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
  16                 vector<string> myArray;
  17                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  18                 return myArray;
  19         }
  20         catch(exception& e) {
  21                 m->errorOut(e, "KruskalWallisCommand", "setParameters");
  22                 exit(1);
  23         }
  24 }
  25 //**********************************************************************************************************************
  26 string KruskalWallisCommand::getHelpString(){
  27         try {
  28                 string helpString = "";
  29                 helpString += "The kruskalwallis command parameter options are \n";
  30         helpString += "Kruskal–Wallis one-way analysis of variance is a non-parametric method for testing whether samples originate from the same distribution.";
  31                 return helpString;
  32         }
  33         catch(exception& e) {
  34                 m->errorOut(e, "KruskalWallisCommand", "getHelpString");
  35                 exit(1);
  36         }
  37 }
  38 //**********************************************************************************************************************
  39 string KruskalWallisCommand::getOutputFileNameTag(string type, string inputName=""){
  40         try {
  41         string outputFileName = "";
  42                 map<string, vector<string> >::iterator it;
  43
  44         //is this a type this command creates
  45         it = outputTypes.find(type);
  46         if (it == outputTypes.end()) {  m->mothurOut("[ERROR]: this command doesn't create a " + type + " output file.\n"); }
  47         else {
  48             if (type == "summary") {  outputFileName =  "cooccurence.summary"; }
  49             else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true;  }
  50         }
  51         return outputFileName;
  52         }
  53         catch(exception& e) {
  54                 m->errorOut(e, "KruskalWallisCommand", "getOutputFileNameTag");
  55                 exit(1);
  56         }
  57 }
  58 //**********************************************************************************************************************
  59 KruskalWallisCommand::KruskalWallisCommand(){
  60         try {
  61                 abort = true; calledHelp = true;
  62                 setParameters();
  63         vector<string> tempOutNames;
  64                 outputTypes["summary"] = tempOutNames;
  65
  66         }
  67         catch(exception& e) {
  68                 m->errorOut(e, "KruskalWallisCommand", "KruskalWallisCommand");
  69                 exit(1);
  70         }
  71 }
  72 //**********************************************************************************************************************
  73 KruskalWallisCommand::KruskalWallisCommand(string option) {
  74         try {
  75                 abort = false; calledHelp = false;
  76
  77                 //allow user to run help
  78                 if(option == "help") { help(); abort = true; calledHelp = true; }
  79                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  80
  81                 else {
  82                         vector<string> myArray = setParameters();
  83
  84                         OptionParser parser(option);
  85                         map<string,string> parameters = parser.getParameters();
  86                         map<string,string>::iterator it;
  87
  88                         ValidParameters validParameter;
  89
  90                         //check to make sure all parameters are valid for command
  91                         for (it = parameters.begin(); it != parameters.end(); it++) {
  92                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  93                         }
  94
  95
  96                         //if the user changes the input directory command factory will send this info to us in the output parameter
  97                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  98                         if (inputDir == "not found"){   inputDir = "";          }
  99                         else {
 100                                 string path;
 101                                 it = parameters.find("shared");
 102                                 //user has given a template file
 103                                 if(it != parameters.end()){
 104                                         path = m->hasPath(it->second);
 105                                         //if the user has not given a path then, add inputdir. else leave path alone.
 106                                         if (path == "") {       parameters["shared"] = inputDir + it->second;           }
 107                                 }
 108                         }
 109
 110             vector<string> tempOutNames;
 111             outputTypes["summary"] = tempOutNames;
 112
 113
 114                 }
 115
 116         }
 117         catch(exception& e) {
 118                 m->errorOut(e, "KruskalWallisCommand", "KruskalWallisCommand");
 119                 exit(1);
 120         }
 121 }
 122 //**********************************************************************************************************************
 123 int KruskalWallisCommand::execute(){
 124         try {
 125                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 126
 127         InputData* input = new InputData(sharedfile, "sharedfile");
 128         vector<SharedRAbundVector*> lookup = input->getSharedRAbundVectors();
 129                 string lastLabel = lookup[0]->getLabel();
 130
 131                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 132                 set<string> processedLabels;
 133                 set<string> userLabels = labels;
 134
 135         ofstream out;
 136                 string outputFileName = outputDir + m->getRootName(m->getSimpleName(sharedfile)) + getOutputFileNameTag("summary");
 137         m->openOutputFile(outputFileName, out);
 138         outputNames.push_back(outputFileName);  outputTypes["summary"].push_back(outputFileName);
 139         out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
 140         out << "H\tpvalue\n";
 141
 142         //math goes here
 143
 144         int N = lookUp.size();
 145         double H;
 146         double tmp = 0.0;
 147         vector<groupRank> vec;
 148         string group;
 149         int count;
 150         double sum;
 151
 152         //merge all groups into a vector
 153         //rank function here
 154         assignRank(vec);
 155
 156         //populate counts and ranSums vectors
 157         for (int i=0;i<N;i++) {
 158             count = 0;
 159             sum = 0;
 160             //group = next group
 161             for(int j;j<vec.size();j++) {
 162                 if (vec[j].group == group) {
 163                     count++;
 164                     sum = sum + vec[j].rank;
 165                 }
 166             }
 167             counts[i] = count;
 168             rankSums[i] = sum;
 169         }
 170
 171         //test statistic
 172         for (int i=0;i<N;i++) { tmp = tmp + (pow(rankSums[i],2) / counts[i]); }
 173
 174         H = (12 / (N*(N+1))) * tmp - (3*(N+1));
 175
 176         //ss = tmp - pow(accumulate(rankSums.begin(), rankSums.end(), 0), 2);
 177
 178         //H = ss / ( (N * (N + 1))/12 );
 179
 180         //correction for ties?
 181
 182         //p-value calculation
 183
 184                 return 0;
 185         }
 186         catch(exception& e) {
 187                 m->errorOut(e, "KruskalWallisCommand", "execute");
 188                 exit(1);
 189         }
 190 }
 191 //**********************************************************************************************************************
 192 void KruskalWallisCommand::assignRank(vector<groupRank> &vec) {
 193     try {
 194         double rank = 1;
 195         double numRanks, avgRank, j;
 196         vector<groupRank>::iterator it, oldit;
 197
 198         sort (vec.begin(), vec.end(), comparevalue);
 199
 200         it = vec.begin();
 201
 202         while ( it != vec.end() ) {
 203             j = rank;
 204             oldit = it;
 205             if (!equalvalue(*it, *(it+1))) {
 206                 (*it).rank = rank;
 207                 rank = rank+1;
 208                 it++; }
 209             else {
 210                 while(equalrank(*it, *(it+1))) {
 211                     j = j + (j+1);
 212                     rank++;
 213                     it++;
 214                 }
 215                 numRanks = double (distance(oldit, it));
 216                 avgRank = j / numRanks;
 217                 while(oldit != it) {
 218                     (*oldit).rank = avgRank;
 219                     oldit++;
 220                 }
 221             }
 222
 223         }
 224
 225
 226     }
 227     catch(exception& e) {
 228                 m->errorOut(e, "KruskalWallisCommand", "getRank");
 229                 exit(1);
 230         }
 231
 232 }
 233 //**********************************************************************************************************************
 234
 235 //**********************************************************************************************************************
 236 //**********************************************************************************************************************
 237 //**********************************************************************************************************************