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::getValidParameters(){
  14         try {
  15                 string Array[] =  {"cutoff","hard","precision","method","phylip","column","name","sorted","showabund","timing","outputdir","inputdir"};
  16                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  17                 return myArray;
  18         }
  19         catch(exception& e) {
  20                 m->errorOut(e, "HClusterCommand", "getValidParameters");
  21                 exit(1);
  22         }
  23 }
  24 //**********************************************************************************************************************
  25 HClusterCommand::HClusterCommand(){
  26         try {
  27                 abort = true;
  28                 //initialize outputTypes
  29                 vector<string> tempOutNames;
  30                 outputTypes["list"] = tempOutNames;
  31                 outputTypes["rabund"] = tempOutNames;
  32                 outputTypes["sabund"] = tempOutNames;
  33         }
  34         catch(exception& e) {
  35                 m->errorOut(e, "HClusterCommand", "HClusterCommand");
  36                 exit(1);
  37         }
  38 }
  39 //**********************************************************************************************************************
  40 vector<string> HClusterCommand::getRequiredParameters(){
  41         try {
  42                 string Array[] =  {"phylip","column","or"};
  43                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  44                 return myArray;
  45         }
  46         catch(exception& e) {
  47                 m->errorOut(e, "HClusterCommand", "getRequiredParameters");
  48                 exit(1);
  49         }
  50 }
  51 //**********************************************************************************************************************
  52 vector<string> HClusterCommand::getRequiredFiles(){
  53         try {
  54                 vector<string> myArray;
  55                 return myArray;
  56         }
  57         catch(exception& e) {
  58                 m->errorOut(e, "HClusterCommand", "getRequiredFiles");
  59                 exit(1);
  60         }
  61 }
  62 //**********************************************************************************************************************
  63 //This function checks to make sure the cluster command has no errors and then clusters based on the method chosen.
  64 HClusterCommand::HClusterCommand(string option)  {
  65         try{
  66                 globaldata = GlobalData::getInstance();
  67                 abort = false;
  68
  69                 //allow user to run help
  70                 if(option == "help") { help(); abort = true; }
  71
  72                 else {
  73                         //valid paramters for this command
  74                         string Array[] =  {"cutoff","hard","precision","method","phylip","column","name","sorted","showabund","timing","outputdir","inputdir"};
  75                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  76
  77                         OptionParser parser(option);
  78                         map<string,string> parameters = parser.getParameters();
  79
  80                         ValidParameters validParameter;
  81                         map<string,string>::iterator it;
  82
  83                         //check to make sure all parameters are valid for command
  84                         for (it = parameters.begin(); it != parameters.end(); it++) {
  85                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {
  86                                         abort = true;
  87                                 }
  88                         }
  89
  90                         //initialize outputTypes
  91                         vector<string> tempOutNames;
  92                         outputTypes["list"] = tempOutNames;
  93                         outputTypes["rabund"] = tempOutNames;
  94                         outputTypes["sabund"] = tempOutNames;
  95
  96                         globaldata->newRead();
  97
  98                         //if the user changes the input directory command factory will send this info to us in the output parameter
  99                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 100                         if (inputDir == "not found"){   inputDir = "";          }
 101                         else {
 102                                 string path;
 103                                 it = parameters.find("phylip");
 104                                 //user has given a template file
 105                                 if(it != parameters.end()){
 106                                         path = m->hasPath(it->second);
 107                                         //if the user has not given a path then, add inputdir. else leave path alone.
 108                                         if (path == "") {       parameters["phylip"] = inputDir + it->second;           }
 109                                 }
 110
 111                                 it = parameters.find("column");
 112                                 //user has given a template file
 113                                 if(it != parameters.end()){
 114                                         path = m->hasPath(it->second);
 115                                         //if the user has not given a path then, add inputdir. else leave path alone.
 116                                         if (path == "") {       parameters["column"] = inputDir + it->second;           }
 117                                 }
 118
 119                                 it = parameters.find("name");
 120                                 //user has given a template file
 121                                 if(it != parameters.end()){
 122                                         path = m->hasPath(it->second);
 123                                         //if the user has not given a path then, add inputdir. else leave path alone.
 124                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 125                                 }
 126                         }
 127
 128                         //if the user changes the output directory command factory will send this info to us in the output parameter
 129                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 130
 131                         //check for required parameters
 132                         phylipfile = validParameter.validFile(parameters, "phylip", true);
 133                         if (phylipfile == "not open") { abort = true; }
 134                         else if (phylipfile == "not found") { phylipfile = ""; }
 135                         else {  distfile = phylipfile;  format = "phylip";      }
 136
 137                         columnfile = validParameter.validFile(parameters, "column", true);
 138                         if (columnfile == "not open") { abort = true; }
 139                         else if (columnfile == "not found") { columnfile = ""; }
 140                         else {  distfile = columnfile; format = "column";       }
 141
 142                         namefile = validParameter.validFile(parameters, "name", true);
 143                         if (namefile == "not open") { abort = true; }
 144                         else if (namefile == "not found") { namefile = ""; }
 145
 146                         if ((phylipfile == "") && (columnfile == "")) { m->mothurOut("When executing a hcluster command you must enter a phylip or a column."); m->mothurOutEndLine(); abort = true; }
 147                         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; }
 148
 149                         if (columnfile != "") {
 150                                 if (namefile == "") {  cout << "You need to provide a namefile if you are going to use the column format." << endl; abort = true; }
 151                         }
 152
 153                         //check for optional parameter and set defaults
 154                         // ...at some point should added some additional type checking...
 155                         //get user cutoff and precision or use defaults
 156                         string temp;
 157                         temp = validParameter.validFile(parameters, "precision", false);
 158                         if (temp == "not found") { temp = "100"; }
 159                         //saves precision legnth for formatting below
 160                         length = temp.length();
 161                         convert(temp, precision);
 162
 163                         temp = validParameter.validFile(parameters, "hard", false);                     if (temp == "not found") { temp = "F"; }
 164                         hard = m->isTrue(temp);
 165
 166                         temp = validParameter.validFile(parameters, "cutoff", false);
 167                         if (temp == "not found") { temp = "10"; }
 168                         convert(temp, cutoff);
 169                         cutoff += (5 / (precision * 10.0));
 170
 171                         method = validParameter.validFile(parameters, "method", false);
 172                         if (method == "not found") { method = "furthest"; }
 173
 174                         if ((method == "furthest") || (method == "nearest") || (method == "average")) { }
 175                         else { m->mothurOut("Not a valid clustering method.  Valid clustering algorithms are furthest, nearest or average."); m->mothurOutEndLine(); abort = true; }
 176
 177                         showabund = validParameter.validFile(parameters, "showabund", false);
 178                         if (showabund == "not found") { showabund = "T"; }
 179
 180                         sort = validParameter.validFile(parameters, "sorted", false);
 181                         if (sort == "not found") { sort = "F"; }
 182                         sorted = m->isTrue(sort);
 183
 184                         timing = validParameter.validFile(parameters, "timing", false);
 185                         if (timing == "not found") { timing = "F"; }
 186
 187
 188                         if (abort == false) {
 189
 190                                 if (outputDir == "") {  outputDir += m->hasPath(distfile); }
 191                                 fileroot = outputDir + m->getRootName(m->getSimpleName(distfile));
 192
 193                                 if (method == "furthest")               { tag = "fn";  }
 194                                 else if (method == "nearest")   { tag = "nn";  }
 195                                 else                                                    { tag = "an";  }
 196
 197                                 m->openOutputFile(fileroot+ tag + ".sabund",    sabundFile);
 198                                 m->openOutputFile(fileroot+ tag + ".rabund",    rabundFile);
 199                                 m->openOutputFile(fileroot+ tag + ".list",              listFile);
 200
 201                                 outputNames.push_back(fileroot+ tag + ".sabund"); outputTypes["sabund"].push_back(fileroot+ tag + ".sabund");
 202                                 outputNames.push_back(fileroot+ tag + ".rabund"); outputTypes["rabund"].push_back(fileroot+ tag + ".rabund");
 203                                 outputNames.push_back(fileroot+ tag + ".list"); outputTypes["list"].push_back(fileroot+ tag + ".list");
 204                         }
 205                 }
 206         }
 207         catch(exception& e) {
 208                 m->errorOut(e, "HClusterCommand", "HClusterCommand");
 209                 exit(1);
 210         }
 211 }
 212
 213 //**********************************************************************************************************************
 214
 215 void HClusterCommand::help(){
 216         try {
 217                 m->mothurOut("The hcluster command parameter options are cutoff, precision, method, phylip, column, name, showabund, timing and sorted. Phylip or column and name are required.\n");
 218                 m->mothurOut("The phylip and column parameter allow you to enter your distance file, and sorted indicates whether your column distance file is already sorted. \n");
 219                 m->mothurOut("The name parameter allows you to enter your name file and is required if your distance file is in column format. \n");
 220                 m->mothurOut("The hcluster command should be in the following format: \n");
 221                 m->mothurOut("hcluster(column=youDistanceFile, name=yourNameFile, method=yourMethod, cutoff=yourCutoff, precision=yourPrecision) \n");
 222                 m->mothurOut("The acceptable hcluster methods are furthest, nearest and average.\n\n");
 223         }
 224         catch(exception& e) {
 225                 m->errorOut(e, "HClusterCommand", "help");
 226                 exit(1);
 227         }
 228 }
 229
 230 //**********************************************************************************************************************
 231
 232 HClusterCommand::~HClusterCommand(){}
 233
 234 //**********************************************************************************************************************
 235
 236 int HClusterCommand::execute(){
 237         try {
 238
 239                 if (abort == true) {    return 0;       }
 240
 241                 if(namefile != ""){
 242                         globaldata->nameMap = new NameAssignment(namefile);
 243                         globaldata->nameMap->readMap();
 244                 }else{
 245                         globaldata->nameMap = NULL;
 246                 }
 247
 248                 time_t estart = time(NULL);
 249
 250                 if (!sorted) {
 251                         read = new ReadCluster(distfile, cutoff, outputDir, true);
 252                         read->setFormat(format);
 253                         read->read(globaldata->nameMap);
 254
 255                         if (m->control_pressed) {
 256                                 delete read;
 257                                 sabundFile.close();
 258                                 rabundFile.close();
 259                                 listFile.close();
 260                                 for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());  } outputTypes.clear();
 261                                 return 0;
 262                         }
 263
 264                         distfile = read->getOutputFile();
 265
 266                         list = read->getListVector();
 267                         delete read;
 268                 }else {
 269                         list = new ListVector(globaldata->nameMap->getListVector());
 270                 }
 271
 272                 if (m->control_pressed) {
 273                         sabundFile.close();
 274                         rabundFile.close();
 275                         listFile.close();
 276                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());  } outputTypes.clear();
 277                         return 0;
 278                 }
 279
 280                 m->mothurOut("It took " + toString(time(NULL) - estart) + " seconds to sort. "); m->mothurOutEndLine();
 281                 estart = time(NULL);
 282
 283                 //list vector made by read contains all sequence names
 284                 if(list != NULL){
 285                         rabund = new RAbundVector(list->getRAbundVector());
 286                 }else{
 287                         m->mothurOut("Error: no list vector!"); m->mothurOutEndLine(); return 0;
 288                 }
 289
 290                 float previousDist = 0.00000;
 291                 float rndPreviousDist = 0.00000;
 292                 oldRAbund = *rabund;
 293                 oldList = *list;
 294
 295                 print_start = true;
 296                 start = time(NULL);
 297
 298                 cluster = new HCluster(rabund, list, method, distfile, globaldata->nameMap, cutoff);
 299                 vector<seqDist> seqs; seqs.resize(1); // to start loop
 300
 301                 if (m->control_pressed) {
 302                                 delete cluster;
 303                                 sabundFile.close();
 304                                 rabundFile.close();
 305                                 listFile.close();
 306                                 for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());  } outputTypes.clear();
 307                                 return 0;
 308                 }
 309
 310
 311                 while (seqs.size() != 0){
 312
 313                         seqs = cluster->getSeqs();
 314
 315                         if (m->control_pressed) {
 316                                 delete cluster;
 317                                 sabundFile.close();
 318                                 rabundFile.close();
 319                                 listFile.close();
 320                                 for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());  } outputTypes.clear();
 321                                 return 0;
 322                         }
 323
 324                         for (int i = 0; i < seqs.size(); i++) {  //-1 means skip me
 325
 326                                 if (seqs[i].seq1 != seqs[i].seq2) {
 327                                         cluster->update(seqs[i].seq1, seqs[i].seq2, seqs[i].dist);
 328
 329                                         if (m->control_pressed) {
 330                                                 delete cluster;
 331                                                 sabundFile.close();
 332                                                 rabundFile.close();
 333                                                 listFile.close();
 334                                                 for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());  } outputTypes.clear();
 335                                                 return 0;
 336                                         }
 337
 338
 339                                         float rndDist;
 340                                         if (hard) {
 341                                                 rndDist = m->ceilDist(seqs[i].dist, precision);
 342                                         }else{
 343                                                 rndDist = m->roundDist(seqs[i].dist, precision);
 344                                         }
 345
 346
 347                                         if((previousDist <= 0.0000) && (seqs[i].dist != previousDist)){
 348                                                 printData("unique");
 349                                         }
 350                                         else if((rndDist != rndPreviousDist)){
 351                                                 printData(toString(rndPreviousDist,  length-1));
 352                                         }
 353
 354                                         previousDist = seqs[i].dist;
 355                                         rndPreviousDist = rndDist;
 356                                         oldRAbund = *rabund;
 357                                         oldList = *list;
 358                                 }
 359                         }
 360                 }
 361
 362                 if (m->control_pressed) {
 363                         delete cluster;
 364                         sabundFile.close();
 365                         rabundFile.close();
 366                         listFile.close();
 367                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());  } outputTypes.clear();
 368                         return 0;
 369                 }
 370
 371                 if(previousDist <= 0.0000){
 372                         printData("unique");
 373                 }
 374                 else if(rndPreviousDist<cutoff){
 375                         printData(toString(rndPreviousDist, length-1));
 376                 }
 377
 378                 //delete globaldata's copy of the sparsematrix and listvector to free up memory
 379                 delete globaldata->gListVector;  globaldata->gListVector = NULL;
 380
 381                 //saves .list file so you can do the collect, rarefaction and summary commands without doing a read.list
 382                 if (globaldata->getFormat() == "phylip") { globaldata->setPhylipFile(""); }
 383                 else if (globaldata->getFormat() == "column") { globaldata->setColumnFile(""); }
 384
 385                 globaldata->setListFile(fileroot+ tag + ".list");
 386                 globaldata->setNameFile("");
 387                 globaldata->setFormat("list");
 388
 389                 sabundFile.close();
 390                 rabundFile.close();
 391                 listFile.close();
 392                 delete cluster;
 393
 394                 if (m->control_pressed) {
 395                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());  } outputTypes.clear();
 396                         return 0;
 397                 }
 398
 399
 400                 m->mothurOutEndLine();
 401                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 402                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 403                 m->mothurOutEndLine();
 404
 405                 m->mothurOut("It took " + toString(time(NULL) - estart) + " seconds to cluster. "); m->mothurOutEndLine();
 406
 407                 return 0;
 408         }
 409         catch(exception& e) {
 410                 m->errorOut(e, "HClusterCommand", "execute");
 411                 exit(1);
 412         }
 413 }
 414
 415 //**********************************************************************************************************************
 416
 417 void HClusterCommand::printData(string label){
 418         try {
 419                 if (m->isTrue(timing)) {
 420                         m->mothurOut("\tTime: " + toString(time(NULL) - start) + "\tsecs for " + toString(oldRAbund.getNumBins())
 421                      + "\tclusters. Updates: " + toString(loops)); m->mothurOutEndLine();
 422                 }
 423                 print_start = true;
 424                 loops = 0;
 425                 start = time(NULL);
 426
 427                 oldRAbund.setLabel(label);
 428                 if (m->isTrue(showabund)) {
 429                         oldRAbund.getSAbundVector().print(cout);
 430                 }
 431                 oldRAbund.print(rabundFile);
 432                 oldRAbund.getSAbundVector().print(sabundFile);
 433
 434                 oldList.setLabel(label);
 435                 oldList.print(listFile);
 436         }
 437         catch(exception& e) {
 438                 m->errorOut(e, "HClusterCommand", "printData");
 439                 exit(1);
 440         }
 441
 442
 443 }
 444 //**********************************************************************************************************************
 445