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