splitabundcommand.cpp

   1 /*
   2  *  splitabundcommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 5/17/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "splitabundcommand.h"
  11
  12 //**********************************************************************************************************************
  13 vector<string> SplitAbundCommand::getValidParameters(){
  14         try {
  15                 string Array[] =  {"name","group","list","label","accnos","groups","fasta","cutoff","outputdir","inputdir"};
  16                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  17                 return myArray;
  18         }
  19         catch(exception& e) {
  20                 m->errorOut(e, "SplitAbundCommand", "getValidParameters");
  21                 exit(1);
  22         }
  23 }
  24 //**********************************************************************************************************************
  25 SplitAbundCommand::SplitAbundCommand(){
  26         try {
  27                 abort = true; calledHelp = true;
  28                 vector<string> tempOutNames;
  29                 outputTypes["list"] = tempOutNames;
  30                 outputTypes["name"] = tempOutNames;
  31                 outputTypes["accnos"] = tempOutNames;
  32                 outputTypes["group"] = tempOutNames;
  33                 outputTypes["fasta"] = tempOutNames;
  34         }
  35         catch(exception& e) {
  36                 m->errorOut(e, "SplitAbundCommand", "SplitAbundCommand");
  37                 exit(1);
  38         }
  39 }
  40 //**********************************************************************************************************************
  41 vector<string> SplitAbundCommand::getRequiredParameters(){
  42         try {
  43                 string Array[] =  {"fasta","list","name","or"};
  44                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  45                 return myArray;
  46         }
  47         catch(exception& e) {
  48                 m->errorOut(e, "SplitAbundCommand", "getRequiredParameters");
  49                 exit(1);
  50         }
  51 }
  52 //**********************************************************************************************************************
  53 vector<string> SplitAbundCommand::getRequiredFiles(){
  54         try {
  55                 vector<string> myArray;
  56                 return myArray;
  57         }
  58         catch(exception& e) {
  59                 m->errorOut(e, "SplitAbundCommand", "getRequiredFiles");
  60                 exit(1);
  61         }
  62 }
  63 //**********************************************************************************************************************
  64 SplitAbundCommand::SplitAbundCommand(string option)  {
  65         try {
  66                 abort = false; calledHelp = false;
  67                 allLines = 1;
  68
  69                 //allow user to run help
  70                 if(option == "help") { help(); abort = true; calledHelp = true; }
  71
  72                 else {
  73                         //valid paramters for this command
  74                         string Array[] =  {"name","group","list","label","accnos","groups","fasta","cutoff","outputdir","inputdir"}; //
  75                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  76
  77                         OptionParser parser(option);
  78                         map<string, string> parameters = parser.getParameters();
  79
  80                         ValidParameters validParameter;
  81                         map<string, string>::iterator it;
  82
  83                         //check to make sure all parameters are valid for command
  84                         for (it = parameters.begin(); it != parameters.end(); it++) {
  85                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  86                         }
  87
  88                         //initialize outputTypes
  89                         vector<string> tempOutNames;
  90                         outputTypes["list"] = tempOutNames;
  91                         outputTypes["name"] = tempOutNames;
  92                         outputTypes["accnos"] = tempOutNames;
  93                         outputTypes["group"] = tempOutNames;
  94                         outputTypes["fasta"] = tempOutNames;
  95
  96                         //if the user changes the input directory command factory will send this info to us in the output parameter
  97                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  98                         if (inputDir == "not found"){   inputDir = "";          }
  99                         else {
 100                                 string path;
 101                                 it = parameters.find("list");
 102                                 //user has given a template file
 103                                 if(it != parameters.end()){
 104                                         path = m->hasPath(it->second);
 105                                         //if the user has not given a path then, add inputdir. else leave path alone.
 106                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 107                                 }
 108
 109                                 it = parameters.find("group");
 110                                 //user has given a template file
 111                                 if(it != parameters.end()){
 112                                         path = m->hasPath(it->second);
 113                                         //if the user has not given a path then, add inputdir. else leave path alone.
 114                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 115                                 }
 116
 117                                 it = parameters.find("fasta");
 118                                 //user has given a template file
 119                                 if(it != parameters.end()){
 120                                         path = m->hasPath(it->second);
 121                                         //if the user has not given a path then, add inputdir. else leave path alone.
 122                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 123                                 }
 124
 125                                 it = parameters.find("name");
 126                                 //user has given a template file
 127                                 if(it != parameters.end()){
 128                                         path = m->hasPath(it->second);
 129                                         //if the user has not given a path then, add inputdir. else leave path alone.
 130                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 131                                 }
 132
 133                         }
 134
 135
 136                         //if the user changes the output directory command factory will send this info to us in the output parameter
 137                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
 138
 139                         //check for required parameters
 140                         listfile = validParameter.validFile(parameters, "list", true);
 141                         if (listfile == "not open") { abort = true; }
 142                         else if (listfile == "not found") { listfile = ""; }
 143                         else{ inputFile = listfile; }
 144
 145                         namefile = validParameter.validFile(parameters, "name", true);
 146                         if (namefile == "not open") { abort = true; }
 147                         else if (namefile == "not found") { namefile = ""; }
 148                         else{ inputFile = namefile; }
 149
 150                         fastafile = validParameter.validFile(parameters, "fasta", true);
 151                         if (fastafile == "not open") { abort = true; }
 152                         else if (fastafile == "not found") { fastafile = ""; m->mothurOut("fasta is a required parameter for the split.abund command. "); m->mothurOutEndLine(); abort = true;  }
 153
 154                         groupfile = validParameter.validFile(parameters, "group", true);
 155                         if (groupfile == "not open") {  groupfile = ""; abort = true; }
 156                         else if (groupfile == "not found") { groupfile = ""; }
 157                         else {
 158                                 groupMap = new GroupMap(groupfile);
 159
 160                                 int error = groupMap->readMap();
 161                                 if (error == 1) { abort = true; }
 162
 163                         }
 164
 165                         groups = validParameter.validFile(parameters, "groups", false);
 166                         if (groups == "not found") { groups = ""; }
 167                         else if (groups == "all") {
 168                                 if (groupfile != "") {  Groups = groupMap->namesOfGroups;  }
 169                                 else {  m->mothurOut("You cannot select groups without a valid groupfile, I will disregard your groups selection. "); m->mothurOutEndLine(); groups = "";   }
 170                         }else {
 171                                 m->splitAtDash(groups, Groups);
 172                         }
 173
 174                         if ((groupfile == "") && (groups != "")) {  m->mothurOut("You cannot select groups without a valid groupfile, I will disregard your groups selection. "); m->mothurOutEndLine(); groups = "";  Groups.clear(); }
 175
 176                         //do you have all files needed
 177                         if ((listfile == "") && (namefile == "")) { m->mothurOut("You must either a listfile or a namefile for the split.abund command. "); m->mothurOutEndLine(); abort = true;  }
 178
 179                         //check for optional parameter and set defaults
 180                         // ...at some point should added some additional type checking...
 181                         label = validParameter.validFile(parameters, "label", false);
 182                         if (label == "not found") { label = "";  allLines = 1; }
 183                         else {
 184                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 185                                 else { allLines = 1;  }
 186                         }
 187
 188                         string temp = validParameter.validFile(parameters, "accnos", false);            if (temp == "not found") { temp = "F"; }
 189                         accnos = m->isTrue(temp);
 190
 191                         temp = validParameter.validFile(parameters, "cutoff", false);                           if (temp == "not found") { temp = "0"; }
 192                         convert(temp, cutoff);
 193
 194                         if (cutoff == 0) {  m->mothurOut("You must provide a cutoff to qualify what is abundant for the split.abund command. "); m->mothurOutEndLine(); abort = true;  }
 195
 196                 }
 197
 198         }
 199         catch(exception& e) {
 200                 m->errorOut(e, "SplitAbundCommand", "SplitAbundCommand");
 201                 exit(1);
 202         }
 203 }
 204 //**********************************************************************************************************************
 205 void SplitAbundCommand::help(){
 206         try {
 207                 m->mothurOut("The split.abund command reads a fasta file and a list or a names file splits the sequences into rare and abundant groups. \n");
 208                 m->mothurOut("The split.abund command parameters are fasta, list, name, cutoff, group, label, groups, cutoff and accnos.\n");
 209                 m->mothurOut("The fasta and a list or name parameter are required, and you must provide a cutoff value.\n");
 210                 m->mothurOut("The cutoff parameter is used to qualify what is abundant and rare.\n");
 211                 m->mothurOut("The group parameter allows you to parse a group file into rare and abundant groups.\n");
 212                 m->mothurOut("The label parameter is used to read specific labels in your listfile you want to use.\n");
 213                 m->mothurOut("The accnos parameter allows you to output a .rare.accnos and .abund.accnos files to use with the get.seqs and remove.seqs commands.\n");
 214                 m->mothurOut("The groups parameter allows you to parse the files into rare and abundant files by group.  \n");
 215                 m->mothurOut("For example if you set groups=A-B-C, you will get a .A.abund, .A.rare, .B.abund, .B.rare, .C.abund, .C.rare files.  \n");
 216                 m->mothurOut("If you want .abund and .rare files for all groups, set groups=all.  \n");
 217                 m->mothurOut("The split.abund command should be used in the following format: split.abund(fasta=yourFasta, list=yourListFile, group=yourGroupFile, label=yourLabels, cutoff=yourCutoff).\n");
 218                 m->mothurOut("Example: split.abund(fasta=abrecovery.fasta, list=abrecovery.fn.list, group=abrecovery.groups, label=0.03, cutoff=2).\n");
 219                 m->mothurOut("Note: No spaces between parameter labels (i.e. list), '=' and parameters (i.e.yourListfile).\n\n");
 220
 221         }
 222         catch(exception& e) {
 223                 m->errorOut(e, "SplitAbundCommand", "help");
 224                 exit(1);
 225         }
 226 }
 227 //**********************************************************************************************************************
 228 SplitAbundCommand::~SplitAbundCommand(){
 229         if (groupfile != "") {  delete groupMap;  }
 230 }
 231 //**********************************************************************************************************************
 232 int SplitAbundCommand::execute(){
 233         try {
 234
 235                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 236
 237                 if (listfile != "") { //you are using a listfile to determine abundance
 238                         if (outputDir == "") { outputDir = m->hasPath(listfile); }
 239
 240                         //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 241                         set<string> processedLabels;
 242                         set<string> userLabels = labels;
 243
 244                         input = new InputData(listfile, "list");
 245                         list = input->getListVector();
 246                         string lastLabel = list->getLabel();
 247
 248                         //do you have a namefile or do we need to similate one?
 249                         if (namefile != "") {  readNamesFile();         }
 250                         else                            { createNameMap(list);  }
 251
 252                         if (m->control_pressed) { delete input; delete list; for (int i = 0; i < outputNames.size(); i++) {     remove(outputNames[i].c_str()); } return 0; }
 253
 254                         while((list != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 255
 256                                 if (m->control_pressed) { delete input; delete list; for (int i = 0; i < outputNames.size(); i++) {     remove(outputNames[i].c_str()); } return 0; }
 257
 258                                 if(allLines == 1 || labels.count(list->getLabel()) == 1){
 259
 260                                                 m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 261                                                 splitList(list);
 262
 263                                                 processedLabels.insert(list->getLabel());
 264                                                 userLabels.erase(list->getLabel());
 265                                 }
 266
 267                                 if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 268                                                 string saveLabel = list->getLabel();
 269
 270                                                 delete list;
 271                                                 list = input->getListVector(lastLabel); //get new list vector to process
 272
 273                                                 m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 274                                                 splitList(list);
 275
 276                                                 processedLabels.insert(list->getLabel());
 277                                                 userLabels.erase(list->getLabel());
 278
 279                                                 //restore real lastlabel to save below
 280                                                 list->setLabel(saveLabel);
 281                                 }
 282
 283
 284                                 lastLabel = list->getLabel();
 285
 286                                 delete list;
 287                                 list = input->getListVector(); //get new list vector to process
 288                         }
 289
 290                         if (m->control_pressed) { delete input;  for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } return 0; }
 291
 292                         //output error messages about any remaining user labels
 293                         set<string>::iterator it;
 294                         bool needToRun = false;
 295                         for (it = userLabels.begin(); it != userLabels.end(); it++) {
 296                                 m->mothurOut("Your file does not include the label " + *it);
 297                                 if (processedLabels.count(lastLabel) != 1) {
 298                                         m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 299                                         needToRun = true;
 300                                 }else {
 301                                         m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 302                                 }
 303
 304                         }
 305
 306                         if (m->control_pressed) { delete input;  for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } return 0; }
 307
 308                         //run last label if you need to
 309                         if (needToRun == true)  {
 310                                 if (list != NULL) {     delete list;    }
 311                                 list = input->getListVector(lastLabel); //get new list vector to process
 312
 313                                 m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 314                                 splitList(list);
 315
 316                                 delete list;
 317                         }
 318
 319                         delete input;
 320
 321                         if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str()); }       return 0;       }
 322
 323                 }else { //you are using the namefile to determine abundance
 324                         if (outputDir == "") { outputDir = m->hasPath(namefile); }
 325
 326                         splitNames();
 327                         writeNames();
 328
 329                         string tag = "";
 330                         if (groupfile != "")                            {  parseGroup(tag);             }
 331                         if (accnos)                                                     {  writeAccnos(tag);    }
 332                         if (fastafile != "")                            {  parseFasta(tag);             }
 333                 }
 334
 335                 m->mothurOutEndLine();
 336                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 337                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 338                 m->mothurOutEndLine();
 339
 340                 return 0;
 341         }
 342         catch(exception& e) {
 343                 m->errorOut(e, "SplitAbundCommand", "execute");
 344                 exit(1);
 345         }
 346 }
 347 /**********************************************************************************************************************/
 348 int SplitAbundCommand::splitList(ListVector* thisList) {
 349         try {
 350                 rareNames.clear();
 351                 abundNames.clear();
 352
 353                 //get rareNames and abundNames
 354                 for (int i = 0; i < thisList->getNumBins(); i++) {
 355                         if (m->control_pressed) { return 0; }
 356
 357                         string bin = thisList->get(i);
 358
 359                         vector<string> names;
 360                         m->splitAtComma(bin, names);  //parses bin into individual sequence names
 361                         int size = names.size();
 362
 363                         if (size <= cutoff) {
 364                                 for (int j = 0; j < names.size(); j++) {  rareNames.insert(names[j]);  }
 365                         }else{
 366                                 for (int j = 0; j < names.size(); j++) {  abundNames.insert(names[j]);  }
 367                         }
 368                 }//end for
 369
 370
 371                 string tag = thisList->getLabel() + ".";
 372
 373                 writeList(thisList, tag);
 374
 375                 if (groupfile != "")                            {  parseGroup(tag);             }
 376                 if (accnos)                                                     {  writeAccnos(tag);    }
 377                 if (fastafile != "")                            {  parseFasta(tag);             }
 378
 379                 return 0;
 380
 381         }
 382         catch(exception& e) {
 383                 m->errorOut(e, "SplitAbundCommand", "splitList");
 384                 exit(1);
 385         }
 386 }
 387 /**********************************************************************************************************************/
 388 int SplitAbundCommand::writeList(ListVector* thisList, string tag) {
 389         try {
 390
 391                 map<string, ofstream*> filehandles;
 392
 393                 if (Groups.size() == 0) {
 394                         SAbundVector* sabund = new SAbundVector();
 395                         *sabund = thisList->getSAbundVector();
 396
 397                         //find out how many bins are rare and how many are abundant so you can process the list vector one bin at a time
 398                         // and don't have to store the bins until you are done with the whole vector, this save alot of space.
 399                         int numRareBins = 0;
 400                         for (int i = 0; i <= sabund->getMaxRank(); i++) {
 401                                 if (i > cutoff) { break; }
 402                                 numRareBins += sabund->get(i);
 403                         }
 404                         int numAbundBins = thisList->getNumBins() - numRareBins;
 405                         delete sabund;
 406
 407                         ofstream aout;
 408                         ofstream rout;
 409
 410                         string rare = outputDir + m->getRootName(m->getSimpleName(listfile)) + tag + "rare.list";
 411                         m->openOutputFile(rare, rout);
 412                         outputNames.push_back(rare); outputTypes["list"].push_back(rare);
 413
 414                         string abund = outputDir + m->getRootName(m->getSimpleName(listfile)) + tag + "abund.list";
 415                         m->openOutputFile(abund, aout);
 416                         outputNames.push_back(abund); outputTypes["list"].push_back(abund);
 417
 418                         if (rareNames.size() != 0)      {  rout << thisList->getLabel() << '\t' << numRareBins << '\t';         }
 419                         if (abundNames.size() != 0) {   aout << thisList->getLabel() << '\t' << numAbundBins << '\t';   }
 420
 421                         for (int i = 0; i < thisList->getNumBins(); i++) {
 422                                 if (m->control_pressed) { break; }
 423
 424                                 string bin = list->get(i);
 425
 426                                 int size = m->getNumNames(bin);
 427
 428                                 if (size <= cutoff) {  rout << bin << '\t';  }
 429                                 else                            {  aout << bin << '\t'; }
 430                         }
 431
 432                         if (rareNames.size() != 0)      { rout << endl; }
 433                         if (abundNames.size() != 0) { aout << endl; }
 434
 435                         rout.close();
 436                         aout.close();
 437
 438                 }else{ //parse names by abundance and group
 439                         string fileroot =  outputDir + m->getRootName(m->getSimpleName(listfile));
 440                         ofstream* temp;
 441                         ofstream* temp2;
 442                         //map<string, bool> wroteFile;
 443                         map<string, ofstream*> filehandles;
 444                         map<string, ofstream*>::iterator it3;
 445
 446                         for (int i=0; i<Groups.size(); i++) {
 447                                 temp = new ofstream;
 448                                 filehandles[Groups[i]+".rare"] = temp;
 449                                 temp2 = new ofstream;
 450                                 filehandles[Groups[i]+".abund"] = temp2;
 451
 452                                 m->openOutputFile(fileroot + Groups[i] + tag + ".rare.list", *(filehandles[Groups[i]+".rare"]));
 453                                 m->openOutputFile(fileroot + Groups[i] + tag + ".abund.list", *(filehandles[Groups[i]+".abund"]));
 454                                 outputNames.push_back(fileroot + Groups[i] + tag + ".rare.list"); outputTypes["list"].push_back(fileroot + Groups[i] + tag + ".rare.list");
 455                                 outputNames.push_back(fileroot + Groups[i] + tag + ".abund.list"); outputTypes["list"].push_back(fileroot + Groups[i] + tag + ".abund.list");
 456                         }
 457
 458                         map<string, string> groupVector;
 459                         map<string, string>::iterator itGroup;
 460                         map<string, int> groupNumBins;
 461
 462                         for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 463                                 groupNumBins[it3->first] = 0;
 464                                 groupVector[it3->first] = "";
 465                         }
 466
 467                         for (int i = 0; i < thisList->getNumBins(); i++) {
 468                                 if (m->control_pressed) { break; }
 469
 470                                 map<string, string> groupBins;
 471                                 string bin = list->get(i);
 472
 473                                 vector<string> names;
 474                                 m->splitAtComma(bin, names);  //parses bin into individual sequence names
 475
 476                                 //parse bin into list of sequences in each group
 477                                 for (int j = 0; j < names.size(); j++) {
 478                                         string rareAbund;
 479                                         if (rareNames.count(names[j]) != 0) { //you are a rare name
 480                                                 rareAbund = ".rare";
 481                                         }else{ //you are a abund name
 482                                                 rareAbund = ".abund";
 483                                         }
 484
 485                                         string group = groupMap->getGroup(names[j]);
 486
 487                                         if (m->inUsersGroups(group, Groups)) { //only add if this is in a group we want
 488                                                 itGroup = groupBins.find(group+rareAbund);
 489                                                 if(itGroup == groupBins.end()) {
 490                                                         groupBins[group+rareAbund] = names[j];  //add first name
 491                                                         groupNumBins[group+rareAbund]++;
 492                                                 }else{ //add another name
 493                                                         groupBins[group+rareAbund] +=  "," + names[j];
 494                                                 }
 495                                         }else if(group == "not found") {
 496                                                 m->mothurOut(names[j] + " is not in your groupfile. Ignoring."); m->mothurOutEndLine();
 497                                         }
 498                                 }
 499
 500
 501                                 for (itGroup = groupBins.begin(); itGroup != groupBins.end(); itGroup++) {
 502                                         groupVector[itGroup->first] +=  itGroup->second + '\t';
 503                                 }
 504                         }
 505
 506                         //end list vector
 507                         for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 508                                 (*(filehandles[it3->first])) << thisList->getLabel() << '\t' << groupNumBins[it3->first] << '\t' << groupVector[it3->first] << endl;  // label numBins  listvector for that group
 509                                 (*(filehandles[it3->first])).close();
 510                                 delete it3->second;
 511                         }
 512                 }
 513
 514                 return 0;
 515
 516         }
 517         catch(exception& e) {
 518                 m->errorOut(e, "SplitAbundCommand", "writeList");
 519                 exit(1);
 520         }
 521 }
 522 /**********************************************************************************************************************/
 523 int SplitAbundCommand::splitNames() { //namefile
 524         try {
 525
 526                 rareNames.clear();
 527                 abundNames.clear();
 528
 529                 //open input file
 530                 ifstream in;
 531                 m->openInputFile(namefile, in);
 532
 533                 while (!in.eof()) {
 534                         if (m->control_pressed) { break; }
 535
 536                         string firstCol, secondCol;
 537                         in >> firstCol >> secondCol; m->gobble(in);
 538
 539                         nameMap[firstCol] = secondCol;
 540
 541                         int size = m->getNumNames(secondCol);
 542
 543                         if (size <= cutoff) {
 544                                 rareNames.insert(firstCol);
 545                         }else{
 546                                 abundNames.insert(firstCol);
 547                         }
 548                 }
 549                 in.close();
 550
 551                 return 0;
 552
 553         }
 554         catch(exception& e) {
 555                 m->errorOut(e, "SplitAbundCommand", "splitNames");
 556                 exit(1);
 557         }
 558 }
 559 /**********************************************************************************************************************/
 560 int SplitAbundCommand::readNamesFile() {
 561         try {
 562                 //open input file
 563                 ifstream in;
 564                 m->openInputFile(namefile, in);
 565
 566                 while (!in.eof()) {
 567                         if (m->control_pressed) { break; }
 568
 569                         string firstCol, secondCol;
 570                         in >> firstCol >> secondCol; m->gobble(in);
 571
 572                         nameMap[firstCol] = secondCol;
 573                 }
 574                 in.close();
 575
 576                 return 0;
 577
 578         }
 579         catch(exception& e) {
 580                 m->errorOut(e, "SplitAbundCommand", "readNamesFile");
 581                 exit(1);
 582         }
 583 }
 584 /**********************************************************************************************************************/
 585 int SplitAbundCommand::createNameMap(ListVector* thisList) {
 586         try {
 587
 588                 if (thisList != NULL) {
 589                         for (int i = 0; i < thisList->getNumBins(); i++) {
 590                                 if (m->control_pressed) { return 0; }
 591
 592                                 string bin = thisList->get(i);
 593
 594                                 vector<string> names;
 595                                 m->splitAtComma(bin, names);  //parses bin into individual sequence names
 596
 597                                 for (int j = 0; j < names.size(); j++) {  nameMap[names[j]] = names[j];  }
 598                         }//end for
 599                 }
 600
 601                 return 0;
 602         }
 603         catch(exception& e) {
 604                 m->errorOut(e, "SplitAbundCommand", "createNameMap");
 605                 exit(1);
 606         }
 607 }
 608 /**********************************************************************************************************************/
 609 int SplitAbundCommand::writeNames() { //namefile
 610         try {
 611
 612                 map<string, ofstream*> filehandles;
 613
 614                 if (Groups.size() == 0) {
 615                         ofstream aout;
 616                         ofstream rout;
 617
 618                         string rare = outputDir + m->getRootName(m->getSimpleName(namefile))  + "rare.names";
 619                         m->openOutputFile(rare, rout);
 620                         outputNames.push_back(rare); outputTypes["name"].push_back(rare);
 621
 622                         string abund = outputDir + m->getRootName(m->getSimpleName(namefile))  + "abund.names";
 623                         m->openOutputFile(abund, aout);
 624                         outputNames.push_back(abund); outputTypes["name"].push_back(abund);
 625
 626                         if (rareNames.size() != 0) {
 627                                 for (set<string>::iterator itRare = rareNames.begin(); itRare != rareNames.end(); itRare++) {
 628                                         rout << (*itRare) << '\t' << nameMap[(*itRare)] << endl;
 629                                 }
 630                         }
 631                         rout.close();
 632
 633                         if (abundNames.size() != 0) {
 634                                 for (set<string>::iterator itAbund = abundNames.begin(); itAbund != abundNames.end(); itAbund++) {
 635                                         aout << (*itAbund) << '\t' << nameMap[(*itAbund)] << endl;
 636                                 }
 637                         }
 638                         aout.close();
 639
 640                 }else{ //parse names by abundance and group
 641                         string fileroot =  outputDir + m->getRootName(m->getSimpleName(namefile));
 642                         ofstream* temp;
 643                         ofstream* temp2;
 644                         map<string, ofstream*> filehandles;
 645                         map<string, ofstream*>::iterator it3;
 646
 647                         for (int i=0; i<Groups.size(); i++) {
 648                                 temp = new ofstream;
 649                                 filehandles[Groups[i]+".rare"] = temp;
 650                                 temp2 = new ofstream;
 651                                 filehandles[Groups[i]+".abund"] = temp2;
 652
 653                                 m->openOutputFile(fileroot + Groups[i] + ".rare.names", *(filehandles[Groups[i]+".rare"]));
 654                                 m->openOutputFile(fileroot + Groups[i] + ".abund.names", *(filehandles[Groups[i]+".abund"]));
 655                         }
 656
 657                         for (map<string, string>::iterator itName = nameMap.begin(); itName != nameMap.end(); itName++) {
 658                                 vector<string> names;
 659                                 m->splitAtComma(itName->second, names);  //parses bin into individual sequence names
 660
 661                                 string rareAbund;
 662                                 if (rareNames.count(itName->first) != 0) { //you are a rare name
 663                                                 rareAbund = ".rare";
 664                                 }else{ //you are a abund name
 665                                                 rareAbund = ".abund";
 666                                 }
 667
 668                                 map<string, string> outputStrings;
 669                                 map<string, string>::iterator itout;
 670                                 for (int i = 0; i < names.size(); i++) {
 671
 672                                         string group = groupMap->getGroup(names[i]);
 673
 674                                         if (m->inUsersGroups(group, Groups)) { //only add if this is in a group we want
 675                                                 itout = outputStrings.find(group+rareAbund);
 676                                                 if (itout == outputStrings.end()) {
 677                                                         outputStrings[group+rareAbund] = names[i] + '\t' + names[i];
 678                                                 }else {   outputStrings[group+rareAbund] += "," + names[i]; }
 679                                         }else if(group == "not found") {
 680                                                 m->mothurOut(names[i] + " is not in your groupfile. Ignoring."); m->mothurOutEndLine();
 681                                         }
 682                                 }
 683
 684                                 for (itout = outputStrings.begin(); itout != outputStrings.end(); itout++) { *(filehandles[itout->first]) << itout->second << endl;     }
 685                         }
 686
 687
 688                         for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 689                                 (*(filehandles[it3->first])).close();
 690                                 outputNames.push_back(fileroot + it3->first + ".names");  outputTypes["name"].push_back(fileroot + it3->first + ".names");
 691                                 delete it3->second;
 692                         }
 693                 }
 694
 695                 return 0;
 696
 697         }
 698         catch(exception& e) {
 699                 m->errorOut(e, "SplitAbundCommand", "writeNames");
 700                 exit(1);
 701         }
 702 }
 703 /**********************************************************************************************************************/
 704 //just write the unique names - if a namesfile is given
 705 int SplitAbundCommand::writeAccnos(string tag) {
 706         try {
 707
 708                 map<string, ofstream*> filehandles;
 709
 710                 if (Groups.size() == 0) {
 711                         ofstream aout;
 712                         ofstream rout;
 713
 714
 715                         string rare = outputDir + m->getRootName(m->getSimpleName(inputFile))  + tag + "rare.accnos";
 716                         m->openOutputFile(rare, rout);
 717                         outputNames.push_back(rare); outputTypes["accnos"].push_back(rare);
 718
 719                         for (set<string>::iterator itRare = rareNames.begin(); itRare != rareNames.end(); itRare++) {
 720                                 rout << (*itRare) << endl;
 721                         }
 722                         rout.close();
 723
 724                         string abund = outputDir + m->getRootName(m->getSimpleName(inputFile)) + tag  + "abund.accnos";
 725                         m->openOutputFile(abund, aout);
 726                         outputNames.push_back(abund); outputTypes["accnos"].push_back(abund);
 727
 728                         for (set<string>::iterator itAbund = abundNames.begin(); itAbund != abundNames.end(); itAbund++) {
 729                                 aout << (*itAbund) << endl;
 730                         }
 731                         aout.close();
 732
 733                 }else{ //parse names by abundance and group
 734                         string fileroot =  outputDir + m->getRootName(m->getSimpleName(inputFile));
 735                         ofstream* temp;
 736                         ofstream* temp2;
 737                         map<string, ofstream*> filehandles;
 738                         map<string, ofstream*>::iterator it3;
 739
 740                         for (int i=0; i<Groups.size(); i++) {
 741                                 temp = new ofstream;
 742                                 filehandles[Groups[i]+".rare"] = temp;
 743                                 temp2 = new ofstream;
 744                                 filehandles[Groups[i]+".abund"] = temp2;
 745
 746                                 m->openOutputFile(fileroot + tag + Groups[i] + ".rare.accnos", *(filehandles[Groups[i]+".rare"]));
 747                                 m->openOutputFile(fileroot + tag + Groups[i] + ".abund.accnos", *(filehandles[Groups[i]+".abund"]));
 748                         }
 749
 750                         //write rare
 751                         for (set<string>::iterator itRare = rareNames.begin(); itRare != rareNames.end(); itRare++) {
 752                                         string group = groupMap->getGroup(*itRare);
 753
 754                                         if (m->inUsersGroups(group, Groups)) { //only add if this is in a group we want
 755                                                 *(filehandles[group+".rare"]) << *itRare << endl;
 756                                         }
 757                         }
 758
 759                         //write abund
 760                         for (set<string>::iterator itAbund = abundNames.begin(); itAbund != abundNames.end(); itAbund++) {
 761                                         string group = groupMap->getGroup(*itAbund);
 762
 763                                         if (m->inUsersGroups(group, Groups)) { //only add if this is in a group we want
 764                                                 *(filehandles[group+".abund"]) << *itAbund << endl;
 765                                         }
 766                         }
 767
 768                         //close files
 769                         for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 770                                 (*(filehandles[it3->first])).close();
 771                                 outputNames.push_back(fileroot + tag + it3->first + ".accnos");  outputTypes["accnos"].push_back(fileroot + tag + it3->first + ".accnos");
 772                                 delete it3->second;
 773                         }
 774                 }
 775
 776                 return 0;
 777
 778         }
 779         catch(exception& e) {
 780                 m->errorOut(e, "SplitAbundCommand", "writeAccnos");
 781                 exit(1);
 782         }
 783 }
 784 /**********************************************************************************************************************/
 785 int SplitAbundCommand::parseGroup(string tag) { //namefile
 786         try {
 787
 788                 map<string, ofstream*> filehandles;
 789
 790                 if (Groups.size() == 0) {
 791                         ofstream aout;
 792                         ofstream rout;
 793
 794                         string rare = outputDir + m->getRootName(m->getSimpleName(groupfile))  + tag + "rare.groups";
 795                         m->openOutputFile(rare, rout);
 796                         outputNames.push_back(rare); outputTypes["group"].push_back(rare);
 797
 798                         string abund = outputDir + m->getRootName(m->getSimpleName(groupfile))  + tag + "abund.groups";
 799                         m->openOutputFile(abund, aout);
 800                         outputNames.push_back(abund); outputTypes["group"].push_back(abund);
 801
 802                         for (map<string, string>::iterator itName = nameMap.begin(); itName != nameMap.end(); itName++) {
 803                                 vector<string> names;
 804                                 m->splitAtComma(itName->second, names);  //parses bin into individual sequence names
 805
 806                                 for (int i = 0; i < names.size(); i++) {
 807
 808                                         string group = groupMap->getGroup(names[i]);
 809
 810                                         if (group == "not found") {
 811                                                 m->mothurOut(names[i] + " is not in your groupfile, ignoring, please correct."); m->mothurOutEndLine();
 812                                         }else {
 813                                                 if (rareNames.count(itName->first) != 0) { //you are a rare name
 814                                                         rout << names[i] << '\t' << group << endl;
 815                                                 }else{ //you are a abund name
 816                                                         aout << names[i] << '\t' << group << endl;
 817                                                 }
 818                                         }
 819                                 }
 820                         }
 821
 822                         rout.close();
 823                         aout.close();
 824
 825                 }else{ //parse names by abundance and group
 826                         string fileroot =  outputDir + m->getRootName(m->getSimpleName(groupfile));
 827                         ofstream* temp;
 828                         ofstream* temp2;
 829                         map<string, ofstream*> filehandles;
 830                         map<string, ofstream*>::iterator it3;
 831
 832                         for (int i=0; i<Groups.size(); i++) {
 833                                 temp = new ofstream;
 834                                 filehandles[Groups[i]+".rare"] = temp;
 835                                 temp2 = new ofstream;
 836                                 filehandles[Groups[i]+".abund"] = temp2;
 837
 838                                 m->openOutputFile(fileroot + tag + Groups[i] + ".rare.groups", *(filehandles[Groups[i]+".rare"]));
 839                                 m->openOutputFile(fileroot + tag + Groups[i] + ".abund.groups", *(filehandles[Groups[i]+".abund"]));
 840                         }
 841
 842                         for (map<string, string>::iterator itName = nameMap.begin(); itName != nameMap.end(); itName++) {
 843                                 vector<string> names;
 844                                 m->splitAtComma(itName->second, names);  //parses bin into individual sequence names
 845
 846                                 string rareAbund;
 847                                 if (rareNames.count(itName->first) != 0) { //you are a rare name
 848                                         rareAbund = ".rare";
 849                                 }else{ //you are a abund name
 850                                         rareAbund = ".abund";
 851                                 }
 852
 853                                 for (int i = 0; i < names.size(); i++) {
 854
 855                                         string group = groupMap->getGroup(names[i]);
 856
 857                                         if (m->inUsersGroups(group, Groups)) { //only add if this is in a group we want
 858                                                 *(filehandles[group+rareAbund]) << names[i] << '\t' << group << endl;
 859                                         }
 860                                 }
 861                         }
 862
 863                         for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 864                                 (*(filehandles[it3->first])).close();
 865                                 outputNames.push_back(fileroot + tag + it3->first + ".groups");  outputTypes["group"].push_back(fileroot + tag + it3->first + ".groups");
 866                                 delete it3->second;
 867                         }
 868                 }
 869
 870                 return 0;
 871
 872         }
 873         catch(exception& e) {
 874                 m->errorOut(e, "SplitAbundCommand", "parseGroups");
 875                 exit(1);
 876         }
 877 }
 878 /**********************************************************************************************************************/
 879 int SplitAbundCommand::parseFasta(string tag) { //namefile
 880         try {
 881
 882                 map<string, ofstream*> filehandles;
 883
 884                 if (Groups.size() == 0) {
 885                         ofstream aout;
 886                         ofstream rout;
 887
 888                         string rare = outputDir + m->getRootName(m->getSimpleName(fastafile))  + tag + "rare.fasta";
 889                         m->openOutputFile(rare, rout);
 890                         outputNames.push_back(rare); outputTypes["fasta"].push_back(rare);
 891
 892                         string abund = outputDir + m->getRootName(m->getSimpleName(fastafile))  + tag + "abund.fasta";
 893                         m->openOutputFile(abund, aout);
 894                         outputNames.push_back(abund); outputTypes["fasta"].push_back(abund);
 895
 896                         //open input file
 897                         ifstream in;
 898                         m->openInputFile(fastafile, in);
 899
 900                         while (!in.eof()) {
 901                                 if (m->control_pressed) { break; }
 902
 903                                 Sequence seq(in); m->gobble(in);
 904
 905                                 if (seq.getName() != "") {
 906
 907                                         map<string, string>::iterator itNames;
 908
 909                                         itNames = nameMap.find(seq.getName());
 910
 911                                         if (itNames == nameMap.end()) {
 912                                                 m->mothurOut(seq.getName() + " is not in your namesfile, ignoring."); m->mothurOutEndLine();
 913                                         }else{
 914                                                 if (rareNames.count(seq.getName()) != 0) { //you are a rare name
 915                                                         seq.printSequence(rout);
 916                                                 }else{ //you are a abund name
 917                                                         seq.printSequence(aout);
 918                                                 }
 919                                         }
 920                                 }
 921                         }
 922                         in.close();
 923                         rout.close();
 924                         aout.close();
 925
 926                 }else{ //parse names by abundance and group
 927                         string fileroot =  outputDir + m->getRootName(m->getSimpleName(fastafile));
 928                         ofstream* temp;
 929                         ofstream* temp2;
 930                         map<string, ofstream*> filehandles;
 931                         map<string, ofstream*>::iterator it3;
 932
 933                         for (int i=0; i<Groups.size(); i++) {
 934                                 temp = new ofstream;
 935                                 filehandles[Groups[i]+".rare"] = temp;
 936                                 temp2 = new ofstream;
 937                                 filehandles[Groups[i]+".abund"] = temp2;
 938
 939                                 m->openOutputFile(fileroot + tag + Groups[i] + ".rare.fasta", *(filehandles[Groups[i]+".rare"]));
 940                                 m->openOutputFile(fileroot + tag + Groups[i] + ".abund.fasta", *(filehandles[Groups[i]+".abund"]));
 941                         }
 942
 943                         //open input file
 944                         ifstream in;
 945                         m->openInputFile(fastafile, in);
 946
 947                         while (!in.eof()) {
 948                                 if (m->control_pressed) { break; }
 949
 950                                 Sequence seq(in); m->gobble(in);
 951
 952                                 if (seq.getName() != "") {
 953                                         map<string, string>::iterator itNames = nameMap.find(seq.getName());
 954
 955                                         if (itNames == nameMap.end()) {
 956                                                 m->mothurOut(seq.getName() + " is not in your namesfile, ignoring."); m->mothurOutEndLine();
 957                                         }else{
 958                                                 vector<string> names;
 959                                                 m->splitAtComma(itNames->second, names);  //parses bin into individual sequence names
 960
 961                                                 string rareAbund;
 962                                                 if (rareNames.count(itNames->first) != 0) { //you are a rare name
 963                                                         rareAbund = ".rare";
 964                                                 }else{ //you are a abund name
 965                                                         rareAbund = ".abund";
 966                                                 }
 967
 968                                                 for (int i = 0; i < names.size(); i++) {
 969
 970                                                         string group = groupMap->getGroup(seq.getName());
 971
 972                                                         if (m->inUsersGroups(group, Groups)) { //only add if this is in a group we want
 973                                                                 seq.printSequence(*(filehandles[group+rareAbund]));
 974                                                         }else if(group == "not found") {
 975                                                                 m->mothurOut(seq.getName() + " is not in your groupfile. Ignoring."); m->mothurOutEndLine();
 976                                                         }
 977                                                 }
 978                                         }
 979                                 }
 980                         }
 981                         in.close();
 982
 983                         for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 984                                 (*(filehandles[it3->first])).close();
 985                                 outputNames.push_back(fileroot + tag + it3->first + ".fasta");  outputTypes["fasta"].push_back(fileroot + tag + it3->first + ".fasta");
 986                                 delete it3->second;
 987                         }
 988                 }
 989
 990                 return 0;
 991
 992         }
 993         catch(exception& e) {
 994                 m->errorOut(e, "SplitAbundCommand", "parseFasta");
 995                 exit(1);
 996         }
 997 }
 998 /**********************************************************************************************************************/
 999