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