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