makebiomcommand.cpp

   1 //
   2 //  makebiomcommand.cpp
   3 //  Mothur
   4 //
   5 //  Created by Sarah Westcott on 4/16/12.
   6 //  Copyright (c) 2012 Schloss Lab. All rights reserved.
   7 //
   8
   9 #include "makebiomcommand.h"
  10 #include "sharedrabundvector.h"
  11 #include "inputdata.h"
  12 #include "sharedutilities.h"
  13
  14 //taken from http://biom-format.org/documentation/biom_format.html
  15 /* Minimal Sparse
  16  {
  17  "id":null,
  18  "format": "Biological Observation Matrix 0.9.1",
  19  "format_url": "http://biom-format.org",
  20  "type": "OTU table",
  21  "generated_by": "QIIME revision 1.4.0-dev",
  22  "date": "2011-12-19T19:00:00",
  23  "rows":[
  24  {"id":"GG_OTU_1", "metadata":null},
  25  {"id":"GG_OTU_2", "metadata":null},
  26  {"id":"GG_OTU_3", "metadata":null},
  27  {"id":"GG_OTU_4", "metadata":null},
  28  {"id":"GG_OTU_5", "metadata":null}
  29  ],
  30  "columns": [
  31  {"id":"Sample1", "metadata":null},
  32  {"id":"Sample2", "metadata":null},
  33  {"id":"Sample3", "metadata":null},
  34  {"id":"Sample4", "metadata":null},
  35  {"id":"Sample5", "metadata":null},
  36  {"id":"Sample6", "metadata":null}
  37  ],
  38  "matrix_type": "sparse",
  39  "matrix_element_type": "int",
  40  "shape": [5, 6],
  41  "data":[[0,2,1],
  42  [1,0,5],
  43  [1,1,1],
  44  [1,3,2],
  45  [1,4,3],
  46  [1,5,1],
  47  [2,2,1],
  48  [2,3,4],
  49  [2,4,2],
  50  [3,0,2],
  51  [3,1,1],
  52  [3,2,1],
  53  [3,5,1],
  54  [4,1,1],
  55  [4,2,1]
  56  ]
  57  }
  58  */
  59 /* Minimal dense
  60  {
  61  "id":null,
  62  "format": "Biological Observation Matrix 0.9.1",
  63  "format_url": "http://biom-format.org",
  64  "type": "OTU table",
  65  "generated_by": "QIIME revision 1.4.0-dev",
  66  "date": "2011-12-19T19:00:00",
  67  "rows":[
  68  {"id":"GG_OTU_1", "metadata":null},
  69  {"id":"GG_OTU_2", "metadata":null},
  70  {"id":"GG_OTU_3", "metadata":null},
  71  {"id":"GG_OTU_4", "metadata":null},
  72  {"id":"GG_OTU_5", "metadata":null}
  73  ],
  74  "columns": [
  75  {"id":"Sample1", "metadata":null},
  76  {"id":"Sample2", "metadata":null},
  77  {"id":"Sample3", "metadata":null},
  78  {"id":"Sample4", "metadata":null},
  79  {"id":"Sample5", "metadata":null},
  80  {"id":"Sample6", "metadata":null}
  81  ],
  82  "matrix_type": "dense",
  83  "matrix_element_type": "int",
  84  "shape": [5,6],
  85  "data":  [[0,0,1,0,0,0],
  86  [5,1,0,2,3,1],
  87  [0,0,1,4,2,0],
  88  [2,1,1,0,0,1],
  89  [0,1,1,0,0,0]]
  90  }
  91  */
  92 //**********************************************************************************************************************
  93 vector<string> MakeBiomCommand::setParameters(){
  94         try {
  95                 CommandParameter pshared("shared", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pshared);
  96         CommandParameter pcontaxonomy("contaxonomy", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pcontaxonomy);
  97         CommandParameter pmetadata("metadata", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pmetadata);
  98                 CommandParameter pgroups("groups", "String", "", "", "", "", "",false,false); parameters.push_back(pgroups);
  99                 CommandParameter plabel("label", "String", "", "", "", "", "",false,false); parameters.push_back(plabel);
 100                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
 101                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
 102         CommandParameter pmatrixtype("matrixtype", "Multiple", "sparse-dense", "sparse", "", "", "",false,false); parameters.push_back(pmatrixtype);
 103
 104                 vector<string> myArray;
 105                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
 106                 return myArray;
 107         }
 108         catch(exception& e) {
 109                 m->errorOut(e, "MakeBiomCommand", "setParameters");
 110                 exit(1);
 111         }
 112 }
 113 //**********************************************************************************************************************
 114 string MakeBiomCommand::getHelpString(){
 115         try {
 116                 string helpString = "";
 117                 helpString += "The make.biom command parameters are shared, contaxonomy, metadata, groups, matrixtype and label.  shared is required, unless you have a valid current file.\n";
 118                 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";
 119                 helpString += "The label parameter allows you to select what distance levels you would like, and are also separated by dashes.\n";
 120                 helpString += "The matrixtype parameter allows you to select what type you would like to make. Choices are sparse and dense, default is sparse.\n";
 121         helpString += "The contaxonomy file is the taxonomy file outputted by classify.otu(list=yourListfile, taxonomy=yourTaxonomyFile). Be SURE that the you are the constaxonomy file distance matches the shared file distance.  ie, for *.0.03.cons.taxonomy set label=0.03. Mothur is smart enough to handle shared files that have been subsampled. It is used to assign taxonomy information to the metadata of rows.\n";
 122         helpString += "The metadata parameter is used to provide experimental parameters to the columns.  Things like 'sample1 gut human_gut'. \n";
 123                 helpString += "The make.biom command should be in the following format: make.biom(shared=yourShared, groups=yourGroups, label=yourLabels).\n";
 124                 helpString += "Example make.biom(shared=abrecovery.an.shared, groups=A-B-C).\n";
 125                 helpString += "The default value for groups is all the groups in your groupfile, and all labels in your inputfile will be used.\n";
 126                 helpString += "The make.biom command outputs a .biom file.\n";
 127                 helpString += "Note: No spaces between parameter labels (i.e. groups), '=' and parameters (i.e.yourGroups).\n";
 128                 return helpString;
 129         }
 130         catch(exception& e) {
 131                 m->errorOut(e, "MakeBiomCommand", "getHelpString");
 132                 exit(1);
 133         }
 134 }
 135 //**********************************************************************************************************************
 136 string MakeBiomCommand::getOutputFileNameTag(string type, string inputName=""){
 137         try {
 138         string outputFileName = "";
 139                 map<string, vector<string> >::iterator it;
 140
 141         //is this a type this command creates
 142         it = outputTypes.find(type);
 143         if (it == outputTypes.end()) {  m->mothurOut("[ERROR]: this command doesn't create a " + type + " output file.\n"); }
 144         else {
 145             if (type == "biom")             {   outputFileName =  "biom";       }
 146             else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true;  }
 147         }
 148         return outputFileName;
 149         }
 150         catch(exception& e) {
 151                 m->errorOut(e, "MakeBiomCommand", "getOutputFileNameTag");
 152                 exit(1);
 153         }
 154 }
 155
 156 //**********************************************************************************************************************
 157 MakeBiomCommand::MakeBiomCommand(){
 158         try {
 159                 abort = true; calledHelp = true;
 160                 setParameters();
 161                 vector<string> tempOutNames;
 162                 outputTypes["biom"] = tempOutNames;
 163         }
 164         catch(exception& e) {
 165                 m->errorOut(e, "MakeBiomCommand", "MakeBiomCommand");
 166                 exit(1);
 167         }
 168 }
 169 //**********************************************************************************************************************
 170 MakeBiomCommand::MakeBiomCommand(string option) {
 171         try {
 172                 abort = false; calledHelp = false;
 173                 allLines = 1;
 174
 175                 //allow user to run help
 176                 if(option == "help") { help(); abort = true; calledHelp = true; }
 177                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 178
 179                 else {
 180                         vector<string> myArray = setParameters();
 181
 182                         OptionParser parser(option);
 183                         map<string,string> parameters = parser.getParameters();
 184                         map<string,string>::iterator it;
 185
 186                         ValidParameters validParameter;
 187
 188                         //check to make sure all parameters are valid for command
 189                         for (it = parameters.begin(); it != parameters.end(); it++) {
 190                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 191                         }
 192
 193                         //initialize outputTypes
 194                         vector<string> tempOutNames;
 195                         outputTypes["biom"] = tempOutNames;
 196
 197                         //if the user changes the input directory command factory will send this info to us in the output parameter
 198                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 199                         if (inputDir == "not found"){   inputDir = "";          }
 200                         else {
 201                                 string path;
 202                                 it = parameters.find("shared");
 203                                 //user has given a template file
 204                                 if(it != parameters.end()){
 205                                         path = m->hasPath(it->second);
 206                                         //if the user has not given a path then, add inputdir. else leave path alone.
 207                                         if (path == "") {       parameters["shared"] = inputDir + it->second;           }
 208                                 }
 209
 210                 it = parameters.find("contaxonomy");
 211                                 //user has given a template file
 212                                 if(it != parameters.end()){
 213                                         path = m->hasPath(it->second);
 214                                         //if the user has not given a path then, add inputdir. else leave path alone.
 215                                         if (path == "") {       parameters["contaxonomy"] = inputDir + it->second;              }
 216                                 }
 217
 218                 it = parameters.find("metadata");
 219                                 //user has given a template file
 220                                 if(it != parameters.end()){
 221                                         path = m->hasPath(it->second);
 222                                         //if the user has not given a path then, add inputdir. else leave path alone.
 223                                         if (path == "") {       parameters["metadata"] = inputDir + it->second;         }
 224                                 }
 225                         }
 226
 227                         //get shared file
 228                         sharedfile = validParameter.validFile(parameters, "shared", true);
 229                         if (sharedfile == "not open") { sharedfile = ""; abort = true; }
 230                         else if (sharedfile == "not found") {
 231                                 //if there is a current shared file, use it
 232                                 sharedfile = m->getSharedFile();
 233                                 if (sharedfile != "") { m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); }
 234                                 else {  m->mothurOut("You have no current sharedfile and the shared parameter is required."); m->mothurOutEndLine(); abort = true; }
 235                         }else { m->setSharedFile(sharedfile); }
 236
 237
 238                         //if the user changes the output directory command factory will send this info to us in the output parameter
 239                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = m->hasPath(sharedfile);             }
 240
 241             contaxonomyfile = validParameter.validFile(parameters, "contaxonomy", true);
 242                         if (contaxonomyfile == "not found") {  contaxonomyfile = "";  }
 243                         else if (contaxonomyfile == "not open") { contaxonomyfile = ""; abort = true; }
 244
 245             metadatafile = validParameter.validFile(parameters, "metadata", true);
 246                         if (metadatafile == "not found") {  metadatafile = "";  }
 247                         else if (metadatafile == "not open") { metadatafile = ""; abort = true; }
 248
 249                         //check for optional parameter and set defaults
 250                         // ...at some point should added some additional type checking...
 251                         label = validParameter.validFile(parameters, "label", false);
 252                         if (label == "not found") { label = ""; }
 253                         else {
 254                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 255                                 else { allLines = 1;  }
 256                         }
 257
 258                         groups = validParameter.validFile(parameters, "groups", false);
 259                         if (groups == "not found") { groups = ""; }
 260                         else {
 261                                 m->splitAtDash(groups, Groups);
 262                                 m->setGroups(Groups);
 263                         }
 264
 265             if ((contaxonomyfile != "") && (labels.size() > 1)) { m->mothurOut("[ERROR]: the contaxonomy parameter cannot be used with multiple labels."); m->mothurOutEndLine(); abort = true; }
 266
 267                         format = validParameter.validFile(parameters, "matrixtype", false);                             if (format == "not found") { format = "sparse"; }
 268
 269                         if ((format != "sparse") && (format != "dense")) {
 270                                 m->mothurOut(format + " is not a valid option for the matrixtype parameter. Options are sparse and dense."); m->mothurOutEndLine(); abort = true;
 271                         }
 272                 }
 273
 274         }
 275         catch(exception& e) {
 276                 m->errorOut(e, "MakeBiomCommand", "MakeBiomCommand");
 277                 exit(1);
 278         }
 279 }
 280 //**********************************************************************************************************************
 281
 282 int MakeBiomCommand::execute(){
 283         try {
 284
 285                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 286
 287                 InputData input(sharedfile, "sharedfile");
 288                 vector<SharedRAbundVector*> lookup = input.getSharedRAbundVectors();
 289                 string lastLabel = lookup[0]->getLabel();
 290
 291         //if user did not specify a label, then use first one
 292         if ((contaxonomyfile != "") && (labels.size() == 0)) {
 293             allLines = 0;
 294             labels.insert(lastLabel);
 295         }
 296
 297         getSampleMetaData(lookup);
 298
 299                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 300                 set<string> processedLabels;
 301                 set<string> userLabels = labels;
 302
 303                 //as long as you are not at the end of the file or done wih the lines you want
 304                 while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 305
 306                         if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]); } for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }  return 0; }
 307
 308                         if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){
 309
 310                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 311                                 getBiom(lookup);
 312
 313                                 processedLabels.insert(lookup[0]->getLabel());
 314                                 userLabels.erase(lookup[0]->getLabel());
 315                         }
 316
 317                         if ((m->anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 318                                 string saveLabel = lookup[0]->getLabel();
 319
 320                                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 321                                 lookup = input.getSharedRAbundVectors(lastLabel);
 322                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 323
 324                                 getBiom(lookup);
 325
 326                                 processedLabels.insert(lookup[0]->getLabel());
 327                                 userLabels.erase(lookup[0]->getLabel());
 328
 329                                 //restore real lastlabel to save below
 330                                 lookup[0]->setLabel(saveLabel);
 331                         }
 332
 333                         lastLabel = lookup[0]->getLabel();
 334
 335                         //prevent memory leak and get next set
 336                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i]; lookup[i] = NULL; }
 337                         lookup = input.getSharedRAbundVectors();
 338                 }
 339
 340         if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]); }  return 0; }
 341
 342                 //output error messages about any remaining user labels
 343                 set<string>::iterator it;
 344                 bool needToRun = false;
 345                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 346                         m->mothurOut("Your file does not include the label " + *it);
 347                         if (processedLabels.count(lastLabel) != 1) {
 348                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 349                                 needToRun = true;
 350                         }else {
 351                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 352                         }
 353                 }
 354
 355                 //run last label if you need to
 356                 if (needToRun == true)  {
 357                         for (int i = 0; i < lookup.size(); i++) { if (lookup[i] != NULL) { delete lookup[i]; } }
 358                         lookup = input.getSharedRAbundVectors(lastLabel);
 359
 360                         m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 361             getBiom(lookup);
 362
 363                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 364                 }
 365
 366         if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]); }  return 0; }
 367
 368         //set sabund file as new current sabundfile
 369         string current = "";
 370                 itTypes = outputTypes.find("biom");
 371                 if (itTypes != outputTypes.end()) {
 372                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setBiomFile(current); }
 373                 }
 374
 375
 376                 m->mothurOutEndLine();
 377                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 378                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 379                 m->mothurOutEndLine();
 380
 381                 return 0;
 382         }
 383         catch(exception& e) {
 384                 m->errorOut(e, "MakeBiomCommand", "execute");
 385                 exit(1);
 386         }
 387 }
 388 //**********************************************************************************************************************
 389 int MakeBiomCommand::getBiom(vector<SharedRAbundVector*>& lookup){
 390         try {
 391
 392         string outputFileName = outputDir + m->getRootName(m->getSimpleName(sharedfile)) + lookup[0]->getLabel() + "." + getOutputFileNameTag("biom");
 393                 ofstream out;
 394                 m->openOutputFile(outputFileName, out);
 395                 outputNames.push_back(outputFileName); outputTypes["biom"].push_back(outputFileName);
 396
 397         string mothurString = "mothur" + toString(m->getVersion());
 398         time_t rawtime;
 399         struct tm * timeinfo;
 400         time ( &rawtime );
 401         timeinfo = localtime ( &rawtime );
 402         string dateString = asctime (timeinfo);
 403         int pos = dateString.find('\n');
 404         if (pos != string::npos) { dateString = dateString.substr(0, pos);}
 405         string spaces = "      ";
 406
 407         //standard
 408         out << "{\n" + spaces + "\"id\":\"" + sharedfile + "-" + lookup[0]->getLabel() + "\",\n" + spaces + "\"format\": \"Biological Observation Matrix 0.9.1\",\n" + spaces + "\"format_url\": \"http://biom-format.org\",\n";
 409         out << spaces + "\"type\": \"OTU table\",\n" + spaces + "\"generated_by\": \"" << mothurString << "\",\n" + spaces + "\"date\": \"" << dateString << "\",\n";
 410
 411         vector<string> metadata = getMetaData(lookup);
 412
 413         if (m->control_pressed) {  out.close(); return 0; }
 414
 415         //get row info
 416         /*"rows":[
 417                 {"id":"GG_OTU_1", "metadata":null},
 418                 {"id":"GG_OTU_2", "metadata":null},
 419                 {"id":"GG_OTU_3", "metadata":null},
 420                 {"id":"GG_OTU_4", "metadata":null},
 421                 {"id":"GG_OTU_5", "metadata":null}
 422                 ],*/
 423         out << spaces + "\"rows\":[\n";
 424         string rowFront = spaces + spaces + "{\"id\":\"";
 425         string rowBack = "\", \"metadata\":";
 426         for (int i = 0; i < m->currentBinLabels.size()-1; i++) {
 427             if (m->control_pressed) {  out.close(); return 0; }
 428             out << rowFront << m->currentBinLabels[i] << rowBack << metadata[i] << "},\n";
 429         }
 430         out << rowFront << m->currentBinLabels[(m->currentBinLabels.size()-1)] << rowBack << metadata[(m->currentBinLabels.size()-1)] << "}\n" + spaces + "],\n";
 431
 432         //get column info
 433         /*"columns": [
 434                     {"id":"Sample1", "metadata":null},
 435                     {"id":"Sample2", "metadata":null},
 436                     {"id":"Sample3", "metadata":null},
 437                     {"id":"Sample4", "metadata":null},
 438                     {"id":"Sample5", "metadata":null},
 439                     {"id":"Sample6", "metadata":null}
 440                     ],*/
 441
 442         string colBack = "\", \"metadata\":";
 443         out << spaces + "\"columns\":[\n";
 444         for (int i = 0; i < lookup.size()-1; i++) {
 445             if (m->control_pressed) {  out.close(); return 0; }
 446             out << rowFront << lookup[i]->getGroup() << colBack << sampleMetadata[i] << "},\n";
 447         }
 448         out << rowFront << lookup[(lookup.size()-1)]->getGroup() << colBack << sampleMetadata[lookup.size()-1] << "}\n" + spaces + "],\n";
 449
 450         out << spaces + "\"matrix_type\": \"" << format << "\",\n" + spaces + "\"matrix_element_type\": \"int\",\n";
 451         out <<  spaces + "\"shape\": [" << m->currentBinLabels.size() << "," << lookup.size() << "],\n";
 452         out << spaces + "\"data\":  [";
 453
 454         vector<string> dataRows;
 455         if (format == "sparse") {
 456             /*"data":[[0,2,1],
 457              [1,0,5],
 458              [1,1,1],
 459              [1,3,2],
 460              [1,4,3],
 461              [1,5,1],
 462              [2,2,1],
 463              [2,3,4],
 464              [2,4,2],
 465              [3,0,2],
 466              [3,1,1],
 467              [3,2,1],
 468              [3,5,1],
 469              [4,1,1],
 470              [4,2,1]
 471              ]*/
 472             string output = "";
 473             for (int i = 0; i < lookup[0]->getNumBins(); i++) {
 474
 475                 if (m->control_pressed) { out.close(); return 0; }
 476
 477                 for (int j = 0; j < lookup.size(); j++) {
 478                     string binInfo = "[" + toString(i) + "," + toString(j) + "," + toString(lookup[j]->getAbundance(i)) + "]";
 479                     //only print non zero values
 480                     if (lookup[j]->getAbundance(i) != 0) { dataRows.push_back(binInfo); }
 481                 }
 482             }
 483         }else {
 484
 485             /* "matrix_type": "dense",
 486              "matrix_element_type": "int",
 487              "shape": [5,6],
 488              "data":  [[0,0,1,0,0,0],
 489              [5,1,0,2,3,1],
 490              [0,0,1,4,2,0],
 491              [2,1,1,0,0,1],
 492              [0,1,1,0,0,0]]*/
 493
 494             for (int i = 0; i < lookup[0]->getNumBins(); i++) {
 495
 496                 if (m->control_pressed) { out.close(); return 0; }
 497
 498                 string binInfo = "[";
 499                 for (int j = 0; j < lookup.size()-1; j++) {
 500                     binInfo += toString(lookup[j]->getAbundance(i)) + ",";
 501                 }
 502                 binInfo += toString(lookup[lookup.size()-1]->getAbundance(i)) + "]";
 503                 dataRows.push_back(binInfo);
 504             }
 505         }
 506
 507         for (int i = 0; i < dataRows.size()-1; i++) {
 508             out << dataRows[i] << ",\n" + spaces  + spaces;
 509         }
 510         out << dataRows[dataRows.size()-1] << "]\n";
 511
 512         out << "}\n";
 513         out.close();
 514
 515         return 0;
 516     }
 517         catch(exception& e) {
 518                 m->errorOut(e, "MakeBiomCommand", "getBiom");
 519                 exit(1);
 520         }
 521 }
 522 //**********************************************************************************************************************
 523 vector<string> MakeBiomCommand::getMetaData(vector<SharedRAbundVector*>& lookup){
 524         try {
 525         vector<string> metadata;
 526
 527         if (contaxonomyfile == "") { for (int i = 0; i < lookup[0]->getNumBins(); i++) {  metadata.push_back("null");  } }
 528         else {
 529
 530             //read constaxonomy file storing in a map, otulabel -> taxonomy
 531             //constaxonomy file will most likely contain more labels than the shared file, because sharedfile could have been subsampled.
 532             ifstream in;
 533             m->openInputFile(contaxonomyfile, in);
 534
 535             //grab headers
 536             m->getline(in); m->gobble(in);
 537
 538             string otuLabel, tax;
 539             int size;
 540             vector<string> otuLabels;
 541             vector<string> taxs;
 542             while (!in.eof()) {
 543
 544                 if (m->control_pressed) { in.close(); return metadata; }
 545
 546                 in >> otuLabel >> size >> tax; m->gobble(in);
 547
 548                 otuLabels.push_back(otuLabel);
 549                 taxs.push_back(tax);
 550             }
 551             in.close();
 552
 553             //should the labels be Otu001 or PhyloType001
 554             string firstBin = m->currentBinLabels[0];
 555             string binTag = "Otu";
 556             if ((firstBin.find("Otu")) == string::npos) { binTag = "PhyloType";  }
 557
 558             //convert list file bin labels to shared file bin labels
 559             //parse tax strings
 560             //save in map
 561             map<string, string> labelTaxMap;
 562             string snumBins = toString(otuLabels.size());
 563             for (int i = 0; i < otuLabels.size(); i++) {
 564
 565                 if (m->control_pressed) { return metadata; }
 566
 567                 //if there is a bin label use it otherwise make one
 568                 if (m->isContainingOnlyDigits(otuLabels[i])) {
 569                     string binLabel = binTag;
 570                     string sbinNumber = otuLabels[i];
 571                     if (sbinNumber.length() < snumBins.length()) {
 572                         int diff = snumBins.length() - sbinNumber.length();
 573                         for (int h = 0; h < diff; h++) { binLabel += "0"; }
 574                     }
 575                     binLabel += sbinNumber;
 576                     labelTaxMap[binLabel] = taxs[i];
 577                 }else {  labelTaxMap[otuLabels[i]] = taxs[i]; }
 578             }
 579
 580
 581             //{"taxonomy":["k__Bacteria", "p__Proteobacteria", "c__Gammaproteobacteria", "o__Enterobacteriales", "f__Enterobacteriaceae", "g__Escherichia", "s__"]}
 582
 583             //traverse the binLabels forming the metadata strings and saving them
 584             //make sure to sanity check
 585             map<string, string>::iterator it;
 586             for (int i = 0; i < m->currentBinLabels.size(); i++) {
 587
 588                 if (m->control_pressed) { return metadata; }
 589
 590                 it = labelTaxMap.find(m->currentBinLabels[i]);
 591
 592                 if (it == labelTaxMap.end()) { m->mothurOut("[ERROR]: can't find taxonomy information for " + m->currentBinLabels[i] + ".\n"); m->control_pressed = true; }
 593                 else {
 594                     vector<string> bootstrapValues;
 595                     string data = "{\"taxonomy\":[";
 596
 597                     vector<string> scores;
 598                     vector<string> taxonomies = parseTax(it->second, scores);
 599
 600                     for (int j = 0; j < taxonomies.size()-1; j ++) { data += "\"" + taxonomies[j] + "\", "; }
 601                     data += "\"" + taxonomies[taxonomies.size()-1] + "\"]";
 602
 603                     //add bootstrap values if available
 604                     if (scores[0] != "null") {
 605                         data += ", \"bootstrap\":[";
 606
 607                         for (int j = 0; j < scores.size()-1; j ++) { data += scores[j] + ", "; }
 608                         data += scores[scores.size()-1] + "]";
 609
 610                     }
 611                     data += "}";
 612
 613                     metadata.push_back(data);
 614                 }
 615             }
 616         }
 617
 618         return metadata;
 619
 620     }
 621         catch(exception& e) {
 622                 m->errorOut(e, "MakeBiomCommand", "getMetadata");
 623                 exit(1);
 624         }
 625
 626 }
 627 //**********************************************************************************************************************
 628 int MakeBiomCommand::getSampleMetaData(vector<SharedRAbundVector*>& lookup){
 629         try {
 630
 631         if (metadatafile == "") { for (int i = 0; i < lookup.size(); i++) {  sampleMetadata.push_back("null");  } }
 632         else {
 633             ifstream in;
 634             m->openInputFile(metadatafile, in);
 635
 636             vector<string> groupNames, metadataLabels;
 637             map<string, vector<string> > lines;
 638
 639             string headerLine = m->getline(in); m->gobble(in);
 640             vector<string> pieces = m->splitWhiteSpace(headerLine);
 641
 642             //save names of columns you are reading
 643             for (int i = 1; i < pieces.size(); i++) {
 644                 metadataLabels.push_back(pieces[i]);
 645             }
 646             int count = metadataLabels.size();
 647
 648             vector<string> groups = m->getGroups();
 649
 650             //read rest of file
 651             while (!in.eof()) {
 652
 653                 if (m->control_pressed) { in.close(); return 0; }
 654
 655                 string group = "";
 656                 in >> group; m->gobble(in);
 657                 groupNames.push_back(group);
 658
 659                 string line = m->getline(in); m->gobble(in);
 660                 vector<string> thisPieces = m->splitWhiteSpaceWithQuotes(line);
 661
 662                 if (thisPieces.size() != count) { m->mothurOut("[ERROR]: expected " + toString(count) + " items of data for sample " + group + " read " + toString(thisPieces.size()) + ", quitting.\n"); }
 663                 else {  if (m->inUsersGroups(group, groups)) { lines[group] = thisPieces; } }
 664
 665                 m->gobble(in);
 666             }
 667             in.close();
 668
 669             map<string, vector<string> >::iterator it;
 670             for (int i = 0; i < lookup.size(); i++) {
 671
 672                 if (m->control_pressed) { return 0; }
 673
 674                 it = lines.find(lookup[i]->getGroup());
 675
 676                 if (it == lines.end()) { m->mothurOut("[ERROR]: can't find metadata information for " + lookup[i]->getGroup() + ", quitting.\n"); m->control_pressed = true; }
 677                 else {
 678                     vector<string> values = it->second;
 679
 680                     string data = "{";
 681                     for (int j = 0; j < metadataLabels.size()-1; j++) {
 682                         values[j] = m->removeQuotes(values[j]);
 683                         data += "\"" + metadataLabels[j] + "\":\"" + values[j] + "\", ";
 684                     }
 685                     values[metadataLabels.size()-1] = m->removeQuotes(values[metadataLabels.size()-1]);
 686                     data += "\"" + metadataLabels[metadataLabels.size()-1] + "\":\"" + values[metadataLabels.size()-1] + "\"}";
 687                     sampleMetadata.push_back(data);
 688                 }
 689             }
 690         }
 691
 692         return 0;
 693
 694     }
 695         catch(exception& e) {
 696                 m->errorOut(e, "MakeBiomCommand", "getSampleMetaData");
 697                 exit(1);
 698         }
 699
 700 }
 701
 702 /**************************************************************************************************/
 703 //returns {Bacteria, Bacteroidetes, ..} and scores is filled with {100, 98, ...} or {null, null, null}
 704 vector<string> MakeBiomCommand::parseTax(string tax, vector<string>& scores) {
 705         try {
 706
 707                 string taxon;
 708         vector<string> taxs;
 709
 710                 while (tax.find_first_of(';') != -1) {
 711
 712                         if (m->control_pressed) { return taxs; }
 713
 714                         //get taxon
 715                         taxon = tax.substr(0,tax.find_first_of(';'));
 716
 717                         int pos = taxon.find_last_of('(');
 718                         if (pos != -1) {
 719                                 //is it a number?
 720                                 int pos2 = taxon.find_last_of(')');
 721                                 if (pos2 != -1) {
 722                                         string confidenceScore = taxon.substr(pos+1, (pos2-(pos+1)));
 723                                         if (m->isNumeric1(confidenceScore)) {
 724                                                 taxon = taxon.substr(0, pos); //rip off confidence
 725                         scores.push_back(confidenceScore);
 726                                         }else{ scores.push_back("null"); }
 727                                 }
 728                         }else{ scores.push_back("null"); }
 729
 730             //strip "" if they are there
 731             pos = taxon.find("\"");
 732             if (pos != string::npos) {
 733                 string newTax = "";
 734                 for (int k = 0; k < taxon.length(); k++) {
 735                     if (taxon[k] != '\"') { newTax += taxon[k]; }
 736                 }
 737                 taxon = newTax;
 738             }
 739
 740             //look for bootstrap value
 741                         taxs.push_back(taxon);
 742             tax = tax.substr(tax.find_first_of(';')+1, tax.length());
 743                 }
 744
 745                 return taxs;
 746         }
 747         catch(exception& e) {
 748                 m->errorOut(e, "MakeBiomCommand", "parseTax");
 749                 exit(1);
 750         }
 751 }
 752
 753 //**********************************************************************************************************************
 754
 755
 756