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",false,false); parameters.push_back(pphylip);
  16                 CommandParameter pname("name", "InputTypes", "", "", "none", "none", "ColumnName",false,false); parameters.push_back(pname);
  17                 CommandParameter pcolumn("column", "InputTypes", "", "", "PhylipColumn", "PhylipColumn", "ColumnName",false,false); parameters.push_back(pcolumn);
  18                 CommandParameter pcutoff("cutoff", "Number", "", "10", "", "", "",false,false); 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 HClusterCommand::HClusterCommand(){
  56         try {
  57                 abort = true; calledHelp = true;
  58                 setParameters();
  59                 vector<string> tempOutNames;
  60                 outputTypes["list"] = tempOutNames;
  61                 outputTypes["rabund"] = tempOutNames;
  62                 outputTypes["sabund"] = tempOutNames;
  63         }
  64         catch(exception& e) {
  65                 m->errorOut(e, "HClusterCommand", "HClusterCommand");
  66                 exit(1);
  67         }
  68 }
  69 //**********************************************************************************************************************
  70 //This function checks to make sure the cluster command has no errors and then clusters based on the method chosen.
  71 HClusterCommand::HClusterCommand(string option)  {
  72         try{
  73                 abort = false; calledHelp = false;
  74
  75                 //allow user to run help
  76                 if(option == "help") { help(); abort = true; calledHelp = true; }
  77                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  78
  79                 else {
  80                         vector<string> myArray = setParameters();
  81
  82                         OptionParser parser(option);
  83                         map<string,string> parameters = parser.getParameters();
  84
  85                         ValidParameters validParameter;
  86                         map<string,string>::iterator it;
  87
  88                         //check to make sure all parameters are valid for command
  89                         for (it = parameters.begin(); it != parameters.end(); it++) {
  90                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {
  91                                         abort = true;
  92                                 }
  93                         }
  94
  95                         //initialize outputTypes
  96                         vector<string> tempOutNames;
  97                         outputTypes["list"] = tempOutNames;
  98                         outputTypes["rabund"] = tempOutNames;
  99                         outputTypes["sabund"] = tempOutNames;
 100
 101                         //if the user changes the input directory command factory will send this info to us in the output parameter
 102                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 103                         if (inputDir == "not found"){   inputDir = "";          }
 104                         else {
 105                                 string path;
 106                                 it = parameters.find("phylip");
 107                                 //user has given a template file
 108                                 if(it != parameters.end()){
 109                                         path = m->hasPath(it->second);
 110                                         //if the user has not given a path then, add inputdir. else leave path alone.
 111                                         if (path == "") {       parameters["phylip"] = inputDir + it->second;           }
 112                                 }
 113
 114                                 it = parameters.find("column");
 115                                 //user has given a template file
 116                                 if(it != parameters.end()){
 117                                         path = m->hasPath(it->second);
 118                                         //if the user has not given a path then, add inputdir. else leave path alone.
 119                                         if (path == "") {       parameters["column"] = inputDir + it->second;           }
 120                                 }
 121
 122                                 it = parameters.find("name");
 123                                 //user has given a template file
 124                                 if(it != parameters.end()){
 125                                         path = m->hasPath(it->second);
 126                                         //if the user has not given a path then, add inputdir. else leave path alone.
 127                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 128                                 }
 129                         }
 130
 131                         //if the user changes the output directory command factory will send this info to us in the output parameter
 132                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 133
 134                         //check for required parameters
 135                         phylipfile = validParameter.validFile(parameters, "phylip", true);
 136                         if (phylipfile == "not open") { abort = true; }
 137                         else if (phylipfile == "not found") { phylipfile = ""; }
 138                         else {  distfile = phylipfile;  format = "phylip";      }
 139
 140                         columnfile = validParameter.validFile(parameters, "column", true);
 141                         if (columnfile == "not open") { abort = true; }
 142                         else if (columnfile == "not found") { columnfile = ""; }
 143                         else {  distfile = columnfile; format = "column";       }
 144
 145                         namefile = validParameter.validFile(parameters, "name", true);
 146                         if (namefile == "not open") { abort = true; }
 147                         else if (namefile == "not found") { namefile = ""; }
 148
 149                         if ((phylipfile == "") && (columnfile == "")) {
 150                                 //is there are current file available for either of these?
 151                                 //give priority to column, then phylip
 152                                 columnfile = m->getColumnFile();
 153                                 if (columnfile != "") {  m->mothurOut("Using " + columnfile + " as input file for the column parameter."); m->mothurOutEndLine(); }
 154                                 else {
 155                                         phylipfile = m->getPhylipFile();
 156                                         if (phylipfile != "") {  m->mothurOut("Using " + phylipfile + " as input file for the phylip parameter."); m->mothurOutEndLine(); }
 157                                         else {
 158                                                 m->mothurOut("No valid current files. You must provide a phylip or column file before you can use the hcluster command."); m->mothurOutEndLine();
 159                                                 abort = true;
 160                                         }
 161                                 }
 162                         }
 163                         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; }
 164
 165                         if (columnfile != "") {
 166                                 if (namefile == "") {
 167                                         namefile = m->getNameFile();
 168                                         if (namefile != "") {  m->mothurOut("Using " + namefile + " as input file for the name parameter."); m->mothurOutEndLine(); }
 169                                         else {
 170                                                 m->mothurOut("You need to provide a namefile if you are going to use the column format."); m->mothurOutEndLine();
 171                                                 abort = true;
 172                                         }
 173                                 }
 174                         }
 175
 176                         //check for optional parameter and set defaults
 177                         // ...at some point should added some additional type checking...
 178                         //get user cutoff and precision or use defaults
 179                         string temp;
 180                         temp = validParameter.validFile(parameters, "precision", false);
 181                         if (temp == "not found") { temp = "100"; }
 182                         //saves precision legnth for formatting below
 183                         length = temp.length();
 184                         convert(temp, precision);
 185
 186                         temp = validParameter.validFile(parameters, "hard", false);                     if (temp == "not found") { temp = "T"; }
 187                         hard = m->isTrue(temp);
 188
 189                         temp = validParameter.validFile(parameters, "cutoff", false);
 190                         if (temp == "not found") { temp = "10"; }
 191                         convert(temp, cutoff);
 192                         cutoff += (5 / (precision * 10.0));
 193
 194                         method = validParameter.validFile(parameters, "method", false);
 195                         if (method == "not found") { method = "average"; }
 196
 197                         if ((method == "furthest") || (method == "nearest") || (method == "average") || (method == "weighted")) { }
 198                         else { m->mothurOut("Not a valid clustering method.  Valid clustering algorithms are furthest, nearest, average or weighted."); m->mothurOutEndLine(); abort = true; }
 199
 200                         showabund = validParameter.validFile(parameters, "showabund", false);
 201                         if (showabund == "not found") { showabund = "T"; }
 202
 203                         sort = validParameter.validFile(parameters, "sorted", false);
 204                         if (sort == "not found") { sort = "F"; }
 205                         sorted = m->isTrue(sort);
 206
 207                         timing = validParameter.validFile(parameters, "timing", false);
 208                         if (timing == "not found") { timing = "F"; }
 209
 210
 211                         if (abort == false) {
 212
 213                                 if (outputDir == "") {  outputDir += m->hasPath(distfile); }
 214                                 fileroot = outputDir + m->getRootName(m->getSimpleName(distfile));
 215
 216                                 if (method == "furthest")               { tag = "fn";  }
 217                                 else if (method == "nearest")   { tag = "nn";  }
 218                                 else if (method == "weighted")  { tag = "wn";  }
 219                                 else                                                    { tag = "an";  }
 220
 221                                 m->openOutputFile(fileroot+ tag + ".sabund",    sabundFile);
 222                                 m->openOutputFile(fileroot+ tag + ".rabund",    rabundFile);
 223                                 m->openOutputFile(fileroot+ tag + ".list",              listFile);
 224
 225                                 outputNames.push_back(fileroot+ tag + ".sabund"); outputTypes["sabund"].push_back(fileroot+ tag + ".sabund");
 226                                 outputNames.push_back(fileroot+ tag + ".rabund"); outputTypes["rabund"].push_back(fileroot+ tag + ".rabund");
 227                                 outputNames.push_back(fileroot+ tag + ".list"); outputTypes["list"].push_back(fileroot+ tag + ".list");
 228                         }
 229                 }
 230         }
 231         catch(exception& e) {
 232                 m->errorOut(e, "HClusterCommand", "HClusterCommand");
 233                 exit(1);
 234         }
 235 }
 236
 237 //**********************************************************************************************************************
 238
 239 int HClusterCommand::execute(){
 240         try {
 241
 242                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 243
 244                 NameAssignment* nameMap = NULL;
 245                 if(namefile != ""){
 246                         nameMap = new NameAssignment(namefile);
 247                         nameMap->readMap();
 248                 }
 249
 250                 time_t estart = time(NULL);
 251
 252                 if (!sorted) {
 253                         read = new ReadCluster(distfile, cutoff, outputDir, true);
 254                         read->setFormat(format);
 255                         read->read(nameMap);
 256
 257                         if (m->control_pressed) {
 258                                 delete read;
 259                                 sabundFile.close();
 260                                 rabundFile.close();
 261                                 listFile.close();
 262                                 for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());  } outputTypes.clear();
 263                                 return 0;
 264                         }
 265
 266                         distfile = read->getOutputFile();
 267
 268                         list = read->getListVector();
 269                         delete read;
 270                 }else {
 271                         list = new ListVector(nameMap->getListVector());
 272                 }
 273
 274                 if (m->control_pressed) {
 275                         sabundFile.close();
 276                         rabundFile.close();
 277                         listFile.close();
 278                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());  } outputTypes.clear();
 279                         return 0;
 280                 }
 281
 282                 m->mothurOut("It took " + toString(time(NULL) - estart) + " seconds to sort. "); m->mothurOutEndLine();
 283                 estart = time(NULL);
 284
 285                 //list vector made by read contains all sequence names
 286                 if(list != NULL){
 287                         rabund = new RAbundVector(list->getRAbundVector());
 288                 }else{
 289                         m->mothurOut("Error: no list vector!"); m->mothurOutEndLine(); return 0;
 290                 }
 291
 292                 float previousDist = 0.00000;
 293                 float rndPreviousDist = 0.00000;
 294                 oldRAbund = *rabund;
 295                 oldList = *list;
 296
 297                 print_start = true;
 298                 start = time(NULL);
 299
 300                 cluster = new HCluster(rabund, list, method, distfile, nameMap, cutoff);
 301                 vector<seqDist> seqs; seqs.resize(1); // to start loop
 302
 303                 if (m->control_pressed) {
 304                                 delete cluster;
 305                                 sabundFile.close();
 306                                 rabundFile.close();
 307                                 listFile.close();
 308                                 for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());  } outputTypes.clear();
 309                                 return 0;
 310                 }
 311
 312
 313                 while (seqs.size() != 0){
 314
 315                         seqs = cluster->getSeqs();
 316
 317                         if (m->control_pressed) {
 318                                 delete cluster;
 319                                 sabundFile.close();
 320                                 rabundFile.close();
 321                                 listFile.close();
 322                                 for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());  } outputTypes.clear();
 323                                 return 0;
 324                         }
 325
 326                         for (int i = 0; i < seqs.size(); i++) {  //-1 means skip me
 327
 328                                 if (seqs[i].seq1 != seqs[i].seq2) {
 329                                         cluster->update(seqs[i].seq1, seqs[i].seq2, seqs[i].dist);
 330
 331                                         if (m->control_pressed) {
 332                                                 delete cluster;
 333                                                 sabundFile.close();
 334                                                 rabundFile.close();
 335                                                 listFile.close();
 336                                                 for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());  } outputTypes.clear();
 337                                                 return 0;
 338                                         }
 339
 340
 341                                         float rndDist;
 342                                         if (hard) {
 343                                                 rndDist = m->ceilDist(seqs[i].dist, precision);
 344                                         }else{
 345                                                 rndDist = m->roundDist(seqs[i].dist, precision);
 346                                         }
 347
 348
 349                                         if((previousDist <= 0.0000) && (seqs[i].dist != previousDist)){
 350                                                 printData("unique");
 351                                         }
 352                                         else if((rndDist != rndPreviousDist)){
 353                                                 printData(toString(rndPreviousDist,  length-1));
 354                                         }
 355
 356                                         previousDist = seqs[i].dist;
 357                                         rndPreviousDist = rndDist;
 358                                         oldRAbund = *rabund;
 359                                         oldList = *list;
 360                                 }
 361                         }
 362                 }
 363
 364                 if (m->control_pressed) {
 365                         delete cluster;
 366                         sabundFile.close();
 367                         rabundFile.close();
 368                         listFile.close();
 369                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());  } outputTypes.clear();
 370                         return 0;
 371                 }
 372
 373                 if(previousDist <= 0.0000){
 374                         printData("unique");
 375                 }
 376                 else if(rndPreviousDist<cutoff){
 377                         printData(toString(rndPreviousDist, length-1));
 378                 }
 379
 380                 sabundFile.close();
 381                 rabundFile.close();
 382                 listFile.close();
 383                 delete cluster;
 384
 385                 if (m->control_pressed) {
 386                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());  } outputTypes.clear();
 387                         return 0;
 388                 }
 389
 390                 //set list file as new current listfile
 391                 string current = "";
 392                 itTypes = outputTypes.find("list");
 393                 if (itTypes != outputTypes.end()) {
 394                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 395                 }
 396
 397                 //set rabund file as new current rabundfile
 398                 itTypes = outputTypes.find("rabund");
 399                 if (itTypes != outputTypes.end()) {
 400                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setRabundFile(current); }
 401                 }
 402
 403                 //set sabund file as new current sabundfile
 404                 itTypes = outputTypes.find("sabund");
 405                 if (itTypes != outputTypes.end()) {
 406                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setSabundFile(current); }
 407                 }
 408
 409
 410                 m->mothurOutEndLine();
 411                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 412                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 413                 m->mothurOutEndLine();
 414
 415                 m->mothurOut("It took " + toString(time(NULL) - estart) + " seconds to cluster. "); m->mothurOutEndLine();
 416
 417                 return 0;
 418         }
 419         catch(exception& e) {
 420                 m->errorOut(e, "HClusterCommand", "execute");
 421                 exit(1);
 422         }
 423 }
 424
 425 //**********************************************************************************************************************
 426
 427 void HClusterCommand::printData(string label){
 428         try {
 429                 if (m->isTrue(timing)) {
 430                         m->mothurOut("\tTime: " + toString(time(NULL) - start) + "\tsecs for " + toString(oldRAbund.getNumBins())
 431                      + "\tclusters. Updates: " + toString(loops)); m->mothurOutEndLine();
 432                 }
 433                 print_start = true;
 434                 loops = 0;
 435                 start = time(NULL);
 436
 437                 oldRAbund.setLabel(label);
 438                 if (m->isTrue(showabund)) {
 439                         oldRAbund.getSAbundVector().print(cout);
 440                 }
 441                 oldRAbund.print(rabundFile);
 442                 oldRAbund.getSAbundVector().print(sabundFile);
 443
 444                 oldList.setLabel(label);
 445                 oldList.print(listFile);
 446         }
 447         catch(exception& e) {
 448                 m->errorOut(e, "HClusterCommand", "printData");
 449                 exit(1);
 450         }
 451
 452
 453 }
 454 //**********************************************************************************************************************
 455