 for(int j=0;j<rabund->get(p);j++){
1325                                         order->push_back(p);
1326                                 }
1327                         }
1328                         random_shuffle(order->begin(), order->end());
1329
1330                         RAbundVector* temp = new RAbundVector(numBins);
1331                         temp->setLabel(rabund->getLabel());
1332
1333                         delete rabund;
1334                         rabund = temp;
1335
1336                         for (int j = 0; j < size; j++) {
1337
1338                                 if (m->control_pressed) { delete order; return 0; }
1339
1340                                 int bin = order->get(j);
1341
1342                                 int abund = rabund->get(bin);
1343                                 rabund->set(bin, (abund+1));
1344                         }
1345
1346                         delete order;
1347                 }
1348
1349                 if (m->control_pressed) { return 0; }
1350
1351                 rabund->print(out);
1352
1353                 return 0;
1354
1355         }
1356         catch(exception& e) {
1357                 m->errorOut(e, "SubSampleCommand", "processRabund");
1358                 exit(1);
1359         }
1360 }
1361 //**********************************************************************************************************************
1362 int SubSampleCommand::getSubSampleSabund() {
1363         try {
1364
1365                 InputData* input = new InputData(sabundfile, "sabund");
1366                 SAbundVector* sabund = input->getSAbundVector();
1367                 string lastLabel = sabund->getLabel();
1368
1369                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
1370                 set<string> processedLabels;
1371                 set<string> userLabels = labels;
1372
1373                 if (size == 0) { //user has not set size, set size = 10%
1374                         size = int((sabund->getNumSeqs()) * 0.10);
1375                 }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; }
1376
1377
1378                 m->mothurOut("Sampling " + toString(size) + " from " + toString(sabund->getNumSeqs()) + "."); m->mothurOutEndLine();
1379
1380                 string thisOutputDir = outputDir;
1381                 if (outputDir == "") {  thisOutputDir += m->hasPath(sabundfile);  }
1382                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(sabundfile)) + "subsample" + m->getExtension(sabundfile);
1383
1384                 ofstream out;
1385                 m->openOutputFile(outputFileName, out);
1386                 outputTypes["sabund"].push_back(outputFileName);  outputNames.push_back(outputFileName);
1387
1388
1389                 //as long as you are not at the end of the file or done wih the lines you want
1390                 while((sabund != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
1391                         if (m->control_pressed) {  delete input; delete sabund; out.close(); return 0;  }
1392
1393                         if(allLines == 1 || labels.count(sabund->getLabel()) == 1){
1394
1395                                 m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
1396
1397                                 processSabund(sabund, out);
1398
1399                                 processedLabels.insert(sabund->getLabel());
1400                                 userLabels.erase(sabund->getLabel());
1401                         }
1402
1403                         if ((m->anyLabelsToProcess(sabund->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
1404                                 string saveLabel = sabund->getLabel();
1405
1406                                 delete sabund;
1407
1408                                 sabund = input->getSAbundVector(lastLabel);
1409                                 m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
1410
1411                                 processSabund(sabund, out);
1412
1413                                 processedLabels.insert(sabund->getLabel());
1414                                 userLabels.erase(sabund->getLabel());
1415
1416                                 //restore real lastlabel to save below
1417                                 sabund->setLabel(saveLabel);
1418                         }
1419
1420                         lastLabel = sabund->getLabel();
1421
1422                         //prevent memory leak
1423                         delete sabund; sabund = NULL;
1424
1425                         //get next line to process
1426                         sabund = input->getSAbundVector();
1427                 }
1428
1429
1430                 if (m->control_pressed) {  out.close(); return 0;  }
1431
1432                 //output error messages about any remaining user labels
1433                 set<string>::iterator it;
1434                 bool needToRun = false;
1435                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
1436                         m->mothurOut("Your file does not include the label " + *it);
1437                         if (processedLabels.count(lastLabel) != 1) {
1438                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
1439                                 needToRun = true;
1440                         }else {
1441                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
1442                         }
1443                 }
1444
1445                 //run last label if you need to
1446                 if (needToRun == true)  {
1447                         if (sabund != NULL) { delete sabund; }
1448
1449                         sabund = input->getSAbundVector(lastLabel);
1450
1451                         m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
1452
1453                         processSabund(sabund, out);
1454
1455                         delete sabund;
1456                 }
1457
1458                 delete input;
1459                 out.close();
1460
1461                 return 0;
1462
1463         }
1464         catch(exception& e) {
1465                 m->errorOut(e, "SubSampleCommand", "getSubSampleSabund");
1466                 exit(1);
1467         }
1468 }
1469 //**********************************************************************************************************************
1470 int SubSampleCommand::processSabund(SAbundVector*& sabund, ofstream& out) {
1471         try {
1472
1473                 RAbundVector* rabund = new RAbundVector();
1474                 *rabund = sabund->getRAbundVector();
1475
1476                 int numBins = rabund->getNumBins();
1477                 int thisSize = rabund->getNumSeqs();
1478
1479                 if (thisSize != size) {
1480
1481                         OrderVector* order = new OrderVector();
1482                         for(int p=0;p<numBins;p++){
1483                                 for(int j=0;j<rabund->get(p);j++){
1484                                         order->push_back(p);
1485                                 }
1486                         }
1487                         random_shuffle(order->begin(), order->end());
1488
1489                         RAbundVector* temp = new RAbundVector(numBins);
1490                         temp->setLabel(rabund->getLabel());
1491
1492                         delete rabund;
1493                         rabund = temp;
1494
1495                         for (int j = 0; j < size; j++) {
1496
1497                                 if (m->control_pressed) { delete order; return 0; }
1498
1499                                 int bin = order->get(j);
1500
1501                                 int abund = rabund->get(bin);
1502                                 rabund->set(bin, (abund+1));
1503                         }
1504
1505                         delete order;
1506                 }
1507
1508                 if (m->control_pressed) { return 0; }
1509
1510                 delete sabund;
1511                 sabund = new SAbundVector();
1512                 *sabund = rabund->getSAbundVector();
1513                 delete rabund;
1514
1515                 sabund->print(out);
1516
1517                 return 0;
1518
1519         }
1520         catch(exception& e) {
1521                 m->errorOut(e, "SubSampleCommand", "processSabund");
1522                 exit(1);
1523         }
1524 }
1525 //**********************************************************************************************************************
1526 int SubSampleCommand::eliminateZeroOTUS(vector<SharedRAbundVector*>& thislookup) {
1527         try {
1528
1529                 vector<SharedRAbundVector*> newLookup;
1530                 for (int i = 0; i < thislookup.size(); i++) {
1531                         SharedRAbundVector* temp = new SharedRAbundVector();
1532                         temp->setLabel(thislookup[i]->getLabel());
1533                         temp->setGroup(thislookup[i]->getGroup());
1534                         newLookup.push_back(temp);
1535                 }
1536
1537                 //for each bin
1538                 vector<string> newBinLabels;
1539                 string snumBins = toString(thislookup[0]->getNumBins());
1540                 for (int i = 0; i < thislookup[0]->getNumBins(); i++) {
1541                         if (m->control_pressed) { for (int j = 0; j < newLookup.size(); j++) {  delete newLookup[j];  } return 0; }
1542
1543                         //look at each sharedRabund and make sure they are not all zero
1544                         bool allZero = true;
1545                         for (int j = 0; j < thislookup.size(); j++) {
1546                                 if (thislookup[j]->getAbundance(i) != 0) { allZero = false;  break;  }
1547                         }
1548
1549                         //if they are not all zero add this bin
1550                         if (!allZero) {
1551                                 for (int j = 0; j < thislookup.size(); j++) {
1552                                         newLookup[j]->push_back(thislookup[j]->getAbundance(i), thislookup[j]->getGroup());
1553                                 }
1554                                 //if there is a bin label use it otherwise make one
1555                                 string binLabel = "Otu";
1556                                 string sbinNumber = toString(i+1);
1557                                 if (sbinNumber.length() < snumBins.length()) {
1558                                         int diff = snumBins.length() - sbinNumber.length();
1559                                         for (int h = 0; h < diff; h++) { binLabel += "0"; }
1560                                 }
1561                                 binLabel += sbinNumber;
1562                                 if (i < m->currentBinLabels.size()) {  binLabel = m->currentBinLabels[i]; }
1563
1564                                 newBinLabels.push_back(binLabel);
1565                         }
1566                 }
1567
1568                 for (int j = 0; j < thislookup.size(); j++) {  delete thislookup[j];  }
1569                 thislookup.clear();
1570
1571                 thislookup = newLookup;
1572                 m->currentBinLabels = newBinLabels;
1573
1574                 return 0;
1575
1576         }
1577         catch(exception& e) {
1578                 m->errorOut(e, "SubSampleCommand", "eliminateZeroOTUS");
1579                 exit(1);
1580         }
1581 }
1582
1583 //**********************************************************************************************************************
1584
1585
1586