hclustercommand.cpp

   1 /*
   2  *  hclustercommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 10/13/09.
   6  *  Copyright 2009 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "hclustercommand.h"
  11
  12 //**********************************************************************************************************************
  13 vector<string> HClusterCommand::setParameters(){
  14         try {
  15                 CommandParameter pphylip("phylip", "InputTypes", "", "", "PhylipColumn", "PhylipColumn", "none","list-rabund-sabund",false,false,true); parameters.push_back(pphylip);
  16                 CommandParameter pname("name", "InputTypes", "", "", "none", "none", "ColumnName","",false,false,true); parameters.push_back(pname);
  17                 CommandParameter pcolumn("column", "InputTypes", "", "", "PhylipColumn", "PhylipColumn", "ColumnName","list-rabund-sabund",false,false,true); parameters.push_back(pcolumn);
  18                 CommandParameter pcutoff("cutoff", "Number", "", "10", "", "", "","",false,false,true); parameters.push_back(pcutoff);
  19                 CommandParameter pprecision("precision", "Number", "", "100", "", "", "","",false,false); parameters.push_back(pprecision);
  20                 CommandParameter pmethod("method", "Multiple", "furthest-nearest-average-weighted", "average", "", "", "","",false,false); parameters.push_back(pmethod);
  21                 CommandParameter phard("hard", "Boolean", "", "T", "", "", "","",false,false); parameters.push_back(phard);
  22                 CommandParameter psorted("sorted", "Boolean", "", "F", "", "", "","",false,false); parameters.push_back(psorted);
  23                 CommandParameter pshowabund("showabund", "Boolean", "", "T", "", "", "","",false,false); parameters.push_back(pshowabund);
  24                 CommandParameter ptiming("timing", "Boolean", "", "F", "", "", "","",false,false); parameters.push_back(ptiming);
  25                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(pinputdir);
  26                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(poutputdir);
  27
  28                 vector<string> myArray;
  29                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  30                 return myArray;
  31         }
  32         catch(exception& e) {
  33                 m->errorOut(e, "HClusterCommand", "setParameters");
  34                 exit(1);
  35         }
  36 }
  37 //**********************************************************************************************************************
  38 string HClusterCommand::getHelpString(){
  39         try {
  40                 string helpString = "";
  41                 helpString += "The hcluster command parameter options are cutoff, precision, method, phylip, column, name, showabund, timing and sorted. Phylip or column and name are required, unless you have valid current files.\n";
  42                 helpString += "The phylip and column parameter allow you to enter your distance file, and sorted indicates whether your column distance file is already sorted. \n";
  43                 helpString += "The name parameter allows you to enter your name file and is required if your distance file is in column format. \n";
  44                 helpString += "The hcluster command should be in the following format: \n";
  45                 helpString += "hcluster(column=youDistanceFile, name=yourNameFile, method=yourMethod, cutoff=yourCutoff, precision=yourPrecision) \n";
  46                 helpString += "The acceptable hcluster methods are furthest, nearest, weighted and average.\n";
  47                 return helpString;
  48         }
  49         catch(exception& e) {
  50                 m->errorOut(e, "HClusterCommand", "getHelpString");
  51                 exit(1);
  52         }
  53 }
  54 //**********************************************************************************************************************
  55 string HClusterCommand::getOutputPattern(string type) {
  56     try {
  57         string pattern = "";
  58
  59         if (type == "list") {  pattern = "[filename],[clustertag],list"; }
  60         else if (type == "rabund") {  pattern = "[filename],[clustertag],rabund"; }
  61         else if (type == "sabund") {  pattern = "[filename],[clustertag],sabund"; }
  62         else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true;  }
  63
  64         return pattern;
  65     }
  66     catch(exception& e) {
  67         m->errorOut(e, "HClusterCommand", "getOutputPattern");
  68         exit(1);
  69     }
  70 }
  71
  72 //**********************************************************************************************************************
  73 HClusterCommand::HClusterCommand(){
  74         try {
  75                 abort = true; calledHelp = true;
  76                 setParameters();
  77                 vector<string> tempOutNames;
  78                 outputTypes["list"] = tempOutNames;
  79                 outputTypes["rabund"] = tempOutNames;
  80                 outputTypes["sabund"] = tempOutNames;
  81         }
  82         catch(exception& e) {
  83                 m->errorOut(e, "HClusterCommand", "HClusterCommand");
  84                 exit(1);
  85         }
  86 }
  87 //**********************************************************************************************************************
  88 //This function checks to make sure the cluster command has no errors and then clusters based on the method chosen.
  89 HClusterCommand::HClusterCommand(string option)  {
  90         try{
  91                 abort = false; calledHelp = false;
  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) {
 109                                         abort = true;
 110                                 }
 111                         }
 112
 113                         //initialize outputTypes
 114                         vector<string> tempOutNames;
 115                         outputTypes["list"] = tempOutNames;
 116                         outputTypes["rabund"] = tempOutNames;
 117                         outputTypes["sabund"] = tempOutNames;
 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
 149                         //if the user changes the output directory command factory will send this info to us in the output parameter
 150                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 151
 152                         //check for required parameters
 153                         phylipfile = validParameter.validFile(parameters, "phylip", true);
 154                         if (phylipfile == "not open") { abort = true; }
 155                         else if (phylipfile == "not found") { phylipfile = ""; }
 156                         else {  distfile = phylipfile;  format = "phylip";      m->setPhylipFile(phylipfile); }
 157
 158                         columnfile = validParameter.validFile(parameters, "column", true);
 159                         if (columnfile == "not open") { abort = true; }
 160                         else if (columnfile == "not found") { columnfile = ""; }
 161                         else {  distfile = columnfile; format = "column";       m->setColumnFile(columnfile); }
 162
 163                         namefile = validParameter.validFile(parameters, "name", true);
 164                         if (namefile == "not open") { abort = true; }
 165                         else if (namefile == "not found") { namefile = ""; }
 166                         else { m->setNameFile(namefile); }
 167
 168                         if ((phylipfile == "") && (columnfile == "")) {
 169                                 //is there are current file available for either of these?
 170                                 //give priority to column, then phylip
 171                                 columnfile = m->getColumnFile();
 172                                 if (columnfile != "") {  m->mothurOut("Using " + columnfile + " as input file for the column parameter."); m->mothurOutEndLine(); }
 173                                 else {
 174                                         phylipfile = m->getPhylipFile();
 175                                         if (phylipfile != "") {  m->mothurOut("Using " + phylipfile + " as input file for the phylip parameter."); m->mothurOutEndLine(); }
 176                                         else {
 177                                                 m->mothurOut("No valid current files. You must provide a phylip or column file before you can use the hcluster command."); m->mothurOutEndLine();
 178                                                 abort = true;
 179                                         }
 180                                 }
 181                         }
 182                         else if ((phylipfile != "") && (columnfile != "")) { m->mothurOut("When executing a hcluster command you must enter ONLY ONE of the following: phylip or column."); m->mothurOutEndLine(); abort = true; }
 183
 184                         if (columnfile != "") {
 185                                 if (namefile == "") {
 186                                         namefile = m->getNameFile();
 187                                         if (namefile != "") {  m->mothurOut("Using " + namefile + " as input file for the name parameter."); m->mothurOutEndLine(); }
 188                                         else {
 189                                                 m->mothurOut("You need to provide a namefile if you are going to use the column format."); m->mothurOutEndLine();
 190                                                 abort = true;
 191                                         }
 192                                 }
 193                         }
 194
 195                         //check for optional parameter and set defaults
 196                         // ...at some point should added some additional type checking...
 197                         //get user cutoff and precision or use defaults
 198                         string temp;
 199                         temp = validParameter.validFile(parameters, "precision", false);
 200                         if (temp == "not found") { temp = "100"; }
 201                         //saves precision legnth for formatting below
 202                         length = temp.length();
 203                         m->mothurConvert(temp, precision);
 204
 205                         temp = validParameter.validFile(parameters, "hard", false);                     if (temp == "not found") { temp = "T"; }
 206                         hard = m->isTrue(temp);
 207
 208                         temp = validParameter.validFile(parameters, "cutoff", false);
 209                         if (temp == "not found") { temp = "10"; }
 210                         m->mothurConvert(temp, cutoff);
 211                         cutoff += (5 / (precision * 10.0));
 212
 213                         method = validParameter.validFile(parameters, "method", false);
 214                         if (method == "not found") { method = "average"; }
 215
 216                         if ((method == "furthest") || (method == "nearest") || (method == "average") || (method == "weighted")) { }
 217                         else { m->mothurOut("Not a valid clustering method.  Valid clustering algorithms are furthest, nearest, average or weighted."); m->mothurOutEndLine(); abort = true; }
 218
 219                         showabund = validParameter.validFile(parameters, "showabund", false);
 220                         if (showabund == "not found") { showabund = "T"; }
 221
 222                         sort = validParameter.validFile(parameters, "sorted", false);
 223                         if (sort == "not found") { sort = "F"; }
 224                         sorted = m->isTrue(sort);
 225
 226                         timing = validParameter.validFile(parameters, "timing", false);
 227                         if (timing == "not found") { timing = "F"; }
 228
 229
 230                         if (abort == false) {
 231
 232                                 if (outputDir == "") {  outputDir += m->hasPath(distfile); }
 233                                 fileroot = outputDir + m->getRootName(m->getSimpleName(distfile));
 234
 235                                 if (method == "furthest")               { tag = "fn";  }
 236                                 else if (method == "nearest")   { tag = "nn";  }
 237                                 else if (method == "weighted")  { tag = "wn";  }
 238                                 else                                                    { tag = "an";  }
 239
 240                 map<string, string> variables;
 241                 variables["[filename]"] = fileroot;
 242                 variables["[clustertag]"] = tag;
 243
 244                                 string sabundFileName = getOutputFileName("sabund",variables);
 245                 string rabundFileName = getOutputFileName("rabund",variables);
 246                 string listFileName = getOutputFileName("list", variables);
 247
 248                 m->openOutputFile(sabundFileName,       sabundFile);
 249                 m->openOutputFile(rabundFileName,       rabundFile);
 250                 m->openOutputFile(listFileName, listFile);
 251
 252                 outputNames.push_back(sabundFileName); outputTypes["sabund"].push_back(sabundFileName);
 253                 outputNames.push_back(rabundFileName); outputTypes["rabund"].push_back(rabundFileName);
 254                 outputNames.push_back(listFileName); outputTypes["list"].push_back(listFileName);
 255                         }
 256                 }
 257         }
 258         catch(exception& e) {
 259                 m->errorOut(e, "HClusterCommand", "HClusterCommand");
 260                 exit(1);
 261         }
 262 }
 263
 264 //**********************************************************************************************************************
 265
 266 int HClusterCommand::execute(){
 267         try {
 268
 269                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 270
 271                 NameAssignment* nameMap = NULL;
 272                 if(namefile != ""){
 273                         nameMap = new NameAssignment(namefile);
 274                         nameMap->readMap();
 275                 }
 276
 277                 time_t estart = time(NULL);
 278
 279                 if (!sorted) {
 280                         read = new ReadCluster(distfile, cutoff, outputDir, true);
 281                         read->setFormat(format);
 282                         read->read(nameMap);
 283
 284                         if (m->control_pressed) {
 285                                 delete read;
 286                                 sabundFile.close();
 287                                 rabundFile.close();
 288                                 listFile.close();
 289                                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]);  } outputTypes.clear();
 290                                 return 0;
 291                         }
 292
 293                         distfile = read->getOutputFile();
 294
 295                         list = read->getListVector();
 296                         delete read;
 297                 }else {
 298                         list = new ListVector(nameMap->getListVector());
 299                 }
 300
 301                 if (m->control_pressed) {
 302                         sabundFile.close();
 303                         rabundFile.close();
 304                         listFile.close();
 305                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]);  } outputTypes.clear();
 306                         return 0;
 307                 }
 308
 309                 m->mothurOut("It took " + toString(time(NULL) - estart) + " seconds to sort. "); m->mothurOutEndLine();
 310                 estart = time(NULL);
 311
 312                 //list vector made by read contains all sequence names
 313                 if(list != NULL){
 314                         rabund = new RAbundVector(list->getRAbundVector());
 315                 }else{
 316                         m->mothurOut("Error: no list vector!"); m->mothurOutEndLine(); return 0;
 317                 }
 318
 319         list->printHeaders(listFile);
 320
 321                 float previousDist = 0.00000;
 322                 float rndPreviousDist = 0.00000;
 323                 oldRAbund = *rabund;
 324                 oldList = *list;
 325
 326                 print_start = true;
 327                 start = time(NULL);
 328
 329                 cluster = new HCluster(rabund, list, method, distfile, nameMap, cutoff);
 330                 vector<seqDist> seqs; seqs.resize(1); // to start loop
 331
 332                 if (m->control_pressed) {
 333                                 delete cluster;
 334                                 sabundFile.close();
 335                                 rabundFile.close();
 336                                 listFile.close();
 337                                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]);  } outputTypes.clear();
 338                                 return 0;
 339                 }
 340
 341                 float saveCutoff = cutoff;
 342
 343                 while (seqs.size() != 0){
 344
 345                         seqs = cluster->getSeqs();
 346
 347                         //to account for cutoff change in average neighbor
 348                         if (seqs.size() != 0) {
 349                                 if (seqs[0].dist > cutoff) { break; }
 350                         }
 351
 352                         if (m->control_pressed) {
 353                                 delete cluster;
 354                                 sabundFile.close();
 355                                 rabundFile.close();
 356                                 listFile.close();
 357                                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]);  } outputTypes.clear();
 358                                 return 0;
 359                         }
 360
 361                         for (int i = 0; i < seqs.size(); i++) {  //-1 means skip me
 362
 363                                 if (seqs[i].seq1 != seqs[i].seq2) {
 364                                         cutoff = cluster->update(seqs[i].seq1, seqs[i].seq2, seqs[i].dist);
 365
 366                                         if (m->control_pressed) {
 367                                                 delete cluster;
 368                                                 sabundFile.close();
 369                                                 rabundFile.close();
 370                                                 listFile.close();
 371                                                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]);  } outputTypes.clear();
 372                                                 return 0;
 373                                         }
 374
 375
 376                                         float rndDist;
 377                                         if (hard) {
 378                                                 rndDist = m->ceilDist(seqs[i].dist, precision);
 379                                         }else{
 380                                                 rndDist = m->roundDist(seqs[i].dist, precision);
 381                                         }
 382
 383
 384                                         if((previousDist <= 0.0000) && (seqs[i].dist != previousDist)){
 385                                                 printData("unique");
 386                                         }
 387                                         else if((rndDist != rndPreviousDist)){
 388                                                 printData(toString(rndPreviousDist,  length-1));
 389                                         }
 390
 391                                         previousDist = seqs[i].dist;
 392                                         rndPreviousDist = rndDist;
 393                                         oldRAbund = *rabund;
 394                                         oldList = *list;
 395                                 }
 396                         }
 397                 }
 398
 399                 if (m->control_pressed) {
 400                         delete cluster;
 401                         sabundFile.close();
 402                         rabundFile.close();
 403                         listFile.close();
 404                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]);  } outputTypes.clear();
 405                         return 0;
 406                 }
 407
 408                 if(previousDist <= 0.0000){
 409                         printData("unique");
 410                 }
 411                 else if(rndPreviousDist<cutoff){
 412                         printData(toString(rndPreviousDist, length-1));
 413                 }
 414
 415                 sabundFile.close();
 416                 rabundFile.close();
 417                 listFile.close();
 418                 delete cluster;
 419
 420                 if (m->control_pressed) {
 421                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]);  } outputTypes.clear();
 422                         return 0;
 423                 }
 424
 425
 426                 if (saveCutoff != cutoff) {
 427                         if (hard)       {  saveCutoff = m->ceilDist(saveCutoff, precision);     }
 428                         else            {       saveCutoff = m->roundDist(saveCutoff, precision);  }
 429
 430                         m->mothurOut("changed cutoff to " + toString(cutoff)); m->mothurOutEndLine();
 431                 }
 432
 433                 //set list file as new current listfile
 434                 string current = "";
 435                 itTypes = outputTypes.find("list");
 436                 if (itTypes != outputTypes.end()) {
 437                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 438                 }
 439
 440                 //set rabund file as new current rabundfile
 441                 itTypes = outputTypes.find("rabund");
 442                 if (itTypes != outputTypes.end()) {
 443                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setRabundFile(current); }
 444                 }
 445
 446                 //set sabund file as new current sabundfile
 447                 itTypes = outputTypes.find("sabund");
 448                 if (itTypes != outputTypes.end()) {
 449                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setSabundFile(current); }
 450                 }
 451
 452
 453                 m->mothurOutEndLine();
 454                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 455                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 456                 m->mothurOutEndLine();
 457
 458                 m->mothurOut("It took " + toString(time(NULL) - estart) + " seconds to cluster. "); m->mothurOutEndLine();
 459
 460                 return 0;
 461         }
 462         catch(exception& e) {
 463                 m->errorOut(e, "HClusterCommand", "execute");
 464                 exit(1);
 465         }
 466 }
 467
 468 //**********************************************************************************************************************
 469
 470 void HClusterCommand::printData(string label){
 471         try {
 472                 if (m->isTrue(timing)) {
 473                         m->mothurOut("\tTime: " + toString(time(NULL) - start) + "\tsecs for " + toString(oldRAbund.getNumBins())
 474                      + "\tclusters. Updates: " + toString(loops)); m->mothurOutEndLine();
 475                 }
 476                 print_start = true;
 477                 loops = 0;
 478                 start = time(NULL);
 479
 480                 oldRAbund.setLabel(label);
 481                 if (m->isTrue(showabund)) {
 482                         oldRAbund.getSAbundVector().print(cout);
 483                 }
 484                 oldRAbund.print(rabundFile);
 485                 oldRAbund.getSAbundVector().print(sabundFile);
 486
 487                 oldList.setLabel(label);
 488                 oldList.print(listFile);
 489         }
 490         catch(exception& e) {
 491                 m->errorOut(e, "HClusterCommand", "printData");
 492                 exit(1);
 493         }
 494
 495
 496 }
 497 //**********************************************************************************************************************
 498