clustercommand.cpp

   1 /*
   2  *  clustercommand.cpp
   3  *  Dotur
   4  *
   5  *  Created by Sarah Westcott on 1/2/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10 #include "clustercommand.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 ClusterCommand::ClusterCommand(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","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
  32                         //check to make sure all parameters are valid for command
  33                         for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
  34                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {
  35                                         abort = true;
  36                                 }
  37                         }
  38
  39                         //if the user changes the output directory command factory will send this info to us in the output parameter
  40                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
  41
  42                         //error checking to make sure they read a distance file
  43                         if ((globaldata->gSparseMatrix == NULL) || (globaldata->gListVector == NULL)) {
  44                                 mothurOut("Before you use the cluster command, you first need to read in a distance matrix."); mothurOutEndLine();
  45                                 abort = true;
  46                         }
  47
  48                         //check for optional parameter and set defaults
  49                         // ...at some point should added some additional type checking...
  50                         //get user cutoff and precision or use defaults
  51                         string temp;
  52                         temp = validParameter.validFile(parameters, "precision", false);
  53                         if (temp == "not found") { temp = "100"; }
  54                         //saves precision legnth for formatting below
  55                         length = temp.length();
  56                         convert(temp, precision);
  57
  58                         temp = validParameter.validFile(parameters, "cutoff", false);
  59                         if (temp == "not found") { temp = "10"; }
  60                         convert(temp, cutoff);
  61                         cutoff += (5 / (precision * 10.0));
  62
  63                         method = validParameter.validFile(parameters, "method", false);
  64                         if (method == "not found") { method = "furthest"; }
  65
  66                         if ((method == "furthest") || (method == "nearest") || (method == "average")) { }
  67                         else { mothurOut("Not a valid clustering method.  Valid clustering algorithms are furthest, nearest or average."); mothurOutEndLine(); abort = true; }
  68
  69                         showabund = validParameter.validFile(parameters, "showabund", false);
  70                         if (showabund == "not found") { showabund = "T"; }
  71
  72                         timing = validParameter.validFile(parameters, "timing", false);
  73                         if (timing == "not found") { timing = "F"; }
  74
  75                         if (abort == false) {
  76
  77
  78                                                         //get matrix, list and rabund for execute
  79                                 if(globaldata->gSparseMatrix != NULL)   {       matrix = globaldata->gSparseMatrix;             }
  80
  81                                 if(globaldata->gListVector != NULL){
  82                                         list = globaldata->gListVector;
  83                                         rabund = new RAbundVector(list->getRAbundVector());
  84                                 }
  85
  86                                 //create cluster
  87                                 if (method == "furthest")       {       cluster = new CompleteLinkage(rabund, list, matrix, cutoff); }
  88                                 else if(method == "nearest"){   cluster = new SingleLinkage(rabund, list, matrix, cutoff); }
  89                                 else if(method == "average"){   cluster = new AverageLinkage(rabund, list, matrix, cutoff);     }
  90                                 tag = cluster->getTag();
  91
  92                                 if (outputDir == "") { outputDir += hasPath(globaldata->inputFileName); }
  93                                 fileroot = outputDir + getRootName(getSimpleName(globaldata->inputFileName));
  94
  95                                 openOutputFile(fileroot+ tag + ".sabund",       sabundFile);
  96                                 openOutputFile(fileroot+ tag + ".rabund",       rabundFile);
  97                                 openOutputFile(fileroot+ tag + ".list",         listFile);
  98                         }
  99                 }
 100         }
 101         catch(exception& e) {
 102                 errorOut(e, "ClusterCommand", "ClusterCommand");
 103                 exit(1);
 104         }
 105 }
 106
 107 //**********************************************************************************************************************
 108
 109 void ClusterCommand::help(){
 110         try {
 111                 mothurOut("The cluster command can only be executed after a successful read.dist command.\n");
 112                 mothurOut("The cluster command parameter options are method, cuttoff, precision, showabund and timing. No parameters are required.\n");
 113                 mothurOut("The cluster command should be in the following format: \n");
 114                 mothurOut("cluster(method=yourMethod, cutoff=yourCutoff, precision=yourPrecision) \n");
 115                 mothurOut("The acceptable cluster methods are furthest, nearest and average.  If no method is provided then furthest is assumed.\n\n");
 116         }
 117         catch(exception& e) {
 118                 errorOut(e, "ClusterCommand", "help");
 119                 exit(1);
 120         }
 121 }
 122
 123 //**********************************************************************************************************************
 124
 125 ClusterCommand::~ClusterCommand(){
 126         if (abort == false) {
 127                 delete cluster;
 128                 delete rabund;
 129         }
 130 }
 131
 132 //**********************************************************************************************************************
 133
 134 int ClusterCommand::execute(){
 135         try {
 136
 137                 if (abort == true) {    return 0;       }
 138
 139                 time_t estart = time(NULL);
 140                 //int ndist = matrix->getNNodes();
 141                 float previousDist = 0.00000;
 142                 float rndPreviousDist = 0.00000;
 143                 oldRAbund = *rabund;
 144                 oldList = *list;
 145
 146                 print_start = true;
 147                 start = time(NULL);
 148                 loops = 0;
 149
 150                 while (matrix->getSmallDist() < cutoff && matrix->getNNodes() > 0){
 151                         if (print_start && isTrue(timing)) {
 152                                 mothurOut("Clustering (" + tag + ") dist " + toString(matrix->getSmallDist()) + "/"
 153                                         + toString(roundDist(matrix->getSmallDist(), precision))
 154                                         + "\t(precision: " + toString(precision) + ", Nodes: " + toString(matrix->getNNodes()) + ")");
 155                                 cout.flush();
 156                                 print_start = false;
 157                         }
 158
 159                         loops++;
 160
 161                         cluster->update(cutoff);
 162                         float dist = matrix->getSmallDist();
 163                         float rndDist = roundDist(dist, precision);
 164
 165                         if(previousDist <= 0.0000 && dist != previousDist){
 166                                 printData("unique");
 167                         }
 168                         else if(rndDist != rndPreviousDist){
 169                                 printData(toString(rndPreviousDist,  length-1));
 170                         }
 171
 172                         previousDist = dist;
 173                         rndPreviousDist = rndDist;
 174                         oldRAbund = *rabund;
 175                         oldList = *list;
 176                 }
 177
 178                 if (print_start && isTrue(timing)) {
 179                         mothurOut("Clustering (" + tag + ") for distance " + toString(previousDist) + "/" + toString(rndPreviousDist)
 180                                          + "\t(precision: " + toString(precision) + ", Nodes: " + toString(matrix->getNNodes()) + ")");
 181                         cout.flush();
 182                         print_start = false;
 183                 }
 184
 185                 if(previousDist <= 0.0000){
 186                         printData("unique");
 187                 }
 188                 else if(rndPreviousDist<cutoff){
 189                         printData(toString(rndPreviousDist, length-1));
 190                 }
 191
 192                 //delete globaldata's copy of the sparsematrix and listvector to free up memory
 193                 delete globaldata->gSparseMatrix;  globaldata->gSparseMatrix = NULL;
 194                 delete globaldata->gListVector;  globaldata->gListVector = NULL;
 195
 196                 //saves .list file so you can do the collect, rarefaction and summary commands without doing a read.list
 197                 if (globaldata->getFormat() == "phylip") { globaldata->setPhylipFile(""); }
 198                 else if (globaldata->getFormat() == "column") { globaldata->setColumnFile(""); }
 199
 200                 globaldata->setListFile(fileroot+ tag + ".list");
 201                 globaldata->setNameFile("");
 202                 globaldata->setFormat("list");
 203
 204                 sabundFile.close();
 205                 rabundFile.close();
 206                 listFile.close();
 207                 //if (isTrue(timing)) {
 208                         mothurOut("It took " + toString(time(NULL) - estart) + " seconds to cluster"); mothurOutEndLine();
 209                 //}
 210                 return 0;
 211         }
 212         catch(exception& e) {
 213                 errorOut(e, "ClusterCommand", "execute");
 214                 exit(1);
 215         }
 216 }
 217
 218 //**********************************************************************************************************************
 219
 220 void ClusterCommand::printData(string label){
 221         try {
 222                 if (isTrue(timing)) {
 223                         mothurOut("\tTime: " + toString(time(NULL) - start) + "\tsecs for " + toString(oldRAbund.getNumBins())
 224                      + "\tclusters. Updates: " + toString(loops)); mothurOutEndLine();
 225                 }
 226                 print_start = true;
 227                 loops = 0;
 228                 start = time(NULL);
 229
 230                 oldRAbund.setLabel(label);
 231                 if (isTrue(showabund)) {
 232                         oldRAbund.getSAbundVector().print(cout);
 233                 }
 234                 oldRAbund.print(rabundFile);
 235                 oldRAbund.getSAbundVector().print(sabundFile);
 236
 237                 oldList.setLabel(label);
 238                 oldList.print(listFile);
 239         }
 240         catch(exception& e) {
 241                 errorOut(e, "ClusterCommand", "printData");
 242                 exit(1);
 243         }
 244
 245
 246 }
 247 //**********************************************************************************************************************