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