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