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                         //valid paramters for this command
  80                         string Array[] =  {"cutoff","hard","precision","method","phylip","column","name","sorted","showabund","timing","outputdir","inputdir"};
  81                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  82
  83                         OptionParser parser(option);
  84                         map<string,string> parameters = parser.getParameters();
  85
  86                         ValidParameters validParameter;
  87                         map<string,string>::iterator it;
  88
  89                         //check to make sure all parameters are valid for command
  90                         for (it = parameters.begin(); it != parameters.end(); it++) {
  91                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {
  92                                         abort = true;
  93                                 }
  94                         }
  95
  96                         //initialize outputTypes
  97                         vector<string> tempOutNames;
  98                         outputTypes["list"] = tempOutNames;
  99                         outputTypes["rabund"] = tempOutNames;
 100                         outputTypes["sabund"] = tempOutNames;
 101
 102                         //if the user changes the input directory command factory will send this info to us in the output parameter
 103                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 104                         if (inputDir == "not found"){   inputDir = "";          }
 105                         else {
 106                                 string path;
 107                                 it = parameters.find("phylip");
 108                                 //user has given a template file
 109                                 if(it != parameters.end()){
 110                                         path = m->hasPath(it->second);
 111                                         //if the user has not given a path then, add inputdir. else leave path alone.
 112                                         if (path == "") {       parameters["phylip"] = inputDir + it->second;           }
 113                                 }
 114
 115                                 it = parameters.find("column");
 116                                 //user has given a template file
 117                                 if(it != parameters.end()){
 118                                         path = m->hasPath(it->second);
 119                                         //if the user has not given a path then, add inputdir. else leave path alone.
 120                                         if (path == "") {       parameters["column"] = inputDir + it->second;           }
 121                                 }
 122
 123                                 it = parameters.find("name");
 124                                 //user has given a template file
 125                                 if(it != parameters.end()){
 126                                         path = m->hasPath(it->second);
 127                                         //if the user has not given a path then, add inputdir. else leave path alone.
 128                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 129                                 }
 130                         }
 131
 132                         //if the user changes the output directory command factory will send this info to us in the output parameter
 133                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 134
 135                         //check for required parameters
 136                         phylipfile = validParameter.validFile(parameters, "phylip", true);
 137                         if (phylipfile == "not open") { abort = true; }
 138                         else if (phylipfile == "not found") { phylipfile = ""; }
 139                         else {  distfile = phylipfile;  format = "phylip";      }
 140
 141                         columnfile = validParameter.validFile(parameters, "column", true);
 142                         if (columnfile == "not open") { abort = true; }
 143                         else if (columnfile == "not found") { columnfile = ""; }
 144                         else {  distfile = columnfile; format = "column";       }
 145
 146                         namefile = validParameter.validFile(parameters, "name", true);
 147                         if (namefile == "not open") { abort = true; }
 148                         else if (namefile == "not found") { namefile = ""; }
 149
 150                         if ((phylipfile == "") && (columnfile == "")) {
 151                                 //is there are current file available for either of these?
 152                                 //give priority to column, then phylip
 153                                 columnfile = m->getColumnFile();
 154                                 if (columnfile != "") {  m->mothurOut("Using " + columnfile + " as input file for the column parameter."); m->mothurOutEndLine(); }
 155                                 else {
 156                                         phylipfile = m->getPhylipFile();
 157                                         if (phylipfile != "") {  m->mothurOut("Using " + phylipfile + " as input file for the phylip parameter."); m->mothurOutEndLine(); }
 158                                         else {
 159                                                 m->mothurOut("No valid current files. You must provide a phylip or column file before you can use the hcluster command."); m->mothurOutEndLine();
 160                                                 abort = true;
 161                                         }
 162                                 }
 163                         }
 164                         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; }
 165
 166                         if (columnfile != "") {
 167                                 if (namefile == "") {
 168                                         namefile = m->getNameFile();
 169                                         if (namefile != "") {  m->mothurOut("Using " + namefile + " as input file for the name parameter."); m->mothurOutEndLine(); }
 170                                         else {
 171                                                 m->mothurOut("You need to provide a namefile if you are going to use the column format."); m->mothurOutEndLine();
 172                                                 abort = true;
 173                                         }
 174                                 }
 175                         }
 176
 177                         //check for optional parameter and set defaults
 178                         // ...at some point should added some additional type checking...
 179                         //get user cutoff and precision or use defaults
 180                         string temp;
 181                         temp = validParameter.validFile(parameters, "precision", false);
 182                         if (temp == "not found") { temp = "100"; }
 183                         //saves precision legnth for formatting below
 184                         length = temp.length();
 185                         convert(temp, precision);
 186
 187                         temp = validParameter.validFile(parameters, "hard", false);                     if (temp == "not found") { temp = "F"; }
 188                         hard = m->isTrue(temp);
 189
 190                         temp = validParameter.validFile(parameters, "cutoff", false);
 191                         if (temp == "not found") { temp = "10"; }
 192                         convert(temp, cutoff);
 193                         cutoff += (5 / (precision * 10.0));
 194
 195                         method = validParameter.validFile(parameters, "method", false);
 196                         if (method == "not found") { method = "furthest"; }
 197
 198                         if ((method == "furthest") || (method == "nearest") || (method == "average") || (method == "weighted")) { }
 199                         else { m->mothurOut("Not a valid clustering method.  Valid clustering algorithms are furthest, nearest, average or weighted."); m->mothurOutEndLine(); abort = true; }
 200
 201                         showabund = validParameter.validFile(parameters, "showabund", false);
 202                         if (showabund == "not found") { showabund = "T"; }
 203
 204                         sort = validParameter.validFile(parameters, "sorted", false);
 205                         if (sort == "not found") { sort = "F"; }
 206                         sorted = m->isTrue(sort);
 207
 208                         timing = validParameter.validFile(parameters, "timing", false);
 209                         if (timing == "not found") { timing = "F"; }
 210
 211
 212                         if (abort == false) {
 213
 214                                 if (outputDir == "") {  outputDir += m->hasPath(distfile); }
 215                                 fileroot = outputDir + m->getRootName(m->getSimpleName(distfile));
 216
 217                                 if (method == "furthest")               { tag = "fn";  }
 218                                 else if (method == "nearest")   { tag = "nn";  }
 219                                 else if (method == "weighted")  { tag = "wn";  }
 220                                 else                                                    { tag = "an";  }
 221
 222                                 m->openOutputFile(fileroot+ tag + ".sabund",    sabundFile);
 223                                 m->openOutputFile(fileroot+ tag + ".rabund",    rabundFile);
 224                                 m->openOutputFile(fileroot+ tag + ".list",              listFile);
 225
 226                                 outputNames.push_back(fileroot+ tag + ".sabund"); outputTypes["sabund"].push_back(fileroot+ tag + ".sabund");
 227                                 outputNames.push_back(fileroot+ tag + ".rabund"); outputTypes["rabund"].push_back(fileroot+ tag + ".rabund");
 228                                 outputNames.push_back(fileroot+ tag + ".list"); outputTypes["list"].push_back(fileroot+ tag + ".list");
 229                         }
 230                 }
 231         }
 232         catch(exception& e) {
 233                 m->errorOut(e, "HClusterCommand", "HClusterCommand");
 234                 exit(1);
 235         }
 236 }
 237
 238 //**********************************************************************************************************************
 239
 240 int HClusterCommand::execute(){
 241         try {
 242
 243                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 244
 245                 NameAssignment* nameMap = NULL;
 246                 if(namefile != ""){
 247                         nameMap = new NameAssignment(namefile);
 248                         nameMap->readMap();
 249                 }
 250
 251                 time_t estart = time(NULL);
 252
 253                 if (!sorted) {
 254                         read = new ReadCluster(distfile, cutoff, outputDir, true);
 255                         read->setFormat(format);
 256                         read->read(nameMap);
 257
 258                         if (m->control_pressed) {
 259                                 delete read;
 260                                 sabundFile.close();
 261                                 rabundFile.close();
 262                                 listFile.close();
 263                                 for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());  } outputTypes.clear();
 264                                 return 0;
 265                         }
 266
 267                         distfile = read->getOutputFile();
 268
 269                         list = read->getListVector();
 270                         delete read;
 271                 }else {
 272                         list = new ListVector(nameMap->getListVector());
 273                 }
 274
 275                 if (m->control_pressed) {
 276                         sabundFile.close();
 277                         rabundFile.close();
 278                         listFile.close();
 279                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());  } outputTypes.clear();
 280                         return 0;
 281                 }
 282
 283                 m->mothurOut("It took " + toString(time(NULL) - estart) + " seconds to sort. "); m->mothurOutEndLine();
 284                 estart = time(NULL);
 285
 286                 //list vector made by read contains all sequence names
 287                 if(list != NULL){
 288                         rabund = new RAbundVector(list->getRAbundVector());
 289                 }else{
 290                         m->mothurOut("Error: no list vector!"); m->mothurOutEndLine(); return 0;
 291                 }
 292
 293                 float previousDist = 0.00000;
 294                 float rndPreviousDist = 0.00000;
 295                 oldRAbund = *rabund;
 296                 oldList = *list;
 297
 298                 print_start = true;
 299                 start = time(NULL);
 300
 301                 cluster = new HCluster(rabund, list, method, distfile, nameMap, cutoff);
 302                 vector<seqDist> seqs; seqs.resize(1); // to start loop
 303
 304                 if (m->control_pressed) {
 305                                 delete cluster;
 306                                 sabundFile.close();
 307                                 rabundFile.close();
 308                                 listFile.close();
 309                                 for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());  } outputTypes.clear();
 310                                 return 0;
 311                 }
 312
 313
 314                 while (seqs.size() != 0){
 315
 316                         seqs = cluster->getSeqs();
 317
 318                         if (m->control_pressed) {
 319                                 delete cluster;
 320                                 sabundFile.close();
 321                                 rabundFile.close();
 322                                 listFile.close();
 323                                 for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());  } outputTypes.clear();
 324                                 return 0;
 325                         }
 326
 327                         for (int i = 0; i < seqs.size(); i++) {  //-1 means skip me
 328
 329                                 if (seqs[i].seq1 != seqs[i].seq2) {
 330                                         cluster->update(seqs[i].seq1, seqs[i].seq2, seqs[i].dist);
 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++) {  remove(outputNames[i].c_str());  } outputTypes.clear();
 338                                                 return 0;
 339                                         }
 340
 341
 342                                         float rndDist;
 343                                         if (hard) {
 344                                                 rndDist = m->ceilDist(seqs[i].dist, precision);
 345                                         }else{
 346                                                 rndDist = m->roundDist(seqs[i].dist, precision);
 347                                         }
 348
 349
 350                                         if((previousDist <= 0.0000) && (seqs[i].dist != previousDist)){
 351                                                 printData("unique");
 352                                         }
 353                                         else if((rndDist != rndPreviousDist)){
 354                                                 printData(toString(rndPreviousDist,  length-1));
 355                                         }
 356
 357                                         previousDist = seqs[i].dist;
 358                                         rndPreviousDist = rndDist;
 359                                         oldRAbund = *rabund;
 360                                         oldList = *list;
 361                                 }
 362                         }
 363                 }
 364
 365                 if (m->control_pressed) {
 366                         delete cluster;
 367                         sabundFile.close();
 368                         rabundFile.close();
 369                         listFile.close();
 370                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());  } outputTypes.clear();
 371                         return 0;
 372                 }
 373
 374                 if(previousDist <= 0.0000){
 375                         printData("unique");
 376                 }
 377                 else if(rndPreviousDist<cutoff){
 378                         printData(toString(rndPreviousDist, length-1));
 379                 }
 380
 381                 sabundFile.close();
 382                 rabundFile.close();
 383                 listFile.close();
 384                 delete cluster;
 385
 386                 if (m->control_pressed) {
 387                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());  } outputTypes.clear();
 388                         return 0;
 389                 }
 390
 391                 //set list file as new current listfile
 392                 string current = "";
 393                 itTypes = outputTypes.find("list");
 394                 if (itTypes != outputTypes.end()) {
 395                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 396                 }
 397
 398                 //set rabund file as new current rabundfile
 399                 itTypes = outputTypes.find("rabund");
 400                 if (itTypes != outputTypes.end()) {
 401                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setRabundFile(current); }
 402                 }
 403
 404                 //set sabund file as new current sabundfile
 405                 itTypes = outputTypes.find("sabund");
 406                 if (itTypes != outputTypes.end()) {
 407                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setSabundFile(current); }
 408                 }
 409
 410
 411                 m->mothurOutEndLine();
 412                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 413                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 414                 m->mothurOutEndLine();
 415
 416                 m->mothurOut("It took " + toString(time(NULL) - estart) + " seconds to cluster. "); m->mothurOutEndLine();
 417
 418                 return 0;
 419         }
 420         catch(exception& e) {
 421                 m->errorOut(e, "HClusterCommand", "execute");
 422                 exit(1);
 423         }
 424 }
 425
 426 //**********************************************************************************************************************
 427
 428 void HClusterCommand::printData(string label){
 429         try {
 430                 if (m->isTrue(timing)) {
 431                         m->mothurOut("\tTime: " + toString(time(NULL) - start) + "\tsecs for " + toString(oldRAbund.getNumBins())
 432                      + "\tclusters. Updates: " + toString(loops)); m->mothurOutEndLine();
 433                 }
 434                 print_start = true;
 435                 loops = 0;
 436                 start = time(NULL);
 437
 438                 oldRAbund.setLabel(label);
 439                 if (m->isTrue(showabund)) {
 440                         oldRAbund.getSAbundVector().print(cout);
 441                 }
 442                 oldRAbund.print(rabundFile);
 443                 oldRAbund.getSAbundVector().print(sabundFile);
 444
 445                 oldList.setLabel(label);
 446                 oldList.print(listFile);
 447         }
 448         catch(exception& e) {
 449                 m->errorOut(e, "HClusterCommand", "printData");
 450                 exit(1);
 451         }
 452
 453
 454 }
 455 //**********************************************************************************************************************
 456