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", "", "", "PhylipColumnShared", "PhylipColumnShared", "none",false,false); parameters.push_back(pshared);
  27                 CommandParameter pphylip("phylip", "InputTypes", "", "", "PhylipColumnShared", "PhylipColumnShared", "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", "", "", "PhylipColumnShared", "PhylipColumnShared", "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                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  96
  97                 else {
  98                         vector<string> myArray = setParameters();
  99
 100                         OptionParser parser(option);
 101                         map<string,string> parameters = parser.getParameters();
 102
 103                         ValidParameters validParameter;
 104                         map<string,string>::iterator it;
 105
 106                         //check to make sure all parameters are valid for command
 107                         for (it = parameters.begin(); it != parameters.end(); it++) {
 108                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 109                         }
 110
 111                         //initialize outputTypes
 112                         vector<string> tempOutNames;
 113                         outputTypes["svg"] = tempOutNames;
 114
 115                         format = "";
 116                         //if the user changes the output directory command factory will send this info to us in the output parameter
 117                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 118
 119                         //if the user changes the input directory command factory will send this info to us in the output parameter
 120                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 121                         if (inputDir == "not found"){   inputDir = "";          }
 122                         else {
 123                                 string path;
 124                                 it = parameters.find("phylip");
 125                                 //user has given a template file
 126                                 if(it != parameters.end()){
 127                                         path = m->hasPath(it->second);
 128                                         //if the user has not given a path then, add inputdir. else leave path alone.
 129                                         if (path == "") {       parameters["phylip"] = inputDir + it->second;           }
 130                                 }
 131
 132                                 it = parameters.find("column");
 133                                 //user has given a template file
 134                                 if(it != parameters.end()){
 135                                         path = m->hasPath(it->second);
 136                                         //if the user has not given a path then, add inputdir. else leave path alone.
 137                                         if (path == "") {       parameters["column"] = inputDir + it->second;           }
 138                                 }
 139
 140                                 it = parameters.find("name");
 141                                 //user has given a template file
 142                                 if(it != parameters.end()){
 143                                         path = m->hasPath(it->second);
 144                                         //if the user has not given a path then, add inputdir. else leave path alone.
 145                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 146                                 }
 147
 148                                 it = parameters.find("shared");
 149                                 //user has given a template file
 150                                 if(it != parameters.end()){
 151                                         path = m->hasPath(it->second);
 152                                         //if the user has not given a path then, add inputdir. else leave path alone.
 153                                         if (path == "") {       parameters["shared"] = inputDir + it->second;           }
 154                                 }
 155                         }
 156
 157                         //required parameters
 158                         phylipfile = validParameter.validFile(parameters, "phylip", true);
 159                         if (phylipfile == "not open") { abort = true; }
 160                         else if (phylipfile == "not found") { phylipfile = ""; }
 161                         else {  format = "phylip";      inputfile = phylipfile; if (outputDir == "") { outputDir += m->hasPath(phylipfile); }  }
 162
 163                         columnfile = validParameter.validFile(parameters, "column", true);
 164                         if (columnfile == "not open") { abort = true; }
 165                         else if (columnfile == "not found") { columnfile = ""; }
 166                         else {  format = "column";      inputfile = columnfile; if (outputDir == "") { outputDir += m->hasPath(columnfile); } }
 167
 168                         sharedfile = validParameter.validFile(parameters, "shared", true);
 169                         if (sharedfile == "not open") { abort = true; }
 170                         else if (sharedfile == "not found") { sharedfile = ""; }
 171                         else {  format = "shared";      inputfile = sharedfile;  if (outputDir == "") { outputDir += m->hasPath(sharedfile); } }
 172
 173                         namefile = validParameter.validFile(parameters, "name", true);
 174                         if (namefile == "not open") { abort = true; }
 175                         else if (namefile == "not found") { namefile = ""; }
 176
 177
 178                         //error checking on files
 179                         if ((sharedfile == "") && ((phylipfile == "") && (columnfile == "")))   {
 180                                 sharedfile = m->getSharedFile();
 181                                 if (sharedfile != "") { format = "shared"; inputfile = sharedfile; m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); }
 182                                 else {
 183                                         //is there are current file available for either of these?
 184                                         //give priority to column, then phylip
 185                                         columnfile = m->getColumnFile();
 186                                         if (columnfile != "") {  format = "column"; inputfile = columnfile; m->mothurOut("Using " + columnfile + " as input file for the column parameter."); m->mothurOutEndLine(); }
 187                                         else {
 188                                                 phylipfile = m->getPhylipFile();
 189                                                 if (phylipfile != "") { format = "phylip";  inputfile = phylipfile; m->mothurOut("Using " + phylipfile + " as input file for the phylip parameter."); m->mothurOutEndLine(); }
 190                                                 else {
 191                                                         m->mothurOut("No valid current files. You must provide a shared or phylip or column file."); m->mothurOutEndLine();
 192                                                         abort = true;
 193                                                 }
 194                                         }
 195                                 }
 196                         }
 197                         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; }
 198
 199                         if (columnfile != "") {
 200                                 if (namefile == "") {
 201                                         namefile = m->getNameFile();
 202                                         if (namefile != "") {  m->mothurOut("Using " + namefile + " as input file for the name parameter."); m->mothurOutEndLine(); }
 203                                         else {
 204                                                 m->mothurOut("You need to provide a namefile if you are going to use the column format."); m->mothurOutEndLine();
 205                                                 abort = true;
 206                                         }
 207                                 }
 208                         }
 209
 210                         //check for optional parameter and set defaults
 211                         // ...at some point should added some additional type checking...
 212
 213                         label = validParameter.validFile(parameters, "label", false);
 214                         if (label == "not found") { label = ""; }
 215                         else {
 216                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 217                                 else { allLines = 1;  }
 218                         }
 219
 220                         calc = validParameter.validFile(parameters, "calc", false);
 221                         if (calc == "not found") { calc = "jest-thetayc";  }
 222                         else {
 223                                 if (calc == "default")  {  calc = "jest-thetayc";  }
 224                         }
 225                         m->splitAtDash(calc, Estimators);
 226
 227                         groups = validParameter.validFile(parameters, "groups", false);
 228                         if (groups == "not found") { groups = ""; }
 229                         else {
 230                                 m->splitAtDash(groups, Groups);
 231                                 m->Groups = Groups;
 232                         }
 233
 234
 235                         if (abort == false) {
 236                                 ValidCalculators validCalculator;
 237
 238                                 int i;
 239                                 for (i=0; i<Estimators.size(); i++) {
 240                                         if (validCalculator.isValidCalculator("heat", Estimators[i]) == true) {
 241                                                 if (Estimators[i] == "jabund") {
 242                                                         heatCalculators.push_back(new JAbund());
 243                                                 }else if (Estimators[i] == "sorabund") {
 244                                                         heatCalculators.push_back(new SorAbund());
 245                                                 }else if (Estimators[i] == "jclass") {
 246                                                         heatCalculators.push_back(new Jclass());
 247                                                 }else if (Estimators[i] == "sorclass") {
 248                                                         heatCalculators.push_back(new SorClass());
 249                                                 }else if (Estimators[i] == "jest") {
 250                                                         heatCalculators.push_back(new Jest());
 251                                                 }else if (Estimators[i] == "sorest") {
 252                                                         heatCalculators.push_back(new SorEst());
 253                                                 }else if (Estimators[i] == "thetayc") {
 254                                                         heatCalculators.push_back(new ThetaYC());
 255                                                 }else if (Estimators[i] == "thetan") {
 256                                                         heatCalculators.push_back(new ThetaN());
 257                                                 }else if (Estimators[i] == "morisitahorn") {
 258                                                         heatCalculators.push_back(new MorHorn());
 259                                                 }else if (Estimators[i] == "braycurtis") {
 260                                                         heatCalculators.push_back(new BrayCurtis());
 261                                                 }
 262                                         }
 263                                 }
 264
 265                         }
 266                 }
 267
 268
 269
 270         }
 271         catch(exception& e) {
 272                 m->errorOut(e, "HeatMapSimCommand", "HeatMapSimCommand");
 273                 exit(1);
 274         }
 275 }
 276 //**********************************************************************************************************************
 277
 278 int HeatMapSimCommand::execute(){
 279         try {
 280
 281                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 282
 283                 heatmap = new HeatMapSim(outputDir, inputfile);
 284
 285                 if (format == "shared") {
 286                         runCommandShared();
 287                 }else{  runCommandDist();       }
 288
 289                 delete heatmap;
 290
 291                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str());  } outputTypes.clear(); return 0; }
 292
 293                 m->mothurOutEndLine();
 294                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 295                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 296                 m->mothurOutEndLine();
 297
 298                 return 0;
 299         }
 300         catch(exception& e) {
 301                 m->errorOut(e, "HeatMapSimCommand", "execute");
 302                 exit(1);
 303         }
 304 }
 305
 306 //**********************************************************************************************************************
 307 int HeatMapSimCommand::runCommandShared() {
 308         try {
 309                 //if the users entered no valid calculators don't execute command
 310                 if (heatCalculators.size() == 0) { m->mothurOut("No valid calculators."); m->mothurOutEndLine(); return 0; }
 311
 312                 input = new InputData(sharedfile, "sharedfile");
 313                 lookup = input->getSharedRAbundVectors();
 314                 string lastLabel = lookup[0]->getLabel();
 315
 316                 if (lookup.size() < 2) { m->mothurOut("You have not provided enough valid groups.  I cannot run the command."); m->mothurOutEndLine(); return 0;}
 317
 318                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 319                 set<string> processedLabels;
 320                 set<string> userLabels = labels;
 321
 322                 if (m->control_pressed) {  delete input;  for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }  m->Groups.clear(); return 0; }
 323
 324                 //as long as you are not at the end of the file or done wih the lines you want
 325                 while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 326
 327                         if (m->control_pressed) { delete input;  for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  } m->Groups.clear(); return 0; }
 328
 329                         if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){
 330
 331                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 332                                 vector<string> outfilenames = heatmap->getPic(lookup, heatCalculators);
 333                                 for(int i = 0; i < outfilenames.size(); i++) { outputNames.push_back(outfilenames[i]);  outputTypes["svg"].push_back(outfilenames[i]); }
 334
 335                                 processedLabels.insert(lookup[0]->getLabel());
 336                                 userLabels.erase(lookup[0]->getLabel());
 337                         }
 338
 339                         if ((m->anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 340                                 string saveLabel = lookup[0]->getLabel();
 341
 342                                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 343                                 lookup = input->getSharedRAbundVectors(lastLabel);
 344
 345                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 346                                 vector<string> outfilenames = heatmap->getPic(lookup, heatCalculators);
 347                                 for(int i = 0; i < outfilenames.size(); i++) { outputNames.push_back(outfilenames[i]); outputTypes["svg"].push_back(outfilenames[i]);  }
 348
 349                                 processedLabels.insert(lookup[0]->getLabel());
 350                                 userLabels.erase(lookup[0]->getLabel());
 351
 352                                 //restore real lastlabel to save below
 353                                 lookup[0]->setLabel(saveLabel);
 354                         }
 355
 356                         //prevent memory leak
 357
 358                         lastLabel = lookup[0]->getLabel();
 359
 360                         //get next line to process
 361                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 362                         lookup = input->getSharedRAbundVectors();
 363
 364                 }
 365
 366
 367                 if (m->control_pressed) {  delete input;  m->Groups.clear();  return 0; }
 368
 369                 //output error messages about any remaining user labels
 370                 set<string>::iterator it;
 371                 bool needToRun = false;
 372                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 373                         m->mothurOut("Your file does not include the label " + *it);
 374                         if (processedLabels.count(lastLabel) != 1) {
 375                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 376                                 needToRun = true;
 377                         }else {
 378                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 379                         }
 380                 }
 381
 382                 if (m->control_pressed) {  delete input;  m->Groups.clear(); return 0; }
 383
 384                 //run last label if you need to
 385                 if (needToRun == true)  {
 386                         for (int i = 0; i < lookup.size(); i++) {  if (lookup[i] != NULL) { delete lookup[i]; } }
 387                         lookup = input->getSharedRAbundVectors(lastLabel);
 388
 389                         m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 390                         vector<string> outfilenames = heatmap->getPic(lookup, heatCalculators);
 391                         for(int i = 0; i < outfilenames.size(); i++) { outputNames.push_back(outfilenames[i]); outputTypes["svg"].push_back(outfilenames[i]);  }
 392
 393                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 394                 }
 395
 396                 if (m->control_pressed) {  delete input;  m->Groups.clear(); return 0; }
 397
 398                 //reset groups parameter
 399                 m->Groups.clear();
 400
 401                 delete input;
 402
 403                 return 0;
 404         }
 405         catch(exception& e) {
 406                 m->errorOut(e, "HeatMapSimCommand", "runCommandShared");
 407                 exit(1);
 408         }
 409 }
 410 //**********************************************************************************************************************
 411 int HeatMapSimCommand::runCommandDist() {
 412         try {
 413
 414                 vector< vector<double> > matrix;
 415                 vector<string> names;
 416                 ifstream in;
 417
 418                 //read distance file and create distance vector and names vector
 419                 if (format == "phylip") {
 420                         //read phylip file
 421                         m->openInputFile(phylipfile, in);
 422
 423                         string name;
 424                         int numSeqs;
 425                         in >> numSeqs >> name;
 426
 427                         //save name
 428                         names.push_back(name);
 429
 430                         //resize the matrix and fill with zeros
 431                         matrix.resize(numSeqs);
 432                         for(int i = 0; i < numSeqs; i++) {
 433                                 matrix[i].resize(numSeqs, 0.0);
 434                         }
 435
 436                         //determine if matrix is square or lower triangle
 437                         //if it is square read the distances for the first sequence
 438                         char d;
 439                         bool square;
 440                         while((d=in.get()) != EOF){
 441
 442                                 //is d a number meaning its square
 443                                 if(isalnum(d)){
 444                                         square = true;
 445                                         in.putback(d);
 446
 447                                         for(int i=0;i<numSeqs;i++){
 448                                                 in >> matrix[0][i];
 449                                         }
 450                                         break;
 451                                 }
 452
 453                                 //is d a line return meaning its lower triangle
 454                                 if(d == '\n'){
 455                                         square = false;
 456                                         break;
 457                                 }
 458                         }
 459
 460                         //read rest of matrix
 461                         if (square == true) {
 462                                 for(int i=1;i<numSeqs;i++){
 463                                         in >> name;
 464                                         names.push_back(name);
 465
 466                                         if (m->control_pressed) { return 0; }
 467
 468                                         for(int j=0;j<numSeqs;j++) { in >> matrix[i][j];  }
 469                                         m->gobble(in);
 470                                 }
 471                         }else {
 472                                 double dist;
 473                                 for(int i=1;i<numSeqs;i++){
 474                                         in >> name;
 475                                         names.push_back(name);
 476
 477                                         if (m->control_pressed) { return 0; }
 478
 479                                         for(int j=0;j<i;j++){
 480                                                 in >> dist;
 481                                                 matrix[i][j] = dist;  matrix[j][i] = dist;
 482                                         }
 483                                         m->gobble(in);
 484                                 }
 485                         }
 486                         in.close();
 487                 }else {
 488                         //read names file
 489                         NameAssignment* nameMap = new NameAssignment(namefile);
 490                         nameMap->readMap();
 491
 492                         //put names in order in vector
 493                         for (int i = 0; i < nameMap->size(); i++) {
 494                                 names.push_back(nameMap->get(i));
 495                         }
 496
 497                         //resize matrix
 498                         matrix.resize(nameMap->size());
 499                         for (int i = 0; i < nameMap->size(); i++) {
 500                                 matrix[i].resize(nameMap->size(), 0.0);
 501                         }
 502
 503                         //read column file
 504                         string first, second;
 505                         double dist;
 506                         m->openInputFile(columnfile, in);
 507
 508                         while (!in.eof()) {
 509                                 in >> first >> second >> dist; m->gobble(in);
 510
 511                                 if (m->control_pressed) { return 0; }
 512
 513                                 map<string, int>::iterator itA = nameMap->find(first);
 514                                 map<string, int>::iterator itB = nameMap->find(second);
 515
 516                                 if(itA == nameMap->end()){  m->mothurOut("AAError: Sequence '" + first + "' was not found in the names file, please correct\n"); exit(1);  }
 517                                 if(itB == nameMap->end()){  m->mothurOut("ABError: Sequence '" + second + "' was not found in the names file, please correct\n"); exit(1);  }
 518
 519                                 //save distance
 520                                 matrix[itA->second][itB->second] = dist;
 521                                 matrix[itB->second][itA->second] = dist;
 522                         }
 523                         in.close();
 524
 525                         delete nameMap;
 526                 }
 527
 528
 529                 string outputFileName = heatmap->getPic(matrix, names);
 530                 outputNames.push_back(outputFileName); //vector<vector<double>>, vector<string>
 531                 outputTypes["svg"].push_back(outputFileName);
 532
 533                 return 0;
 534         }
 535         catch(exception& e) {
 536                 m->errorOut(e, "HeatMapSimCommand", "runCommandDist");
 537                 exit(1);
 538         }
 539 }
 540 //**********************************************************************************************************************
 541
 542
 543
 544
 545
 546