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("constaxonomy", "InputTypes", "", "", "none", "none", "none","",false,false); parameters.push_back(pcontaxonomy);
  97         //CommandParameter preference("referencetax", "InputTypes", "", "", "none", "none", "none","",false,false); parameters.push_back(preference);
  98         CommandParameter pmetadata("metadata", "InputTypes", "", "", "none", "none", "none","",false,false); parameters.push_back(pmetadata);
  99                 CommandParameter pgroups("groups", "String", "", "", "", "", "","",false,false); parameters.push_back(pgroups);
 100                 CommandParameter plabel("label", "String", "", "", "", "", "","",false,false); parameters.push_back(plabel);
 101         //CommandParameter ppicrust("picrust", "Boolean", "", "F", "", "", "","",false,false); parameters.push_back(ppicrust);
 102         CommandParameter pinputdir("inputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(pinputdir);
 103                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(poutputdir);
 104         CommandParameter pmatrixtype("matrixtype", "Multiple", "sparse-dense", "sparse", "", "", "","",false,false); parameters.push_back(pmatrixtype);
 105
 106                 vector<string> myArray;
 107                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
 108                 return myArray;
 109         }
 110         catch(exception& e) {
 111                 m->errorOut(e, "MakeBiomCommand", "setParameters");
 112                 exit(1);
 113         }
 114 }
 115 //**********************************************************************************************************************
 116 string MakeBiomCommand::getHelpString(){
 117         try {
 118                 string helpString = "";
 119                 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"; //, picrust and referencetax
 120                 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";
 121                 helpString += "The label parameter allows you to select what distance levels you would like, and are also separated by dashes.\n";
 122                 helpString += "The matrixtype parameter allows you to select what type you would like to make. Choices are sparse and dense, default is sparse.\n";
 123         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";
 124         helpString += "The metadata parameter is used to provide experimental parameters to the columns.  Things like 'sample1 gut human_gut'. \n";
 125         //helpString += "The picrust parameter is used to indicate the biom file is for input to picrust.  NOTE: Picrust requires a greengenes taxonomy. \n";
 126         //helpString += "The referencetax parameter is used with the picrust parameter.  Picrust requires the name of the reference taxonomy sequence to be in the biom file. \n";
 127                 helpString += "The make.biom command should be in the following format: make.biom(shared=yourShared, groups=yourGroups, label=yourLabels).\n";
 128                 helpString += "Example make.biom(shared=abrecovery.an.shared, groups=A-B-C).\n";
 129                 helpString += "The default value for groups is all the groups in your groupfile, and all labels in your inputfile will be used.\n";
 130                 helpString += "The make.biom command outputs a .biom file.\n";
 131                 helpString += "Note: No spaces between parameter labels (i.e. groups), '=' and parameters (i.e.yourGroups).\n";
 132                 return helpString;
 133         }
 134         catch(exception& e) {
 135                 m->errorOut(e, "MakeBiomCommand", "getHelpString");
 136                 exit(1);
 137         }
 138 }
 139 //**********************************************************************************************************************
 140 string MakeBiomCommand::getOutputPattern(string type) {
 141     try {
 142         string pattern = "";
 143
 144         if (type == "biom") {  pattern = "[filename],[distance],biom"; }
 145         else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true;  }
 146
 147         return pattern;
 148     }
 149     catch(exception& e) {
 150         m->errorOut(e, "MakeBiomCommand", "getOutputPattern");
 151         exit(1);
 152     }
 153 }
 154
 155 //**********************************************************************************************************************
 156 MakeBiomCommand::MakeBiomCommand(){
 157         try {
 158                 abort = true; calledHelp = true;
 159                 setParameters();
 160                 vector<string> tempOutNames;
 161                 outputTypes["biom"] = tempOutNames;
 162         }
 163         catch(exception& e) {
 164                 m->errorOut(e, "MakeBiomCommand", "MakeBiomCommand");
 165                 exit(1);
 166         }
 167 }
 168 //**********************************************************************************************************************
 169 MakeBiomCommand::MakeBiomCommand(string option) {
 170         try {
 171                 abort = false; calledHelp = false;
 172                 allLines = 1;
 173
 174                 //allow user to run help
 175                 if(option == "help") { help(); abort = true; calledHelp = true; }
 176                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 177
 178                 else {
 179                         vector<string> myArray = setParameters();
 180
 181                         OptionParser parser(option);
 182                         map<string,string> parameters = parser.getParameters();
 183                         map<string,string>::iterator it;
 184
 185                         ValidParameters validParameter;
 186
 187                         //check to make sure all parameters are valid for command
 188                         for (it = parameters.begin(); it != parameters.end(); it++) {
 189                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 190                         }
 191
 192                         //initialize outputTypes
 193                         vector<string> tempOutNames;
 194                         outputTypes["biom"] = tempOutNames;
 195
 196                         //if the user changes the input directory command factory will send this info to us in the output parameter
 197                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 198                         if (inputDir == "not found"){   inputDir = "";          }
 199                         else {
 200                                 string path;
 201                                 it = parameters.find("shared");
 202                                 //user has given a template file
 203                                 if(it != parameters.end()){
 204                                         path = m->hasPath(it->second);
 205                                         //if the user has not given a path then, add inputdir. else leave path alone.
 206                                         if (path == "") {       parameters["shared"] = inputDir + it->second;           }
 207                                 }
 208
 209                 it = parameters.find("constaxonomy");
 210                                 //user has given a template file
 211                                 if(it != parameters.end()){
 212                                         path = m->hasPath(it->second);
 213                                         //if the user has not given a path then, add inputdir. else leave path alone.
 214                                         if (path == "") {       parameters["constaxonomy"] = inputDir + it->second;             }
 215                                 }
 216
 217                 it = parameters.find("referencetax");
 218                                 //user has given a template file
 219                                 if(it != parameters.end()){
 220                                         path = m->hasPath(it->second);
 221                                         //if the user has not given a path then, add inputdir. else leave path alone.
 222                                         if (path == "") {       parameters["referencetax"] = inputDir + it->second;             }
 223                                 }
 224
 225                 it = parameters.find("metadata");
 226                                 //user has given a template file
 227                                 if(it != parameters.end()){
 228                                         path = m->hasPath(it->second);
 229                                         //if the user has not given a path then, add inputdir. else leave path alone.
 230                                         if (path == "") {       parameters["metadata"] = inputDir + it->second;         }
 231                                 }
 232                         }
 233
 234                         //get shared file
 235                         sharedfile = validParameter.validFile(parameters, "shared", true);
 236                         if (sharedfile == "not open") { sharedfile = ""; abort = true; }
 237                         else if (sharedfile == "not found") {
 238                                 //if there is a current shared file, use it
 239                                 sharedfile = m->getSharedFile();
 240                                 if (sharedfile != "") { m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); }
 241                                 else {  m->mothurOut("You have no current sharedfile and the shared parameter is required."); m->mothurOutEndLine(); abort = true; }
 242                         }else { m->setSharedFile(sharedfile); }
 243
 244
 245                         //if the user changes the output directory command factory will send this info to us in the output parameter
 246                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = m->hasPath(sharedfile);             }
 247
 248             contaxonomyfile = validParameter.validFile(parameters, "constaxonomy", true);
 249                         if (contaxonomyfile == "not found") {  contaxonomyfile = "";  }
 250                         else if (contaxonomyfile == "not open") { contaxonomyfile = ""; abort = true; }
 251
 252             //referenceTax = validParameter.validFile(parameters, "referencetax", true);
 253                         //if (referenceTax == "not found") {  referenceTax = "";  }
 254                         //else if (referenceTax == "not open") { referenceTax = ""; abort = true; }
 255
 256             metadatafile = validParameter.validFile(parameters, "metadata", true);
 257                         if (metadatafile == "not found") {  metadatafile = "";  }
 258                         else if (metadatafile == "not open") { metadatafile = ""; abort = true; }
 259
 260                         //check for optional parameter and set defaults
 261                         // ...at some point should added some additional type checking...
 262                         label = validParameter.validFile(parameters, "label", false);
 263                         if (label == "not found") { label = ""; }
 264                         else {
 265                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 266                                 else { allLines = 1;  }
 267                         }
 268
 269             //string temp = validParameter.validFile(parameters, "picrust", false);                     if (temp == "not found"){       temp = "f";                             }
 270                         //picrust = m->isTrue(temp);
 271             //if (picrust && ((contaxonomyfile == "") || (referenceTax == ""))) {
 272                 //m->mothurOut("[ERROR]: the picrust parameter requires a consensus taxonomy with greengenes taxonomy the reference."); m->mothurOutEndLine(); abort = true;
 273            //}
 274             picrust=false;
 275
 276                         groups = validParameter.validFile(parameters, "groups", false);
 277                         if (groups == "not found") { groups = ""; }
 278                         else {
 279                                 m->splitAtDash(groups, Groups);
 280                                 m->setGroups(Groups);
 281                         }
 282
 283             if ((contaxonomyfile != "") && (labels.size() > 1)) { m->mothurOut("[ERROR]: the contaxonomy parameter cannot be used with multiple labels."); m->mothurOutEndLine(); abort = true; }
 284
 285                         format = validParameter.validFile(parameters, "matrixtype", false);                             if (format == "not found") { format = "sparse"; }
 286
 287                         if ((format != "sparse") && (format != "dense")) {
 288                                 m->mothurOut(format + " is not a valid option for the matrixtype parameter. Options are sparse and dense."); m->mothurOutEndLine(); abort = true;
 289                         }
 290                 }
 291
 292         }
 293         catch(exception& e) {
 294                 m->errorOut(e, "MakeBiomCommand", "MakeBiomCommand");
 295                 exit(1);
 296         }
 297 }
 298 //**********************************************************************************************************************
 299
 300 int MakeBiomCommand::execute(){
 301         try {
 302
 303                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 304
 305                 InputData input(sharedfile, "sharedfile");
 306                 vector<SharedRAbundVector*> lookup = input.getSharedRAbundVectors();
 307                 string lastLabel = lookup[0]->getLabel();
 308
 309         //if user did not specify a label, then use first one
 310         if ((contaxonomyfile != "") && (labels.size() == 0)) {
 311             allLines = 0;
 312             labels.insert(lastLabel);
 313         }
 314
 315         getSampleMetaData(lookup);
 316
 317                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 318                 set<string> processedLabels;
 319                 set<string> userLabels = labels;
 320
 321                 //as long as you are not at the end of the file or done wih the lines you want
 322                 while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 323
 324                         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; }
 325
 326                         if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){
 327
 328                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 329                                 getBiom(lookup);
 330
 331                                 processedLabels.insert(lookup[0]->getLabel());
 332                                 userLabels.erase(lookup[0]->getLabel());
 333                         }
 334
 335                         if ((m->anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 336                                 string saveLabel = lookup[0]->getLabel();
 337
 338                                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 339                                 lookup = input.getSharedRAbundVectors(lastLabel);
 340                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 341
 342                                 getBiom(lookup);
 343
 344                                 processedLabels.insert(lookup[0]->getLabel());
 345                                 userLabels.erase(lookup[0]->getLabel());
 346
 347                                 //restore real lastlabel to save below
 348                                 lookup[0]->setLabel(saveLabel);
 349                         }
 350
 351                         lastLabel = lookup[0]->getLabel();
 352
 353                         //prevent memory leak and get next set
 354                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i]; lookup[i] = NULL; }
 355                         lookup = input.getSharedRAbundVectors();
 356                 }
 357
 358         if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]); }  return 0; }
 359
 360                 //output error messages about any remaining user labels
 361                 set<string>::iterator it;
 362                 bool needToRun = false;
 363                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 364                         m->mothurOut("Your file does not include the label " + *it);
 365                         if (processedLabels.count(lastLabel) != 1) {
 366                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 367                                 needToRun = true;
 368                         }else {
 369                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 370                         }
 371                 }
 372
 373                 //run last label if you need to
 374                 if (needToRun == true)  {
 375                         for (int i = 0; i < lookup.size(); i++) { if (lookup[i] != NULL) { delete lookup[i]; } }
 376                         lookup = input.getSharedRAbundVectors(lastLabel);
 377
 378                         m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 379             getBiom(lookup);
 380
 381                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 382                 }
 383
 384         if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]); }  return 0; }
 385
 386         //set sabund file as new current sabundfile
 387         string current = "";
 388                 itTypes = outputTypes.find("biom");
 389                 if (itTypes != outputTypes.end()) {
 390                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setBiomFile(current); }
 391                 }
 392
 393
 394                 m->mothurOutEndLine();
 395                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 396                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 397                 m->mothurOutEndLine();
 398
 399                 return 0;
 400         }
 401         catch(exception& e) {
 402                 m->errorOut(e, "MakeBiomCommand", "execute");
 403                 exit(1);
 404         }
 405 }
 406 //**********************************************************************************************************************
 407 int MakeBiomCommand::getBiom(vector<SharedRAbundVector*>& lookup){
 408         try {
 409         map<string, string> variables;
 410         variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(sharedfile));
 411         variables["[distance]"] = lookup[0]->getLabel();
 412         string outputFileName = getOutputFileName("biom",variables);
 413                 ofstream out;
 414                 m->openOutputFile(outputFileName, out);
 415                 outputNames.push_back(outputFileName); outputTypes["biom"].push_back(outputFileName);
 416
 417         string mothurString = "mothur" + toString(m->getVersion());
 418         time_t rawtime;
 419         struct tm * timeinfo;
 420         time ( &rawtime );
 421         timeinfo = localtime ( &rawtime );
 422         string dateString = asctime (timeinfo);
 423         int pos = dateString.find('\n');
 424         if (pos != string::npos) { dateString = dateString.substr(0, pos);}
 425         string spaces = "      ";
 426
 427         //standard
 428         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";
 429         out << spaces + "\"type\": \"OTU table\",\n" + spaces + "\"generated_by\": \"" << mothurString << "\",\n" + spaces + "\"date\": \"" << dateString << "\",\n";
 430
 431         int numBins = lookup[0]->getNumBins();
 432         vector<string> picrustLabels;
 433         vector<string> metadata = getMetaData(lookup, picrustLabels);
 434
 435         if (m->control_pressed) {  out.close(); return 0; }
 436
 437         //get row info
 438         /*"rows":[
 439                 {"id":"GG_OTU_1", "metadata":null},
 440                 {"id":"GG_OTU_2", "metadata":null},
 441                 {"id":"GG_OTU_3", "metadata":null},
 442                 {"id":"GG_OTU_4", "metadata":null},
 443                 {"id":"GG_OTU_5", "metadata":null}
 444                 ],*/
 445         out << spaces + "\"rows\":[\n";
 446         string rowFront = spaces + spaces + "{\"id\":\"";
 447         string rowBack = "\", \"metadata\":";
 448         for (int i = 0; i < numBins-1; i++) {
 449             if (m->control_pressed) {  out.close(); return 0; }
 450             if (!picrust) { out << rowFront << m->currentSharedBinLabels[i] << rowBack << metadata[i] << "},\n"; }
 451             else {  out << rowFront << picrustLabels[i] << rowBack << metadata[i] << "},\n"; }
 452         }
 453         if (!picrust) {  out << rowFront << m->currentSharedBinLabels[(numBins-1)] << rowBack << metadata[(numBins-1)] << "}\n" + spaces + "],\n"; }
 454         else {  out << rowFront << picrustLabels[(numBins-1)] << rowBack << metadata[(numBins-1)] << "}\n" + spaces + "],\n"; }
 455         //get column info
 456         /*"columns": [
 457                     {"id":"Sample1", "metadata":null},
 458                     {"id":"Sample2", "metadata":null},
 459                     {"id":"Sample3", "metadata":null},
 460                     {"id":"Sample4", "metadata":null},
 461                     {"id":"Sample5", "metadata":null},
 462                     {"id":"Sample6", "metadata":null}
 463                     ],*/
 464
 465         string colBack = "\", \"metadata\":";
 466         out << spaces + "\"columns\":[\n";
 467         for (int i = 0; i < lookup.size()-1; i++) {
 468             if (m->control_pressed) {  out.close(); return 0; }
 469             out << rowFront << lookup[i]->getGroup() << colBack << sampleMetadata[i] << "},\n";
 470         }
 471         out << rowFront << lookup[(lookup.size()-1)]->getGroup() << colBack << sampleMetadata[lookup.size()-1] << "}\n" + spaces + "],\n";
 472
 473         out << spaces + "\"matrix_type\": \"" << format << "\",\n" + spaces + "\"matrix_element_type\": \"int\",\n";
 474         out <<  spaces + "\"shape\": [" << numBins << "," << lookup.size() << "],\n";
 475         out << spaces + "\"data\":  [";
 476
 477         vector<string> dataRows;
 478         if (format == "sparse") {
 479             /*"data":[[0,2,1],
 480              [1,0,5],
 481              [1,1,1],
 482              [1,3,2],
 483              [1,4,3],
 484              [1,5,1],
 485              [2,2,1],
 486              [2,3,4],
 487              [2,4,2],
 488              [3,0,2],
 489              [3,1,1],
 490              [3,2,1],
 491              [3,5,1],
 492              [4,1,1],
 493              [4,2,1]
 494              ]*/
 495             string output = "";
 496             for (int i = 0; i < lookup[0]->getNumBins(); i++) {
 497
 498                 if (m->control_pressed) { out.close(); return 0; }
 499
 500                 for (int j = 0; j < lookup.size(); j++) {
 501                     string binInfo = "[" + toString(i) + "," + toString(j) + "," + toString(lookup[j]->getAbundance(i)) + "]";
 502                     //only print non zero values
 503                     if (lookup[j]->getAbundance(i) != 0) { dataRows.push_back(binInfo); }
 504                 }
 505             }
 506         }else {
 507
 508             /* "matrix_type": "dense",
 509              "matrix_element_type": "int",
 510              "shape": [5,6],
 511              "data":  [[0,0,1,0,0,0],
 512              [5,1,0,2,3,1],
 513              [0,0,1,4,2,0],
 514              [2,1,1,0,0,1],
 515              [0,1,1,0,0,0]]*/
 516
 517             for (int i = 0; i < lookup[0]->getNumBins(); i++) {
 518
 519                 if (m->control_pressed) { out.close(); return 0; }
 520
 521                 string binInfo = "[";
 522                 for (int j = 0; j < lookup.size()-1; j++) {
 523                     binInfo += toString(lookup[j]->getAbundance(i)) + ",";
 524                 }
 525                 binInfo += toString(lookup[lookup.size()-1]->getAbundance(i)) + "]";
 526                 dataRows.push_back(binInfo);
 527             }
 528         }
 529
 530         for (int i = 0; i < dataRows.size()-1; i++) {
 531             out << dataRows[i] << ",\n" + spaces  + spaces;
 532         }
 533         out << dataRows[dataRows.size()-1] << "]\n";
 534
 535         out << "}\n";
 536         out.close();
 537
 538         return 0;
 539     }
 540         catch(exception& e) {
 541                 m->errorOut(e, "MakeBiomCommand", "getBiom");
 542                 exit(1);
 543         }
 544 }
 545 //**********************************************************************************************************************
 546 vector<string> MakeBiomCommand::getMetaData(vector<SharedRAbundVector*>& lookup, vector<string>& picrustLabels){
 547         try {
 548         vector<string> metadata;
 549
 550         if (contaxonomyfile == "") { for (int i = 0; i < lookup[0]->getNumBins(); i++) {  metadata.push_back("null");  } }
 551         else {
 552
 553             //read constaxonomy file storing in a map, otulabel -> taxonomy
 554             //constaxonomy file will most likely contain more labels than the shared file, because sharedfile could have been subsampled.
 555             ifstream in;
 556             m->openInputFile(contaxonomyfile, in);
 557
 558             //grab headers
 559             m->getline(in); m->gobble(in);
 560
 561             string otuLabel, tax;
 562             int size;
 563             vector<string> otuLabels;
 564             vector<string> taxs;
 565             while (!in.eof()) {
 566
 567                 if (m->control_pressed) { in.close(); return metadata; }
 568
 569                 in >> otuLabel >> size >> tax; m->gobble(in);
 570
 571                 otuLabels.push_back(otuLabel);
 572                 taxs.push_back(tax);
 573             }
 574             in.close();
 575
 576             //should the labels be Otu001 or PhyloType001
 577             string firstBin = m->currentSharedBinLabels[0];
 578             string binTag = "Otu";
 579             if ((firstBin.find("Otu")) == string::npos) { binTag = "PhyloType";  }
 580
 581             //convert list file bin labels to shared file bin labels
 582             //parse tax strings
 583             //save in map
 584             map<string, string> labelTaxMap;
 585             string snumBins = toString(otuLabels.size());
 586             for (int i = 0; i < otuLabels.size(); i++) {
 587
 588                 if (m->control_pressed) { return metadata; }
 589
 590                 //if there is a bin label use it otherwise make one
 591                 if (m->isContainingOnlyDigits(otuLabels[i])) {
 592                     string binLabel = binTag;
 593                     string sbinNumber = otuLabels[i];
 594                     if (sbinNumber.length() < snumBins.length()) {
 595                         int diff = snumBins.length() - sbinNumber.length();
 596                         for (int h = 0; h < diff; h++) { binLabel += "0"; }
 597                     }
 598                     binLabel += sbinNumber;
 599                     binLabel = m->getSimpleLabel(binLabel);
 600                     labelTaxMap[binLabel] = taxs[i];
 601                 }else {  labelTaxMap[m->getSimpleLabel(otuLabels[i])] = taxs[i]; }
 602             }
 603
 604
 605             //{"taxonomy":["k__Bacteria", "p__Proteobacteria", "c__Gammaproteobacteria", "o__Enterobacteriales", "f__Enterobacteriaceae", "g__Escherichia", "s__"]}
 606
 607             //traverse the binLabels forming the metadata strings and saving them
 608             //make sure to sanity check
 609             map<string, string>::iterator it;
 610             for (int i = 0; i < lookup[0]->getNumBins(); i++) {
 611
 612                 if (m->control_pressed) { return metadata; }
 613
 614                 it = labelTaxMap.find(m->getSimpleLabel(m->currentSharedBinLabels[i]));
 615
 616                 if (it == labelTaxMap.end()) { m->mothurOut("[ERROR]: can't find taxonomy information for " + m->currentSharedBinLabels[i] + ".\n"); m->control_pressed = true; }
 617                 else {
 618                     if (picrust) {
 619                         string temp = it->second; m->removeConfidences(temp);
 620                         picrustLabels.push_back(temp);
 621                     }
 622                     vector<string> bootstrapValues;
 623                     string data = "{\"taxonomy\":[";
 624
 625                     vector<string> scores;
 626                     vector<string> taxonomies = parseTax(it->second, scores);
 627
 628                     for (int j = 0; j < taxonomies.size()-1; j ++) { data += "\"" + taxonomies[j] + "\", "; }
 629                     data += "\"" + taxonomies[taxonomies.size()-1] + "\"]";
 630
 631                     //add bootstrap values if available
 632                     if (scores[0] != "null") {
 633                         data += ", \"bootstrap\":[";
 634
 635                         for (int j = 0; j < scores.size()-1; j ++) { data += scores[j] + ", "; }
 636                         data += scores[scores.size()-1] + "]";
 637
 638                     }
 639                     data += "}";
 640
 641                     metadata.push_back(data);
 642                 }
 643             }
 644         }
 645
 646         return metadata;
 647
 648     }
 649         catch(exception& e) {
 650                 m->errorOut(e, "MakeBiomCommand", "getMetadata");
 651                 exit(1);
 652         }
 653
 654 }
 655 //**********************************************************************************************************************
 656 int MakeBiomCommand::getSampleMetaData(vector<SharedRAbundVector*>& lookup){
 657         try {
 658         sampleMetadata.clear();
 659         if (metadatafile == "") {  for (int i = 0; i < lookup.size(); i++) {  sampleMetadata.push_back("null");  } }
 660         else {
 661             ifstream in;
 662             m->openInputFile(metadatafile, in);
 663
 664             vector<string> groupNames, metadataLabels;
 665             map<string, vector<string> > lines;
 666
 667             string headerLine = m->getline(in); m->gobble(in);
 668             vector<string> pieces = m->splitWhiteSpace(headerLine);
 669
 670             //save names of columns you are reading
 671             for (int i = 1; i < pieces.size(); i++) {
 672                 metadataLabels.push_back(pieces[i]);
 673             }
 674             int count = metadataLabels.size();
 675
 676             vector<string> groups = m->getGroups();
 677
 678             //read rest of file
 679             while (!in.eof()) {
 680
 681                 if (m->control_pressed) { in.close(); return 0; }
 682
 683                 string group = "";
 684                 in >> group; m->gobble(in);
 685                 groupNames.push_back(group);
 686
 687                 string line = m->getline(in); m->gobble(in);
 688                 vector<string> thisPieces = m->splitWhiteSpaceWithQuotes(line);
 689
 690                 if (thisPieces.size() != count) { m->mothurOut("[ERROR]: expected " + toString(count) + " items of data for sample " + group + " read " + toString(thisPieces.size()) + ", quitting.\n"); }
 691                 else {  if (m->inUsersGroups(group, groups)) { lines[group] = thisPieces; } }
 692
 693                 m->gobble(in);
 694             }
 695             in.close();
 696
 697             map<string, vector<string> >::iterator it;
 698             for (int i = 0; i < lookup.size(); i++) {
 699
 700                 if (m->control_pressed) { return 0; }
 701
 702                 it = lines.find(lookup[i]->getGroup());
 703
 704                 if (it == lines.end()) { m->mothurOut("[ERROR]: can't find metadata information for " + lookup[i]->getGroup() + ", quitting.\n"); m->control_pressed = true; }
 705                 else {
 706                     vector<string> values = it->second;
 707
 708                     string data = "{";
 709                     for (int j = 0; j < metadataLabels.size()-1; j++) {
 710                         values[j] = m->removeQuotes(values[j]);
 711                         data += "\"" + metadataLabels[j] + "\":\"" + values[j] + "\", ";
 712                     }
 713                     values[metadataLabels.size()-1] = m->removeQuotes(values[metadataLabels.size()-1]);
 714                     data += "\"" + metadataLabels[metadataLabels.size()-1] + "\":\"" + values[metadataLabels.size()-1] + "\"}";
 715                     sampleMetadata.push_back(data);
 716                 }
 717             }
 718         }
 719
 720         return 0;
 721
 722     }
 723         catch(exception& e) {
 724                 m->errorOut(e, "MakeBiomCommand", "getSampleMetaData");
 725                 exit(1);
 726         }
 727
 728 }
 729
 730 /**************************************************************************************************/
 731 //returns {Bacteria, Bacteroidetes, ..} and scores is filled with {100, 98, ...} or {null, null, null}
 732 vector<string> MakeBiomCommand::parseTax(string tax, vector<string>& scores) {
 733         try {
 734
 735                 string taxon;
 736         vector<string> taxs;
 737
 738                 while (tax.find_first_of(';') != -1) {
 739
 740                         if (m->control_pressed) { return taxs; }
 741
 742                         //get taxon
 743                         taxon = tax.substr(0,tax.find_first_of(';'));
 744
 745                         int pos = taxon.find_last_of('(');
 746                         if (pos != -1) {
 747                                 //is it a number?
 748                                 int pos2 = taxon.find_last_of(')');
 749                                 if (pos2 != -1) {
 750                                         string confidenceScore = taxon.substr(pos+1, (pos2-(pos+1)));
 751                                         if (m->isNumeric1(confidenceScore)) {
 752                                                 taxon = taxon.substr(0, pos); //rip off confidence
 753                         scores.push_back(confidenceScore);
 754                                         }else{ scores.push_back("null"); }
 755                                 }
 756                         }else{ scores.push_back("null"); }
 757
 758             //strip "" if they are there
 759             pos = taxon.find("\"");
 760             if (pos != string::npos) {
 761                 string newTax = "";
 762                 for (int k = 0; k < taxon.length(); k++) {
 763                     if (taxon[k] != '\"') { newTax += taxon[k]; }
 764                 }
 765                 taxon = newTax;
 766             }
 767
 768             //look for bootstrap value
 769                         taxs.push_back(taxon);
 770             tax = tax.substr(tax.find_first_of(';')+1, tax.length());
 771                 }
 772
 773                 return taxs;
 774         }
 775         catch(exception& e) {
 776                 m->errorOut(e, "MakeBiomCommand", "parseTax");
 777                 exit(1);
 778         }
 779 }
 780
 781 //**********************************************************************************************************************
 782
 783
 784