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