kruskalwalliscommand.cpp

   1 /*
   2  * File:   kruskalwalliscommand.cpp
   3  * Author: kiverson
   4  *
   5  * Created on June 26, 2012, 11:06 AM
   6  */
   7
   8 #include "kruskalwalliscommand.h"
   9
  10 //**********************************************************************************************************************
  11 vector<string> KruskalWallisCommand::setParameters(){
  12         try {
  13                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(pinputdir);
  14                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(poutputdir);
  15         CommandParameter pgroups("groups", "String", "", "", "", "", "","",false,false,true); parameters.push_back(pgroups);
  16         CommandParameter pshared("shared", "InputTypes", "", "", "none", "none", "none","summary",false,true,true); parameters.push_back(pshared);
  17
  18                 vector<string> myArray;
  19                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  20                 return myArray;
  21         }
  22         catch(exception& e) {
  23                 m->errorOut(e, "KruskalWallisCommand", "setParameters");
  24                 exit(1);
  25         }
  26 }
  27 //**********************************************************************************************************************
  28 string KruskalWallisCommand::getHelpString(){
  29         try {
  30                 string helpString = "";
  31                 helpString += "The kruskalwallis command parameter options are \n";
  32         helpString += "Kruskal–Wallis one-way analysis of variance is a non-parametric method for testing whether samples originate from the same distribution.";
  33                 return helpString;
  34         }
  35         catch(exception& e) {
  36                 m->errorOut(e, "KruskalWallisCommand", "getHelpString");
  37                 exit(1);
  38         }
  39 }
  40 //**********************************************************************************************************************
  41 string KruskalWallisCommand::getOutputPattern(string type) {
  42     try {
  43         string pattern = "";
  44
  45         if (type == "summary") {  pattern = "[filename],cooccurence.summary"; }
  46         else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true;  }
  47
  48         return pattern;
  49     }
  50     catch(exception& e) {
  51         m->errorOut(e, "KruskalWallisCommand", "getOutputPattern");
  52         exit(1);
  53     }
  54 }
  55 //**********************************************************************************************************************
  56 KruskalWallisCommand::KruskalWallisCommand(){
  57         try {
  58                 abort = true; calledHelp = true;
  59                 setParameters();
  60         vector<string> tempOutNames;
  61                 outputTypes["summary"] = tempOutNames;
  62
  63         }
  64         catch(exception& e) {
  65                 m->errorOut(e, "KruskalWallisCommand", "KruskalWallisCommand");
  66                 exit(1);
  67         }
  68 }
  69 //**********************************************************************************************************************
  70 KruskalWallisCommand::KruskalWallisCommand(string option) {
  71         try {
  72                 abort = false; calledHelp = false;
  73
  74                 //allow user to run help
  75                 if(option == "help") { help(); abort = true; calledHelp = true; }
  76                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  77
  78                 else {
  79                         vector<string> myArray = setParameters();
  80
  81                         OptionParser parser(option);
  82                         map<string,string> parameters = parser.getParameters();
  83                         map<string,string>::iterator it;
  84
  85                         ValidParameters validParameter;
  86
  87                         //check to make sure all parameters are valid for command
  88                         for (it = parameters.begin(); it != parameters.end(); it++) {
  89                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  90                         }
  91
  92             //get shared file
  93                         sharedfile = validParameter.validFile(parameters, "shared", true);
  94                         if (sharedfile == "not open") { sharedfile = ""; abort = true; }
  95                         else if (sharedfile == "not found") {
  96                                 //if there is a current shared file, use it
  97                                 sharedfile = m->getSharedFile();
  98                                 if (sharedfile != "") { m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); }
  99                                 else {  m->mothurOut("You have no current sharedfile and the shared parameter is required."); m->mothurOutEndLine(); abort = true; }
 100                         }else { m->setSharedFile(sharedfile); }
 101
 102             //if the user changes the output directory command factory will send this info to us in the output parameter
 103                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = m->hasPath(sharedfile);             }
 104
 105             groups = validParameter.validFile(parameters, "groups", false);
 106             if (groups == "not found") { groups = "";   }
 107             else {
 108             m->splitAtDash(groups, Groups);
 109             }
 110             m->setGroups(Groups);
 111
 112                         //if the user changes the input directory command factory will send this info to us in the output parameter
 113                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 114                         if (inputDir == "not found"){   inputDir = "";          }
 115                         else {
 116                                 string path;
 117                                 it = parameters.find("shared");
 118                                 //user has given a template file
 119                                 if(it != parameters.end()){
 120                                         path = m->hasPath(it->second);
 121                                         //if the user has not given a path then, add inputdir. else leave path alone.
 122                                         if (path == "") {       parameters["shared"] = inputDir + it->second;           }
 123                                 }
 124                         }
 125
 126             vector<string> tempOutNames;
 127             outputTypes["summary"] = tempOutNames;
 128
 129
 130                 }
 131
 132         }
 133         catch(exception& e) {
 134                 m->errorOut(e, "KruskalWallisCommand", "KruskalWallisCommand");
 135                 exit(1);
 136         }
 137 }
 138 //**********************************************************************************************************************
 139 int KruskalWallisCommand::execute(){
 140         try {
 141                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 142
 143         InputData* input = new InputData(sharedfile, "sharedfile");
 144         vector<SharedRAbundVector*> lookup = input->getSharedRAbundVectors();
 145                 string lastLabel = lookup[0]->getLabel();
 146
 147
 148                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 149                 set<string> processedLabels;
 150                 set<string> userLabels = labels;
 151
 152         ofstream out;
 153         map<string,string> variables;
 154         variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(sharedfile));
 155                 string outputFileName = getOutputFileName("summary",variables);
 156         m->openOutputFile(outputFileName, out);
 157         outputNames.push_back(outputFileName);  outputTypes["summary"].push_back(outputFileName);
 158         out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
 159         out << "H\tpvalue\n";
 160
 161         //math goes here
 162
 163         int N = m->getNumGroups();
 164         double H;
 165         double tmp = 0.0;
 166         vector<groupRank> vec;
 167         vector<string> groups = m->getGroups();
 168         string group;
 169         int count;
 170         double sum;
 171
 172         //merge all groups into a vector
 173
 174
 175
 176         //rank function here
 177         assignRank(vec);
 178
 179         //populate counts and ranSums vectors
 180         for (int i=0;i<N;i++) {
 181             count = 0;
 182             sum = 0;
 183             group = groups[i];
 184             for(int j;j<vec.size();j++) {
 185                 if (vec[j].group == group) {
 186                     count++;
 187                     sum = sum + vec[j].rank;
 188                 }
 189             }
 190             counts[i] = count;
 191             rankSums[i] = sum;
 192         }
 193
 194         //test statistic
 195         for (int i=0;i<N;i++) { tmp = tmp + (pow(rankSums[i],2) / counts[i]); }
 196
 197         H = (12 / (N*(N+1))) * tmp - (3*(N+1));
 198
 199         //ss = tmp - pow(accumulate(rankSums.begin(), rankSums.end(), 0), 2);
 200
 201         //H = ss / ( (N * (N + 1))/12 );
 202
 203         //correction for ties?
 204
 205         //p-value calculation
 206
 207                 return 0;
 208         }
 209         catch(exception& e) {
 210                 m->errorOut(e, "KruskalWallisCommand", "execute");
 211                 exit(1);
 212         }
 213 }
 214 //**********************************************************************************************************************
 215 void KruskalWallisCommand::assignRank(vector<groupRank> &vec) {
 216     try {
 217         double rank = 1;
 218         double numRanks, avgRank, j;
 219         vector<groupRank>::iterator it, oldit;
 220
 221         sort (vec.begin(), vec.end(), comparevalue);
 222
 223         it = vec.begin();
 224
 225         while ( it != vec.end() ) {
 226             j = rank;
 227             oldit = it;
 228             if (!equalvalue(*it, *(it+1))) {
 229                 (*it).rank = rank;
 230                 rank = rank+1;
 231                 it++; }
 232             else {
 233                 while(equalrank(*it, *(it+1))) {
 234                     j = j + (j+1);
 235                     rank++;
 236                     it++;
 237                 }
 238                 numRanks = double (distance(oldit, it));
 239                 avgRank = j / numRanks;
 240                 while(oldit != it) {
 241                     (*oldit).rank = avgRank;
 242                     oldit++;
 243                 }
 244             }
 245
 246         }
 247
 248
 249     }
 250     catch(exception& e) {
 251                 m->errorOut(e, "KruskalWallisCommand", "getRank");
 252                 exit(1);
 253         }
 254
 255 }
 256 //**********************************************************************************************************************
 257 void KruskalWallisCommand::assignValue(vector<groupRank> &vec) {
 258
 259 }
 260 //**********************************************************************************************************************
 261 //**********************************************************************************************************************
 262 //**********************************************************************************************************************
 263