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 ClusterDoturCommand::ClusterDoturCommand(){
  53         try {
  54                 abort = true; calledHelp = true;
  55                 setParameters();
  56                 vector<string> tempOutNames;
  57                 outputTypes["list"] = tempOutNames;
  58                 outputTypes["rabund"] = tempOutNames;
  59                 outputTypes["sabund"] = tempOutNames;
  60         }
  61         catch(exception& e) {
  62                 m->errorOut(e, "ClusterDoturCommand", "ClusterCommand");
  63                 exit(1);
  64         }
  65 }
  66 //**********************************************************************************************************************
  67 //This function checks to make sure the cluster command has no errors and then clusters based on the method chosen.
  68 ClusterDoturCommand::ClusterDoturCommand(string option)  {
  69         try{
  70
  71                 abort = false; calledHelp = false;
  72
  73                 //allow user to run help
  74                 if(option == "help") { help(); abort = true; calledHelp = true; }
  75                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  76
  77                 else {
  78                         vector<string> myArray = setParameters();
  79
  80                         OptionParser parser(option);
  81                         map<string,string> parameters = parser.getParameters();
  82
  83                         ValidParameters validParameter;
  84
  85                         //check to make sure all parameters are valid for command
  86                         map<string,string>::iterator it;
  87                         for (it = parameters.begin(); it != parameters.end(); it++) {
  88                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {
  89                                         abort = true;
  90                                 }
  91                         }
  92
  93                         //if the user changes the input directory command factory will send this info to us in the output parameter
  94                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  95                         if (inputDir == "not found"){   inputDir = "";          }
  96                         else {
  97                                 string path;
  98                                 it = parameters.find("phylip");
  99                                 //user has given a template file
 100                                 if(it != parameters.end()){
 101                                         path = m->hasPath(it->second);
 102                                         //if the user has not given a path then, add inputdir. else leave path alone.
 103                                         if (path == "") {       parameters["phylip"] = inputDir + it->second;           }
 104                                 }
 105
 106                                 it = parameters.find("name");
 107                                 //user has given a template file
 108                                 if(it != parameters.end()){
 109                                         path = m->hasPath(it->second);
 110                                         //if the user has not given a path then, add inputdir. else leave path alone.
 111                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 112                                 }
 113
 114                         }
 115
 116                         //initialize outputTypes
 117                         vector<string> tempOutNames;
 118                         outputTypes["list"] = tempOutNames;
 119                         outputTypes["rabund"] = tempOutNames;
 120                         outputTypes["sabund"] = tempOutNames;
 121
 122                         //if the user changes the output directory command factory will send this info to us in the output parameter
 123                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 124
 125                         //check for required parameters
 126                         phylipfile = validParameter.validFile(parameters, "phylip", true);
 127                         if (phylipfile == "not open") { abort = true; }
 128                         else if (phylipfile == "not found") {
 129                                 phylipfile = m->getPhylipFile();
 130                                 if (phylipfile != "") {  m->mothurOut("Using " + phylipfile + " as input file for the phylip parameter."); m->mothurOutEndLine(); }
 131                                 else {
 132                                         m->mothurOut("You need to provide a phylip file with the cluster.classic command."); m->mothurOutEndLine();
 133                                         abort = true;
 134                                 }
 135                         }
 136
 137
 138                         //check for optional parameter and set defaults
 139                         namefile = validParameter.validFile(parameters, "name", true);
 140                         if (namefile == "not open") { abort = true; }
 141                         else if (namefile == "not found") { namefile = ""; }
 142
 143                         string temp;
 144                         temp = validParameter.validFile(parameters, "precision", false);
 145                         if (temp == "not found") { temp = "100"; }
 146                         //saves precision legnth for formatting below
 147                         length = temp.length();
 148                         convert(temp, precision);
 149
 150                         temp = validParameter.validFile(parameters, "cutoff", false);
 151                         if (temp == "not found") { temp = "10"; }
 152                         convert(temp, cutoff);
 153                         cutoff += (5 / (precision * 10.0));
 154
 155                         temp = validParameter.validFile(parameters, "hard", false);                     if (temp == "not found") { temp = "T"; }
 156                         hard = m->isTrue(temp);
 157
 158                         temp = validParameter.validFile(parameters, "sim", false);                              if (temp == "not found") { temp = "F"; }
 159                         sim = m->isTrue(temp);
 160
 161                         method = validParameter.validFile(parameters, "method", false);
 162                         if (method == "not found") { method = "average"; }
 163
 164                         if ((method == "furthest") || (method == "nearest") || (method == "average") || (method == "weighted")) {
 165                                 if (method == "furthest") { tag = "fn"; }
 166                                 else if (method == "nearest") { tag = "nn"; }
 167                                 else if (method == "average") { tag = "an"; }
 168                                 else if (method == "weighted") { tag = "wn"; }
 169                         }else { m->mothurOut("Not a valid clustering method.  Valid clustering algorithms are furthest, nearest, average, weighted."); m->mothurOutEndLine(); abort = true; }
 170                 }
 171         }
 172         catch(exception& e) {
 173                 m->errorOut(e, "ClusterDoturCommand", "ClusterCommand");
 174                 exit(1);
 175         }
 176 }
 177 //**********************************************************************************************************************
 178
 179 int ClusterDoturCommand::execute(){
 180         try {
 181
 182                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 183
 184                 if(namefile != ""){
 185                         nameMap = new NameAssignment(namefile);
 186                         nameMap->readMap();
 187                 }else{
 188                         nameMap = NULL;
 189                 }
 190
 191                 //reads phylip file storing data in 2D vector, also fills list and rabund
 192                 ClusterClassic* cluster = new ClusterClassic(cutoff, method, sim);
 193                 cluster->readPhylipFile(phylipfile, nameMap);
 194
 195                 if (m->control_pressed) { delete cluster; delete list; delete rabund; return 0; }
 196
 197                 list = cluster->getListVector();
 198                 rabund = cluster->getRAbundVector();
 199
 200                 if (outputDir == "") { outputDir += m->hasPath(phylipfile); }
 201                 fileroot = outputDir + m->getRootName(m->getSimpleName(phylipfile));
 202
 203                 m->openOutputFile(fileroot+ tag + ".sabund",    sabundFile);
 204                 m->openOutputFile(fileroot+ tag + ".rabund",    rabundFile);
 205                 m->openOutputFile(fileroot+ tag + ".list",              listFile);
 206
 207                 outputNames.push_back(fileroot+ tag + ".sabund"); outputTypes["sabund"].push_back(fileroot+ tag + ".sabund");
 208                 outputNames.push_back(fileroot+ tag + ".rabund"); outputTypes["rabund"].push_back(fileroot+ tag + ".rabund");
 209                 outputNames.push_back(fileroot+ tag + ".list"); outputTypes["list"].push_back(fileroot+ tag + ".list");
 210
 211                 float previousDist = 0.00000;
 212                 float rndPreviousDist = 0.00000;
 213                 oldRAbund = *rabund;
 214                 oldList = *list;
 215
 216                 //double saveCutoff = cutoff;
 217
 218                 int estart = time(NULL);
 219
 220                 while ((cluster->getSmallDist() < cutoff) && (cluster->getNSeqs() > 1)){
 221                         if (m->control_pressed) { delete cluster; delete list; delete rabund; sabundFile.close();rabundFile.close();listFile.close();  for (int i = 0; i < outputNames.size(); i++) {   remove(outputNames[i].c_str());         } outputTypes.clear();  return 0;  }
 222
 223                         cluster->update(cutoff);
 224
 225                         float dist = cluster->getSmallDist();
 226                         float rndDist;
 227                         if (hard) {
 228                                 rndDist = m->ceilDist(dist, precision);
 229                         }else{
 230                                 rndDist = m->roundDist(dist, precision);
 231                         }
 232
 233                         if(previousDist <= 0.0000 && dist != previousDist){
 234                                 printData("unique");
 235                         }
 236                         else if(rndDist != rndPreviousDist){
 237                                 printData(toString(rndPreviousDist,  length-1));
 238                         }
 239
 240                         previousDist = dist;
 241                         rndPreviousDist = rndDist;
 242                         oldRAbund = *rabund;
 243                         oldList = *list;
 244                 }
 245
 246                 if(previousDist <= 0.0000){
 247                         printData("unique");
 248                 }
 249                 else if(rndPreviousDist<cutoff){
 250                         printData(toString(rndPreviousDist, length-1));
 251                 }
 252
 253                 sabundFile.close();
 254                 rabundFile.close();
 255                 listFile.close();
 256
 257                 delete cluster; delete nameMap; delete list; delete rabund;
 258
 259                 //if (saveCutoff != cutoff) {
 260                 //      if (hard)       {  saveCutoff = m->ceilDist(saveCutoff, precision);     }
 261                 //      else            {       saveCutoff = m->roundDist(saveCutoff, precision);  }
 262                 //      m->mothurOut("changed cutoff to " + toString(cutoff)); m->mothurOutEndLine();
 263                 //}
 264
 265                 //set list file as new current listfile
 266                 string current = "";
 267                 itTypes = outputTypes.find("list");
 268                 if (itTypes != outputTypes.end()) {
 269                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 270                 }
 271
 272                 //set rabund file as new current rabundfile
 273                 itTypes = outputTypes.find("rabund");
 274                 if (itTypes != outputTypes.end()) {
 275                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setRabundFile(current); }
 276                 }
 277
 278                 //set sabund file as new current sabundfile
 279                 itTypes = outputTypes.find("sabund");
 280                 if (itTypes != outputTypes.end()) {
 281                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setSabundFile(current); }
 282                 }
 283
 284                 m->mothurOutEndLine();
 285                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 286                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 287                 m->mothurOutEndLine();
 288
 289                 m->mothurOut("It took " + toString(time(NULL) - estart) + " seconds to cluster"); m->mothurOutEndLine();
 290
 291                 return 0;
 292         }
 293         catch(exception& e) {
 294                 m->errorOut(e, "ClusterDoturCommand", "execute");
 295                 exit(1);
 296         }
 297 }
 298
 299 //**********************************************************************************************************************
 300
 301 void ClusterDoturCommand::printData(string label){
 302         try {
 303
 304                 oldRAbund.setLabel(label);
 305                 oldRAbund.print(rabundFile);
 306                 oldRAbund.getSAbundVector().print(sabundFile);
 307
 308                 oldRAbund.getSAbundVector().print(cout);
 309
 310                 oldList.setLabel(label);
 311                 oldList.print(listFile);
 312         }
 313         catch(exception& e) {
 314                 m->errorOut(e, "ClusterDoturCommand", "printData");
 315                 exit(1);
 316         }
 317 }
 318 //**********************************************************************************************************************