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::getValidParameters(){
  15         try {
  16                 string AlignArray[] =  {"phylip","name","hard","cutoff","precision","method","outputdir","inputdir"};
  17                 vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
  18                 return myArray;
  19         }
  20         catch(exception& e) {
  21                 m->errorOut(e, "ClusterDoturCommand", "getValidParameters");
  22                 exit(1);
  23         }
  24 }
  25 //**********************************************************************************************************************
  26 ClusterDoturCommand::ClusterDoturCommand(){
  27         try {
  28                 abort = true; calledHelp = true;
  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, "ClusterDoturCommand", "ClusterCommand");
  36                 exit(1);
  37         }
  38 }
  39 //**********************************************************************************************************************
  40 vector<string> ClusterDoturCommand::getRequiredParameters(){
  41         try {
  42                 string Array[] =  {"phylip"};
  43                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  44                 return myArray;
  45         }
  46         catch(exception& e) {
  47                 m->errorOut(e, "ClusterDoturCommand", "getRequiredParameters");
  48                 exit(1);
  49         }
  50 }
  51 //**********************************************************************************************************************
  52 vector<string> ClusterDoturCommand::getRequiredFiles(){
  53         try {
  54                 vector<string> myArray;
  55                 return myArray;
  56         }
  57         catch(exception& e) {
  58                 m->errorOut(e, "ClusterDoturCommand", "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 ClusterDoturCommand::ClusterDoturCommand(string option)  {
  65         try{
  66
  67                 abort = false; calledHelp = false;
  68
  69                 //allow user to run help
  70                 if(option == "help") { help(); abort = true; calledHelp = true; }
  71
  72                 else {
  73                         //valid paramters for this command
  74                         string Array[] =  {"phylip","name","cutoff","hard","precision","method","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
  82                         //check to make sure all parameters are valid for command
  83                         map<string,string>::iterator it;
  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                         //if the user changes the input directory command factory will send this info to us in the output parameter
  91                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  92                         if (inputDir == "not found"){   inputDir = "";          }
  93                         else {
  94                                 string path;
  95                                 it = parameters.find("phylip");
  96                                 //user has given a template file
  97                                 if(it != parameters.end()){
  98                                         path = m->hasPath(it->second);
  99                                         //if the user has not given a path then, add inputdir. else leave path alone.
 100                                         if (path == "") {       parameters["phylip"] = inputDir + it->second;           }
 101                                 }
 102
 103                                 it = parameters.find("name");
 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["name"] = inputDir + it->second;             }
 109                                 }
 110
 111                         }
 112
 113                         //initialize outputTypes
 114                         vector<string> tempOutNames;
 115                         outputTypes["list"] = tempOutNames;
 116                         outputTypes["rabund"] = tempOutNames;
 117                         outputTypes["sabund"] = tempOutNames;
 118
 119                         //if the user changes the output directory command factory will send this info to us in the output parameter
 120                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 121
 122                         //check for required parameters
 123                         phylipfile = validParameter.validFile(parameters, "phylip", true);
 124                         if (phylipfile == "not open") { abort = true; }
 125                         else if (phylipfile == "not found") { phylipfile = ""; m->mothurOut("When executing the cluster.dotur command you must enter a phylip file."); m->mothurOutEndLine(); abort = true; }
 126
 127
 128                         //check for optional parameter and set defaults
 129                         namefile = validParameter.validFile(parameters, "name", true);
 130                         if (namefile == "not open") { abort = true; }
 131                         else if (namefile == "not found") { namefile = ""; }
 132
 133                         string temp;
 134                         temp = validParameter.validFile(parameters, "precision", false);
 135                         if (temp == "not found") { temp = "100"; }
 136                         //saves precision legnth for formatting below
 137                         length = temp.length();
 138                         convert(temp, precision);
 139
 140                         temp = validParameter.validFile(parameters, "cutoff", false);
 141                         if (temp == "not found") { temp = "10"; }
 142                         convert(temp, cutoff);
 143                         cutoff += (5 / (precision * 10.0));
 144
 145                         temp = validParameter.validFile(parameters, "hard", false);                     if (temp == "not found") { temp = "F"; }
 146                         hard = m->isTrue(temp);
 147
 148                         method = validParameter.validFile(parameters, "method", false);
 149                         if (method == "not found") { method = "furthest"; }
 150
 151                         if ((method == "furthest") || (method == "nearest") || (method == "average") || (method == "weighted")) {
 152                                 if (method == "furthest") { tag = "fn"; }
 153                                 else if (method == "nearest") { tag = "nn"; }
 154                                 else if (method == "average") { tag = "an"; }
 155                                 else if (method == "weighted") { tag = "wn"; }
 156                         }else { m->mothurOut("Not a valid clustering method.  Valid clustering algorithms are furthest, nearest, average, weighted."); m->mothurOutEndLine(); abort = true; }
 157                 }
 158         }
 159         catch(exception& e) {
 160                 m->errorOut(e, "ClusterDoturCommand", "ClusterCommand");
 161                 exit(1);
 162         }
 163 }
 164
 165 //**********************************************************************************************************************
 166
 167 void ClusterDoturCommand::help(){
 168         try {
 169                 m->mothurOut("The cluster.classic command clusters using the algorithm from dotur. \n");
 170                 m->mothurOut("The cluster.classic command parameter options are phylip, name, method, cuttoff, hard, precision. Phylip is required.\n");
 171                 m->mothurOut("The cluster.classic command should be in the following format: \n");
 172                 m->mothurOut("cluster.classic(phylip=yourDistanceMatrix, method=yourMethod, cutoff=yourCutoff, precision=yourPrecision) \n");
 173                 m->mothurOut("The acceptable cluster methods are furthest, nearest, weighted and average.  If no method is provided then furthest is assumed.\n\n");
 174
 175         }
 176         catch(exception& e) {
 177                 m->errorOut(e, "ClusterDoturCommand", "help");
 178                 exit(1);
 179         }
 180 }
 181
 182 //**********************************************************************************************************************
 183
 184 ClusterDoturCommand::~ClusterDoturCommand(){}
 185
 186 //**********************************************************************************************************************
 187
 188 int ClusterDoturCommand::execute(){
 189         try {
 190
 191                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 192
 193                 if(namefile != ""){
 194                         nameMap = new NameAssignment(namefile);
 195                         nameMap->readMap();
 196                 }else{
 197                         nameMap = NULL;
 198                 }
 199
 200                 //reads phylip file storing data in 2D vector, also fills list and rabund
 201                 ClusterClassic* cluster = new ClusterClassic(cutoff, method);
 202                 cluster->readPhylipFile(phylipfile, nameMap);
 203
 204                 if (m->control_pressed) { delete cluster; delete list; delete rabund; return 0; }
 205
 206                 list = cluster->getListVector();
 207                 rabund = cluster->getRAbundVector();
 208
 209                 if (outputDir == "") { outputDir += m->hasPath(phylipfile); }
 210                 fileroot = outputDir + m->getRootName(m->getSimpleName(phylipfile));
 211
 212                 m->openOutputFile(fileroot+ tag + ".sabund",    sabundFile);
 213                 m->openOutputFile(fileroot+ tag + ".rabund",    rabundFile);
 214                 m->openOutputFile(fileroot+ tag + ".list",              listFile);
 215
 216                 outputNames.push_back(fileroot+ tag + ".sabund"); outputTypes["sabund"].push_back(fileroot+ tag + ".sabund");
 217                 outputNames.push_back(fileroot+ tag + ".rabund"); outputTypes["rabund"].push_back(fileroot+ tag + ".rabund");
 218                 outputNames.push_back(fileroot+ tag + ".list"); outputTypes["list"].push_back(fileroot+ tag + ".list");
 219
 220                 float previousDist = 0.00000;
 221                 float rndPreviousDist = 0.00000;
 222                 oldRAbund = *rabund;
 223                 oldList = *list;
 224
 225                 //double saveCutoff = cutoff;
 226
 227                 int estart = time(NULL);
 228
 229                 while ((cluster->getSmallDist() < cutoff) && (cluster->getNSeqs() > 1)){
 230                         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;  }
 231
 232                         cluster->update(cutoff);
 233
 234                         float dist = cluster->getSmallDist();
 235                         float rndDist;
 236                         if (hard) {
 237                                 rndDist = m->ceilDist(dist, precision);
 238                         }else{
 239                                 rndDist = m->roundDist(dist, precision);
 240                         }
 241
 242                         if(previousDist <= 0.0000 && dist != previousDist){
 243                                 printData("unique");
 244                         }
 245                         else if(rndDist != rndPreviousDist){
 246                                 printData(toString(rndPreviousDist,  length-1));
 247                         }
 248
 249                         previousDist = dist;
 250                         rndPreviousDist = rndDist;
 251                         oldRAbund = *rabund;
 252                         oldList = *list;
 253                 }
 254
 255                 if(previousDist <= 0.0000){
 256                         printData("unique");
 257                 }
 258                 else if(rndPreviousDist<cutoff){
 259                         printData(toString(rndPreviousDist, length-1));
 260                 }
 261
 262                 sabundFile.close();
 263                 rabundFile.close();
 264                 listFile.close();
 265
 266                 delete cluster; delete nameMap; delete list; delete rabund;
 267
 268                 //if (saveCutoff != cutoff) {
 269                 //      if (hard)       {  saveCutoff = m->ceilDist(saveCutoff, precision);     }
 270                 //      else            {       saveCutoff = m->roundDist(saveCutoff, precision);  }
 271                 //      m->mothurOut("changed cutoff to " + toString(cutoff)); m->mothurOutEndLine();
 272                 //}
 273
 274                 m->mothurOutEndLine();
 275                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 276                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 277                 m->mothurOutEndLine();
 278
 279                 m->mothurOut("It took " + toString(time(NULL) - estart) + " seconds to cluster"); m->mothurOutEndLine();
 280
 281                 return 0;
 282         }
 283         catch(exception& e) {
 284                 m->errorOut(e, "ClusterDoturCommand", "execute");
 285                 exit(1);
 286         }
 287 }
 288
 289 //**********************************************************************************************************************
 290
 291 void ClusterDoturCommand::printData(string label){
 292         try {
 293
 294                 oldRAbund.setLabel(label);
 295                 oldRAbund.print(rabundFile);
 296                 oldRAbund.getSAbundVector().print(sabundFile);
 297
 298                 oldRAbund.getSAbundVector().print(cout);
 299
 300                 oldList.setLabel(label);
 301                 oldList.print(listFile);
 302         }
 303         catch(exception& e) {
 304                 m->errorOut(e, "ClusterDoturCommand", "printData");
 305                 exit(1);
 306         }
 307 }
 308 //**********************************************************************************************************************