clusterdoturcommand.cpp

   1 /*
   2  *  clusterdoturcommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 10/27/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "clusterdoturcommand.h"
  11 #include "clusterclassic.h"
  12
  13 //**********************************************************************************************************************
  14 vector<string> ClusterDoturCommand::setParameters(){
  15         try {
  16                 CommandParameter pphylip("phylip", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pphylip);
  17                 CommandParameter pname("name", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pname);
  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", "average", "", "", "",false,false); parameters.push_back(pmethod);
  21                 CommandParameter phard("hard", "Boolean", "", "T", "", "", "",false,false); parameters.push_back(phard);
  22                 CommandParameter psim("sim", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(psim);
  23                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  24                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  25
  26                 vector<string> myArray;
  27                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  28                 return myArray;
  29         }
  30         catch(exception& e) {
  31                 m->errorOut(e, "ClusterDoturCommand", "setParameters");
  32                 exit(1);
  33         }
  34 }
  35 //**********************************************************************************************************************
  36 string ClusterDoturCommand::getHelpString(){
  37         try {
  38                 string helpString = "";
  39                 helpString += "The cluster.classic command clusters using the algorithm from dotur. \n";
  40                 helpString += "The cluster.classic command parameter options are phylip, name, method, cuttoff, hard, sim, precision. Phylip is required, unless you have a valid current file.\n";
  41                 helpString += "The cluster.classic command should be in the following format: \n";
  42                 helpString += "cluster.classic(phylip=yourDistanceMatrix, method=yourMethod, cutoff=yourCutoff, precision=yourPrecision) \n";
  43                 helpString += "The acceptable cluster methods are furthest, nearest, weighted and average.  If no method is provided then average is assumed.\n";
  44                 return helpString;
  45         }
  46         catch(exception& e) {
  47                 m->errorOut(e, "ClusterDoturCommand", "getHelpString");
  48                 exit(1);
  49         }
  50 }
  51 //**********************************************************************************************************************
  52 string ClusterDoturCommand::getOutputFileNameTag(string type, string inputName=""){
  53         try {
  54         string outputFileName = "";
  55                 map<string, vector<string> >::iterator it;
  56
  57         //is this a type this command creates
  58         it = outputTypes.find(type);
  59         if (it == outputTypes.end()) {  m->mothurOut("[ERROR]: this command doesn't create a " + type + " output file.\n"); }
  60         else {
  61             if (type == "list") {  outputFileName =  "list"; }
  62             else if (type == "rabund") {  outputFileName =  "rabund"; }
  63             else if (type == "sabund") {  outputFileName =  "sabund"; }
  64             else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true;  }
  65         }
  66         return outputFileName;
  67         }
  68         catch(exception& e) {
  69                 m->errorOut(e, "ClusterDoturCommand", "getOutputFileNameTag");
  70                 exit(1);
  71         }
  72 }
  73 //**********************************************************************************************************************
  74 ClusterDoturCommand::ClusterDoturCommand(){
  75         try {
  76                 abort = true; calledHelp = true;
  77                 setParameters();
  78                 vector<string> tempOutNames;
  79                 outputTypes["list"] = tempOutNames;
  80                 outputTypes["rabund"] = tempOutNames;
  81                 outputTypes["sabund"] = tempOutNames;
  82         }
  83         catch(exception& e) {
  84                 m->errorOut(e, "ClusterDoturCommand", "ClusterCommand");
  85                 exit(1);
  86         }
  87 }
  88 //**********************************************************************************************************************
  89 //This function checks to make sure the cluster command has no errors and then clusters based on the method chosen.
  90 ClusterDoturCommand::ClusterDoturCommand(string option)  {
  91         try{
  92
  93                 abort = false; calledHelp = false;
  94
  95                 //allow user to run help
  96                 if(option == "help") { help(); abort = true; calledHelp = true; }
  97                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  98
  99                 else {
 100                         vector<string> myArray = setParameters();
 101
 102                         OptionParser parser(option);
 103                         map<string,string> parameters = parser.getParameters();
 104
 105                         ValidParameters validParameter;
 106
 107                         //check to make sure all parameters are valid for command
 108                         map<string,string>::iterator it;
 109                         for (it = parameters.begin(); it != parameters.end(); it++) {
 110                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {
 111                                         abort = true;
 112                                 }
 113                         }
 114
 115                         //if the user changes the input directory command factory will send this info to us in the output parameter
 116                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 117                         if (inputDir == "not found"){   inputDir = "";          }
 118                         else {
 119                                 string path;
 120                                 it = parameters.find("phylip");
 121                                 //user has given a template file
 122                                 if(it != parameters.end()){
 123                                         path = m->hasPath(it->second);
 124                                         //if the user has not given a path then, add inputdir. else leave path alone.
 125                                         if (path == "") {       parameters["phylip"] = inputDir + it->second;           }
 126                                 }
 127
 128                                 it = parameters.find("name");
 129                                 //user has given a template file
 130                                 if(it != parameters.end()){
 131                                         path = m->hasPath(it->second);
 132                                         //if the user has not given a path then, add inputdir. else leave path alone.
 133                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 134                                 }
 135
 136                         }
 137
 138                         //initialize outputTypes
 139                         vector<string> tempOutNames;
 140                         outputTypes["list"] = tempOutNames;
 141                         outputTypes["rabund"] = tempOutNames;
 142                         outputTypes["sabund"] = tempOutNames;
 143
 144                         //if the user changes the output directory command factory will send this info to us in the output parameter
 145                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 146
 147                         //check for required parameters
 148                         phylipfile = validParameter.validFile(parameters, "phylip", true);
 149                         if (phylipfile == "not open") { abort = true; }
 150                         else if (phylipfile == "not found") {
 151                                 phylipfile = m->getPhylipFile();
 152                                 if (phylipfile != "") {  m->mothurOut("Using " + phylipfile + " as input file for the phylip parameter."); m->mothurOutEndLine(); }
 153                                 else {
 154                                         m->mothurOut("You need to provide a phylip file with the cluster.classic command."); m->mothurOutEndLine();
 155                                         abort = true;
 156                                 }
 157                         }else { m->setPhylipFile(phylipfile); }
 158
 159
 160                         //check for optional parameter and set defaults
 161                         namefile = validParameter.validFile(parameters, "name", true);
 162                         if (namefile == "not open") { abort = true; }
 163                         else if (namefile == "not found") { namefile = ""; }
 164                         else { m->setNameFile(namefile); }
 165
 166                         string temp;
 167                         temp = validParameter.validFile(parameters, "precision", false);
 168                         if (temp == "not found") { temp = "100"; }
 169                         //saves precision legnth for formatting below
 170                         length = temp.length();
 171                         m->mothurConvert(temp, precision);
 172
 173                         temp = validParameter.validFile(parameters, "cutoff", false);
 174                         if (temp == "not found") { temp = "10"; }
 175                         m->mothurConvert(temp, cutoff);
 176                         cutoff += (5 / (precision * 10.0));
 177
 178                         temp = validParameter.validFile(parameters, "hard", false);                     if (temp == "not found") { temp = "T"; }
 179                         hard = m->isTrue(temp);
 180
 181                         temp = validParameter.validFile(parameters, "sim", false);                              if (temp == "not found") { temp = "F"; }
 182                         sim = m->isTrue(temp);
 183
 184                         method = validParameter.validFile(parameters, "method", false);
 185                         if (method == "not found") { method = "average"; }
 186
 187                         if ((method == "furthest") || (method == "nearest") || (method == "average") || (method == "weighted")) {
 188                                 if (method == "furthest") { tag = "fn"; }
 189                                 else if (method == "nearest") { tag = "nn"; }
 190                                 else if (method == "average") { tag = "an"; }
 191                                 else if (method == "weighted") { tag = "wn"; }
 192                         }else { m->mothurOut("Not a valid clustering method.  Valid clustering algorithms are furthest, nearest, average, weighted."); m->mothurOutEndLine(); abort = true; }
 193                 }
 194         }
 195         catch(exception& e) {
 196                 m->errorOut(e, "ClusterDoturCommand", "ClusterCommand");
 197                 exit(1);
 198         }
 199 }
 200 //**********************************************************************************************************************
 201
 202 int ClusterDoturCommand::execute(){
 203         try {
 204
 205                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 206
 207                 if(namefile != ""){
 208                         nameMap = new NameAssignment(namefile);
 209                         nameMap->readMap();
 210                 }else{
 211                         nameMap = NULL;
 212                 }
 213
 214                 //reads phylip file storing data in 2D vector, also fills list and rabund
 215                 ClusterClassic* cluster = new ClusterClassic(cutoff, method, sim);
 216                 cluster->readPhylipFile(phylipfile, nameMap);
 217
 218                 if (m->control_pressed) { delete cluster; delete list; delete rabund; return 0; }
 219
 220                 list = cluster->getListVector();
 221                 rabund = cluster->getRAbundVector();
 222
 223                 if (outputDir == "") { outputDir += m->hasPath(phylipfile); }
 224                 fileroot = outputDir + m->getRootName(m->getSimpleName(phylipfile));
 225
 226         string sabundFileName = fileroot+ tag + "." + getOutputFileNameTag("sabund");
 227         string rabundFileName = fileroot+ tag + "." + getOutputFileNameTag("rabund");
 228         string listFileName = fileroot+ tag + "." + getOutputFileNameTag("list");
 229
 230                 m->openOutputFile(sabundFileName,       sabundFile);
 231                 m->openOutputFile(rabundFileName,       rabundFile);
 232                 m->openOutputFile(listFileName, listFile);
 233
 234                 outputNames.push_back(sabundFileName); outputTypes["sabund"].push_back(sabundFileName);
 235                 outputNames.push_back(rabundFileName); outputTypes["rabund"].push_back(rabundFileName);
 236                 outputNames.push_back(listFileName); outputTypes["list"].push_back(listFileName);
 237
 238                 float previousDist = 0.00000;
 239                 float rndPreviousDist = 0.00000;
 240                 oldRAbund = *rabund;
 241                 oldList = *list;
 242
 243                 //double saveCutoff = cutoff;
 244
 245                 int estart = time(NULL);
 246
 247                 while ((cluster->getSmallDist() < cutoff) && (cluster->getNSeqs() > 1)){
 248                         if (m->control_pressed) { delete cluster; delete list; delete rabund; sabundFile.close();rabundFile.close();listFile.close();  for (int i = 0; i < outputNames.size(); i++) {   m->mothurRemove(outputNames[i]);        } outputTypes.clear();  return 0;  }
 249
 250                         cluster->update(cutoff);
 251
 252                         float dist = cluster->getSmallDist();
 253                         float rndDist;
 254                         if (hard) {
 255                                 rndDist = m->ceilDist(dist, precision);
 256                         }else{
 257                                 rndDist = m->roundDist(dist, precision);
 258                         }
 259
 260                         if(previousDist <= 0.0000 && dist != previousDist){
 261                                 printData("unique");
 262                         }
 263                         else if(rndDist != rndPreviousDist){
 264                                 printData(toString(rndPreviousDist,  length-1));
 265                         }
 266
 267                         previousDist = dist;
 268                         rndPreviousDist = rndDist;
 269                         oldRAbund = *rabund;
 270                         oldList = *list;
 271                 }
 272
 273                 if(previousDist <= 0.0000){
 274                         printData("unique");
 275                 }
 276                 else if(rndPreviousDist<cutoff){
 277                         printData(toString(rndPreviousDist, length-1));
 278                 }
 279
 280                 sabundFile.close();
 281                 rabundFile.close();
 282                 listFile.close();
 283
 284                 delete cluster; delete nameMap; delete list; delete rabund;
 285
 286                 //if (saveCutoff != cutoff) {
 287                 //      if (hard)       {  saveCutoff = m->ceilDist(saveCutoff, precision);     }
 288                 //      else            {       saveCutoff = m->roundDist(saveCutoff, precision);  }
 289                 //      m->mothurOut("changed cutoff to " + toString(cutoff)); m->mothurOutEndLine();
 290                 //}
 291
 292                 //set list file as new current listfile
 293                 string current = "";
 294                 itTypes = outputTypes.find("list");
 295                 if (itTypes != outputTypes.end()) {
 296                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 297                 }
 298
 299                 //set rabund file as new current rabundfile
 300                 itTypes = outputTypes.find("rabund");
 301                 if (itTypes != outputTypes.end()) {
 302                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setRabundFile(current); }
 303                 }
 304
 305                 //set sabund file as new current sabundfile
 306                 itTypes = outputTypes.find("sabund");
 307                 if (itTypes != outputTypes.end()) {
 308                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setSabundFile(current); }
 309                 }
 310
 311                 m->mothurOutEndLine();
 312                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 313                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 314                 m->mothurOutEndLine();
 315
 316                 m->mothurOut("It took " + toString(time(NULL) - estart) + " seconds to cluster"); m->mothurOutEndLine();
 317
 318                 return 0;
 319         }
 320         catch(exception& e) {
 321                 m->errorOut(e, "ClusterDoturCommand", "execute");
 322                 exit(1);
 323         }
 324 }
 325
 326 //**********************************************************************************************************************
 327
 328 void ClusterDoturCommand::printData(string label){
 329         try {
 330
 331                 oldRAbund.setLabel(label);
 332                 oldRAbund.print(rabundFile);
 333                 oldRAbund.getSAbundVector().print(sabundFile);
 334
 335                 oldRAbund.getSAbundVector().print(cout);
 336
 337                 oldList.setLabel(label);
 338                 oldList.print(listFile);
 339         }
 340         catch(exception& e) {
 341                 m->errorOut(e, "ClusterDoturCommand", "printData");
 342                 exit(1);
 343         }
 344 }
 345 //**********************************************************************************************************************