heatmapsimcommand.cpp

   1 /*
   2  *  heatmapsimcommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by Sarah Westcott on 6/8/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10 #include "heatmapsimcommand.h"
  11 #include "sharedjabund.h"
  12 #include "sharedsorabund.h"
  13 #include "sharedjclass.h"
  14 #include "sharedsorclass.h"
  15 #include "sharedjest.h"
  16 #include "sharedsorest.h"
  17 #include "sharedthetayc.h"
  18 #include "sharedthetan.h"
  19 #include "sharedmorisitahorn.h"
  20 #include "sharedbraycurtis.h"
  21
  22 //**********************************************************************************************************************
  23 vector<string> HeatMapSimCommand::getValidParameters(){
  24         try {
  25                 string Array[] =  {"groups","label", "calc","phylip","column","name","outputdir","inputdir"};
  26                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  27                 return myArray;
  28         }
  29         catch(exception& e) {
  30                 m->errorOut(e, "HeatMapSimCommand", "getValidParameters");
  31                 exit(1);
  32         }
  33 }
  34 //**********************************************************************************************************************
  35 HeatMapSimCommand::HeatMapSimCommand(){
  36         try {
  37                 abort = true; calledHelp = true;
  38                 vector<string> tempOutNames;
  39                 outputTypes["svg"] = tempOutNames;
  40         }
  41         catch(exception& e) {
  42                 m->errorOut(e, "HeatMapSimCommand", "HeatMapSimCommand");
  43                 exit(1);
  44         }
  45 }
  46 //**********************************************************************************************************************
  47 vector<string> HeatMapSimCommand::getRequiredParameters(){
  48         try {
  49                 vector<string> myArray;
  50                 return myArray;
  51         }
  52         catch(exception& e) {
  53                 m->errorOut(e, "HeatMapSimCommand", "getRequiredParameters");
  54                 exit(1);
  55         }
  56 }
  57 //**********************************************************************************************************************
  58 vector<string> HeatMapSimCommand::getRequiredFiles(){
  59         try {
  60                 vector<string> myArray;
  61                 return myArray;
  62         }
  63         catch(exception& e) {
  64                 m->errorOut(e, "HeatMapSimCommand", "getRequiredFiles");
  65                 exit(1);
  66         }
  67 }
  68 //**********************************************************************************************************************
  69
  70 HeatMapSimCommand::HeatMapSimCommand(string option)  {
  71         try {
  72                 globaldata = GlobalData::getInstance();
  73                 abort = false; calledHelp = false;
  74                 allLines = 1;
  75                 labels.clear();
  76                 Groups.clear();
  77                 Estimators.clear();
  78
  79                 //allow user to run help
  80                 if(option == "help") { validCalculator = new ValidCalculators(); help(); abort = true; calledHelp = true; }
  81
  82                 else {
  83                         //valid paramters for this command
  84                         string AlignArray[] =  {"groups","label", "calc","phylip","column","name","outputdir","inputdir"};
  85                         vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
  86
  87                         OptionParser parser(option);
  88                         map<string,string> parameters = parser.getParameters();
  89
  90                         ValidParameters validParameter;
  91                         map<string,string>::iterator it;
  92
  93                         //check to make sure all parameters are valid for command
  94                         for (it = parameters.begin(); it != parameters.end(); it++) {
  95                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  96                         }
  97
  98                         //initialize outputTypes
  99                         vector<string> tempOutNames;
 100                         outputTypes["svg"] = tempOutNames;
 101
 102                         format = "";
 103                         //if the user changes the output directory command factory will send this info to us in the output parameter
 104                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 105
 106                         //if the user changes the input directory command factory will send this info to us in the output parameter
 107                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 108                         if (inputDir == "not found"){   inputDir = "";          }
 109                         else {
 110                                 string path;
 111                                 it = parameters.find("phylip");
 112                                 //user has given a template file
 113                                 if(it != parameters.end()){
 114                                         path = m->hasPath(it->second);
 115                                         //if the user has not given a path then, add inputdir. else leave path alone.
 116                                         if (path == "") {       parameters["phylip"] = inputDir + it->second;           }
 117                                 }
 118
 119                                 it = parameters.find("column");
 120                                 //user has given a template file
 121                                 if(it != parameters.end()){
 122                                         path = m->hasPath(it->second);
 123                                         //if the user has not given a path then, add inputdir. else leave path alone.
 124                                         if (path == "") {       parameters["column"] = inputDir + it->second;           }
 125                                 }
 126
 127                                 it = parameters.find("name");
 128                                 //user has given a template file
 129                                 if(it != parameters.end()){
 130                                         path = m->hasPath(it->second);
 131                                         //if the user has not given a path then, add inputdir. else leave path alone.
 132                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 133                                 }
 134                         }
 135
 136                         //required parameters
 137                         phylipfile = validParameter.validFile(parameters, "phylip", true);
 138                         if (phylipfile == "not open") { abort = true; }
 139                         else if (phylipfile == "not found") { phylipfile = ""; }
 140                         else {  format = "phylip";      if (outputDir == "") { outputDir += m->hasPath(phylipfile); }  }
 141
 142                         columnfile = validParameter.validFile(parameters, "column", true);
 143                         if (columnfile == "not open") { abort = true; }
 144                         else if (columnfile == "not found") { columnfile = ""; }
 145                         else {  format = "column";      if (outputDir == "") { outputDir += m->hasPath(columnfile); } }
 146
 147                         namefile = validParameter.validFile(parameters, "name", true);
 148                         if (namefile == "not open") { abort = true; }
 149                         else if (namefile == "not found") { namefile = ""; }
 150
 151
 152                         //error checking on files
 153                         if ((globaldata->getSharedFile() == "") && ((phylipfile == "") && (columnfile == "")))  { m->mothurOut("You must run the read.otu command or provide a distance file before running the heatmap.sim command."); m->mothurOutEndLine(); abort = true; }
 154                         else if ((phylipfile != "") && (columnfile != "")) { m->mothurOut("When running the heatmap.sim command with a distance file you may not use both the column and the phylip parameters."); m->mothurOutEndLine(); abort = true; }
 155
 156                         if (columnfile != "") {
 157                                 if (namefile == "") {  m->mothurOut("You need to provide a namefile if you are going to use the column format."); m->mothurOutEndLine(); abort = true; }
 158                         }
 159
 160                         if (format == "") { format = "shared"; }
 161
 162                         //check for optional parameter and set defaults
 163                         // ...at some point should added some additional type checking...
 164                         if (format == "shared") {
 165                                 if (outputDir == "") { outputDir += m->hasPath(globaldata->getSharedFile()); }
 166
 167                                 label = validParameter.validFile(parameters, "label", false);
 168                                 if (label == "not found") { label = ""; }
 169                                 else {
 170                                         if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 171                                         else { allLines = 1;  }
 172                                 }
 173
 174                                 //if the user has not specified any labels use the ones from read.otu
 175                                 if (label == "") {
 176                                         allLines = globaldata->allLines;
 177                                         labels = globaldata->labels;
 178                                 }
 179
 180                                 calc = validParameter.validFile(parameters, "calc", false);
 181                                 if (calc == "not found") { calc = "jest-thetayc";  }
 182                                 else {
 183                                         if (calc == "default")  {  calc = "jest-thetayc";  }
 184                                 }
 185                                 m->splitAtDash(calc, Estimators);
 186
 187                                 groups = validParameter.validFile(parameters, "groups", false);
 188                                 if (groups == "not found") { groups = ""; }
 189                                 else {
 190                                         m->splitAtDash(groups, Groups);
 191                                         globaldata->Groups = Groups;
 192                                 }
 193                         }
 194
 195                         if (abort == false) {
 196                                 validCalculator = new ValidCalculators();
 197
 198                                 int i;
 199                                 for (i=0; i<Estimators.size(); i++) {
 200                                         if (validCalculator->isValidCalculator("heat", Estimators[i]) == true) {
 201                                                 if (Estimators[i] == "jabund") {
 202                                                         heatCalculators.push_back(new JAbund());
 203                                                 }else if (Estimators[i] == "sorabund") {
 204                                                         heatCalculators.push_back(new SorAbund());
 205                                                 }else if (Estimators[i] == "jclass") {
 206                                                         heatCalculators.push_back(new Jclass());
 207                                                 }else if (Estimators[i] == "sorclass") {
 208                                                         heatCalculators.push_back(new SorClass());
 209                                                 }else if (Estimators[i] == "jest") {
 210                                                         heatCalculators.push_back(new Jest());
 211                                                 }else if (Estimators[i] == "sorest") {
 212                                                         heatCalculators.push_back(new SorEst());
 213                                                 }else if (Estimators[i] == "thetayc") {
 214                                                         heatCalculators.push_back(new ThetaYC());
 215                                                 }else if (Estimators[i] == "thetan") {
 216                                                         heatCalculators.push_back(new ThetaN());
 217                                                 }else if (Estimators[i] == "morisitahorn") {
 218                                                         heatCalculators.push_back(new MorHorn());
 219                                                 }else if (Estimators[i] == "braycurtis") {
 220                                                         heatCalculators.push_back(new BrayCurtis());
 221                                                 }
 222                                         }
 223                                 }
 224
 225                         }
 226                 }
 227
 228
 229
 230         }
 231         catch(exception& e) {
 232                 m->errorOut(e, "HeatMapSimCommand", "HeatMapSimCommand");
 233                 exit(1);
 234         }
 235 }
 236
 237 //**********************************************************************************************************************
 238
 239 void HeatMapSimCommand::help(){
 240         try {
 241                 m->mothurOut("The heatmap.sim command can only be executed after a successful read.otu command, or by providing a distance file.\n");
 242                 m->mothurOut("The heatmap.sim command parameters are phylip, column, name, groups, calc and label.  No parameters are required.\n");
 243                 m->mothurOut("There are two ways to use the heatmap.sim command. The first is with the read.otu command. \n");
 244                 m->mothurOut("With the read.otu command you may use the groups, label and calc parameters. \n");
 245                 m->mothurOut("The groups parameter allows you to specify which of the groups in your groupfile you would like included in your heatmap.\n");
 246                 m->mothurOut("The group names are separated by dashes. The label parameter allows you to select what distance levels you would like a heatmap created for, and is also separated by dashes.\n");
 247                 m->mothurOut("The heatmap.sim command should be in the following format: heatmap.sim(groups=yourGroups, calc=yourCalc, label=yourLabels).\n");
 248                 m->mothurOut("Example heatmap.sim(groups=A-B-C, calc=jabund).\n");
 249                 m->mothurOut("The default value for groups is all the groups in your groupfile, and all labels in your inputfile will be used.\n");
 250                 validCalculator->printCalc("heat", cout);
 251                 m->mothurOut("The default value for calc is jclass-thetayc.\n");
 252                 m->mothurOut("The heatmap.sim command outputs a .svg file for each calculator you choose at each label you specify.\n");
 253                 m->mothurOut("The second way to use the heatmap.sim command is with a distance file representing the distance bewteen your groups. \n");
 254                 m->mothurOut("Using the command this way, the phylip or column parameter are required, and only one may be used.  If you use a column file the name filename is required. \n");
 255                 m->mothurOut("The heatmap.sim command should be in the following format: heatmap.sim(phylip=yourDistanceFile).\n");
 256                 m->mothurOut("Example heatmap.sim(phylip=amazonGroups.dist).\n");
 257                 m->mothurOut("Note: No spaces between parameter labels (i.e. groups), '=' and parameters (i.e.yourGroups).\n\n");
 258
 259         }
 260         catch(exception& e) {
 261                 m->errorOut(e, "HeatMapSimCommand", "help");
 262                 exit(1);
 263         }
 264 }
 265
 266 //**********************************************************************************************************************
 267
 268 HeatMapSimCommand::~HeatMapSimCommand(){}
 269
 270 //**********************************************************************************************************************
 271
 272 int HeatMapSimCommand::execute(){
 273         try {
 274
 275                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 276
 277                 heatmap = new HeatMapSim(outputDir);
 278
 279                 if (format == "shared") {
 280                         runCommandShared();
 281                 }else if (format == "phylip") {
 282                         globaldata->inputFileName = phylipfile;
 283                         runCommandDist();
 284                 }else if (format == "column") {
 285                         globaldata->inputFileName = columnfile;
 286                         runCommandDist();
 287                 }
 288
 289                 delete heatmap;
 290                 delete validCalculator;
 291
 292                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str());  } outputTypes.clear(); return 0; }
 293
 294                 m->mothurOutEndLine();
 295                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 296                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 297                 m->mothurOutEndLine();
 298
 299                 return 0;
 300         }
 301         catch(exception& e) {
 302                 m->errorOut(e, "HeatMapSimCommand", "execute");
 303                 exit(1);
 304         }
 305 }
 306
 307 //**********************************************************************************************************************
 308 int HeatMapSimCommand::runCommandShared() {
 309         try {
 310                 //if the users entered no valid calculators don't execute command
 311                 if (heatCalculators.size() == 0) { m->mothurOut("No valid calculators."); m->mothurOutEndLine(); return 0; }
 312
 313                 //you have groups
 314                 read = new ReadOTUFile(globaldata->inputFileName);
 315                 read->read(&*globaldata);
 316
 317                 input = globaldata->ginput;
 318                 lookup = input->getSharedRAbundVectors();
 319                 string lastLabel = lookup[0]->getLabel();
 320
 321                 if (lookup.size() < 2) { m->mothurOut("You have not provided enough valid groups.  I cannot run the command."); m->mothurOutEndLine(); return 0;}
 322
 323                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 324                 set<string> processedLabels;
 325                 set<string> userLabels = labels;
 326
 327                 if (m->control_pressed) { delete read; delete input; globaldata->ginput = NULL; for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }  globaldata->Groups.clear(); return 0; }
 328
 329                 //as long as you are not at the end of the file or done wih the lines you want
 330                 while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 331
 332                         if (m->control_pressed) { delete read; delete input; globaldata->ginput = NULL; for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  } globaldata->Groups.clear(); return 0; }
 333
 334                         if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){
 335
 336                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 337                                 vector<string> outfilenames = heatmap->getPic(lookup, heatCalculators);
 338                                 for(int i = 0; i < outfilenames.size(); i++) { outputNames.push_back(outfilenames[i]);  outputTypes["svg"].push_back(outfilenames[i]); }
 339
 340                                 processedLabels.insert(lookup[0]->getLabel());
 341                                 userLabels.erase(lookup[0]->getLabel());
 342                         }
 343
 344                         if ((m->anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 345                                 string saveLabel = lookup[0]->getLabel();
 346
 347                                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 348                                 lookup = input->getSharedRAbundVectors(lastLabel);
 349
 350                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 351                                 vector<string> outfilenames = heatmap->getPic(lookup, heatCalculators);
 352                                 for(int i = 0; i < outfilenames.size(); i++) { outputNames.push_back(outfilenames[i]); outputTypes["svg"].push_back(outfilenames[i]);  }
 353
 354                                 processedLabels.insert(lookup[0]->getLabel());
 355                                 userLabels.erase(lookup[0]->getLabel());
 356
 357                                 //restore real lastlabel to save below
 358                                 lookup[0]->setLabel(saveLabel);
 359                         }
 360
 361                         //prevent memory leak
 362
 363                         lastLabel = lookup[0]->getLabel();
 364
 365                         //get next line to process
 366                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 367                         lookup = input->getSharedRAbundVectors();
 368
 369                 }
 370
 371
 372                 if (m->control_pressed) { delete read; delete input; globaldata->ginput = NULL; globaldata->Groups.clear();  return 0; }
 373
 374                 //output error messages about any remaining user labels
 375                 set<string>::iterator it;
 376                 bool needToRun = false;
 377                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 378                         m->mothurOut("Your file does not include the label " + *it);
 379                         if (processedLabels.count(lastLabel) != 1) {
 380                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 381                                 needToRun = true;
 382                         }else {
 383                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 384                         }
 385                 }
 386
 387                 if (m->control_pressed) { delete read; delete input; globaldata->ginput = NULL;  globaldata->Groups.clear(); return 0; }
 388
 389                 //run last label if you need to
 390                 if (needToRun == true)  {
 391                         for (int i = 0; i < lookup.size(); i++) {  if (lookup[i] != NULL) { delete lookup[i]; } }
 392                         lookup = input->getSharedRAbundVectors(lastLabel);
 393
 394                         m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 395                         vector<string> outfilenames = heatmap->getPic(lookup, heatCalculators);
 396                         for(int i = 0; i < outfilenames.size(); i++) { outputNames.push_back(outfilenames[i]); outputTypes["svg"].push_back(outfilenames[i]);  }
 397
 398                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 399                 }
 400
 401                 if (m->control_pressed) { delete read; delete input; globaldata->ginput = NULL;  globaldata->Groups.clear(); return 0; }
 402
 403                 //reset groups parameter
 404                 globaldata->Groups.clear();
 405
 406                 delete input;  globaldata->ginput = NULL;
 407                 delete read;
 408
 409                 return 0;
 410         }
 411         catch(exception& e) {
 412                 m->errorOut(e, "HeatMapSimCommand", "runCommandShared");
 413                 exit(1);
 414         }
 415 }
 416 //**********************************************************************************************************************
 417 int HeatMapSimCommand::runCommandDist() {
 418         try {
 419
 420                 vector< vector<double> > matrix;
 421                 vector<string> names;
 422                 ifstream in;
 423
 424                 //read distance file and create distance vector and names vector
 425                 if (format == "phylip") {
 426                         //read phylip file
 427                         m->openInputFile(phylipfile, in);
 428
 429                         string name;
 430                         int numSeqs;
 431                         in >> numSeqs >> name;
 432
 433                         //save name
 434                         names.push_back(name);
 435
 436                         //resize the matrix and fill with zeros
 437                         matrix.resize(numSeqs);
 438                         for(int i = 0; i < numSeqs; i++) {
 439                                 matrix[i].resize(numSeqs, 0.0);
 440                         }
 441
 442                         //determine if matrix is square or lower triangle
 443                         //if it is square read the distances for the first sequence
 444                         char d;
 445                         bool square;
 446                         while((d=in.get()) != EOF){
 447
 448                                 //is d a number meaning its square
 449                                 if(isalnum(d)){
 450                                         square = true;
 451                                         in.putback(d);
 452
 453                                         for(int i=0;i<numSeqs;i++){
 454                                                 in >> matrix[0][i];
 455                                         }
 456                                         break;
 457                                 }
 458
 459                                 //is d a line return meaning its lower triangle
 460                                 if(d == '\n'){
 461                                         square = false;
 462                                         break;
 463                                 }
 464                         }
 465
 466                         //read rest of matrix
 467                         if (square == true) {
 468                                 for(int i=1;i<numSeqs;i++){
 469                                         in >> name;
 470                                         names.push_back(name);
 471
 472                                         if (m->control_pressed) { return 0; }
 473
 474                                         for(int j=0;j<numSeqs;j++) { in >> matrix[i][j];  }
 475                                         m->gobble(in);
 476                                 }
 477                         }else {
 478                                 double dist;
 479                                 for(int i=1;i<numSeqs;i++){
 480                                         in >> name;
 481                                         names.push_back(name);
 482
 483                                         if (m->control_pressed) { return 0; }
 484
 485                                         for(int j=0;j<i;j++){
 486                                                 in >> dist;
 487                                                 matrix[i][j] = dist;  matrix[j][i] = dist;
 488                                         }
 489                                         m->gobble(in);
 490                                 }
 491                         }
 492                         in.close();
 493                 }else {
 494                         //read names file
 495                         NameAssignment* nameMap = new NameAssignment(namefile);
 496                         nameMap->readMap();
 497
 498                         //put names in order in vector
 499                         for (int i = 0; i < nameMap->size(); i++) {
 500                                 names.push_back(nameMap->get(i));
 501                         }
 502
 503                         //resize matrix
 504                         matrix.resize(nameMap->size());
 505                         for (int i = 0; i < nameMap->size(); i++) {
 506                                 matrix[i].resize(nameMap->size(), 0.0);
 507                         }
 508
 509                         //read column file
 510                         string first, second;
 511                         double dist;
 512                         m->openInputFile(columnfile, in);
 513
 514                         while (!in.eof()) {
 515                                 in >> first >> second >> dist; m->gobble(in);
 516
 517                                 if (m->control_pressed) { return 0; }
 518
 519                                 map<string, int>::iterator itA = nameMap->find(first);
 520                                 map<string, int>::iterator itB = nameMap->find(second);
 521
 522                                 if(itA == nameMap->end()){  cerr << "AAError: Sequence '" << first << "' was not found in the names file, please correct\n"; exit(1);  }
 523                                 if(itB == nameMap->end()){  cerr << "ABError: Sequence '" << second << "' was not found in the names file, please correct\n"; exit(1);  }
 524
 525                                 //save distance
 526                                 matrix[itA->second][itB->second] = dist;
 527                                 matrix[itB->second][itA->second] = dist;
 528                         }
 529                         in.close();
 530
 531                         delete nameMap;
 532                 }
 533
 534
 535                 string outputFileName = heatmap->getPic(matrix, names);
 536                 outputNames.push_back(outputFileName); //vector<vector<double>>, vector<string>
 537                 outputTypes["svg"].push_back(outputFileName);
 538
 539                 return 0;
 540         }
 541         catch(exception& e) {
 542                 m->errorOut(e, "HeatMapSimCommand", "runCommandDist");
 543                 exit(1);
 544         }
 545 }
 546 //**********************************************************************************************************************
 547
 548
 549
 550
 551
 552