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