homovacommand.cpp

   1 /*
   2  *  homovacommand.cpp
   3  *  mothur
   4  *
   5  *  Created by westcott on 2/8/11.
   6  *  Copyright 2011 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "homovacommand.h"
  11 #include "groupmap.h"
  12 #include "readphylipvector.h"
  13
  14 //**********************************************************************************************************************
  15 vector<string> HomovaCommand::setParameters(){
  16         try {
  17                 CommandParameter pdesign("design", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pdesign);
  18                 CommandParameter pphylip("phylip", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pphylip);
  19                 CommandParameter piters("iters", "Number", "", "1000", "", "", "",false,false); parameters.push_back(piters);
  20                 CommandParameter palpha("alpha", "Number", "", "0.05", "", "", "",false,false); parameters.push_back(palpha);
  21                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  22                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  23
  24                 vector<string> myArray;
  25                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  26                 return myArray;
  27         }
  28         catch(exception& e) {
  29                 m->errorOut(e, "HomovaCommand", "setParameters");
  30                 exit(1);
  31         }
  32 }
  33 //**********************************************************************************************************************
  34 string HomovaCommand::getHelpString(){
  35         try {
  36                 string helpString = "";
  37                 helpString += "Referenced: Stewart CN, Excoffier L (1996). Assessing population genetic structure and variability with RAPD data: Application to Vaccinium macrocarpon (American Cranberry). J Evol Biol 9: 153-71.\n";
  38                 helpString += "The homova command outputs a .homova file. \n";
  39                 helpString += "The homova command parameters are phylip, iters, and alpha.  The phylip and design parameters are required, unless valid current files exist.\n";
  40                 helpString += "The design parameter allows you to assign your samples to groups when you are running homova. It is required. \n";
  41                 helpString += "The design file looks like the group file.  It is a 2 column tab delimited file, where the first column is the sample name and the second column is the group the sample belongs to.\n";
  42                 helpString += "The iters parameter allows you to set number of randomization for the P value.  The default is 1000. \n";
  43                 helpString += "The homova command should be in the following format: homova(phylip=file.dist, design=file.design).\n";
  44                 helpString += "Note: No spaces between parameter labels (i.e. iters), '=' and parameters (i.e. 1000).\n";
  45                 return helpString;
  46         }
  47         catch(exception& e) {
  48                 m->errorOut(e, "HomovaCommand", "getHelpString");
  49                 exit(1);
  50         }
  51 }
  52
  53 //**********************************************************************************************************************
  54
  55 HomovaCommand::HomovaCommand(){
  56         try {
  57                 abort = true; calledHelp = true;
  58                 setParameters();
  59                 vector<string> tempOutNames;
  60                 outputTypes["homova"] = tempOutNames;
  61         }
  62         catch(exception& e) {
  63                 m->errorOut(e, "HomovaCommand", "HomovaCommand");
  64                 exit(1);
  65         }
  66 }
  67 //**********************************************************************************************************************
  68
  69 HomovaCommand::HomovaCommand(string option) {
  70         try {
  71                 abort = false; calledHelp = false;
  72
  73                 //allow user to run help
  74                 if(option == "help") { help(); abort = true; calledHelp = true; }
  75
  76                 else {
  77                         vector<string> myArray = setParameters();
  78
  79                         OptionParser parser(option);
  80                         map<string,string> parameters = parser.getParameters();
  81
  82                         ValidParameters validParameter;
  83
  84                         //check to make sure all parameters are valid for command
  85                         map<string,string>::iterator it;
  86                         for (it = parameters.begin(); it != parameters.end(); it++) {
  87                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  88                         }
  89
  90                         //initialize outputTypes
  91                         vector<string> tempOutNames;
  92                         outputTypes["homova"] = tempOutNames;
  93
  94                         //if the user changes the output directory command factory will send this info to us in the output parameter
  95                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
  96
  97                         //if the user changes the input directory command factory will send this info to us in the output parameter
  98                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  99                         if (inputDir == "not found"){   inputDir = "";          }
 100                         else {
 101                                 string path;
 102                                 it = parameters.find("design");
 103                                 //user has given a template file
 104                                 if(it != parameters.end()){
 105                                         path = m->hasPath(it->second);
 106                                         //if the user has not given a path then, add inputdir. else leave path alone.
 107                                         if (path == "") {       parameters["design"] = inputDir + it->second;           }
 108                                 }
 109
 110                                 it = parameters.find("phylip");
 111                                 //user has given a template file
 112                                 if(it != parameters.end()){
 113                                         path = m->hasPath(it->second);
 114                                         //if the user has not given a path then, add inputdir. else leave path alone.
 115                                         if (path == "") {       parameters["phylip"] = inputDir + it->second;           }
 116                                 }
 117                         }
 118
 119                         phylipFileName = validParameter.validFile(parameters, "phylip", true);
 120                         if (phylipFileName == "not open") { phylipFileName = ""; abort = true; }
 121                         else if (phylipFileName == "not found") {
 122                                 //if there is a current phylip file, use it
 123                                 phylipFileName = m->getPhylipFile();
 124                                 if (phylipFileName != "") { m->mothurOut("Using " + phylipFileName + " as input file for the phylip parameter."); m->mothurOutEndLine(); }
 125                                 else {  m->mothurOut("You have no current phylip file and the phylip parameter is required."); m->mothurOutEndLine(); abort = true; }
 126
 127                         }
 128
 129                         //check for required parameters
 130                         designFileName = validParameter.validFile(parameters, "design", true);
 131                         if (designFileName == "not open") { abort = true; }
 132                         else if (designFileName == "not found") {
 133                                 //if there is a current design file, use it
 134                                 designFileName = m->getDesignFile();
 135                                 if (designFileName != "") { m->mothurOut("Using " + designFileName + " as input file for the design parameter."); m->mothurOutEndLine(); }
 136                                 else {  m->mothurOut("You have no current design file and the design parameter is required."); m->mothurOutEndLine(); abort = true; }
 137                         }
 138
 139                         string temp = validParameter.validFile(parameters, "iters", false);
 140                         if (temp == "not found") { temp = "1000"; }
 141                         convert(temp, iters);
 142
 143                         temp = validParameter.validFile(parameters, "alpha", false);
 144                         if (temp == "not found") { temp = "0.05"; }
 145                         convert(temp, experimentwiseAlpha);
 146                 }
 147
 148         }
 149         catch(exception& e) {
 150                 m->errorOut(e, "HomovaCommand", "HomovaCommand");
 151                 exit(1);
 152         }
 153 }
 154 //**********************************************************************************************************************
 155
 156 int HomovaCommand::execute(){
 157         try {
 158
 159                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 160
 161                 //read design file
 162                 designMap = new GroupMap(designFileName);
 163                 designMap->readDesignMap();
 164
 165                 if (outputDir == "") { outputDir = m->hasPath(phylipFileName); }
 166
 167                 //read in distance matrix and square it
 168                 ReadPhylipVector readMatrix(phylipFileName);
 169                 vector<string> sampleNames = readMatrix.read(distanceMatrix);
 170
 171                 for(int i=0;i<distanceMatrix.size();i++){
 172                         for(int j=0;j<i;j++){
 173                                 distanceMatrix[i][j] *= distanceMatrix[i][j];
 174                         }
 175                 }
 176
 177                 //link designMap to rows/columns in distance matrix
 178                 map<string, vector<int> > origGroupSampleMap;
 179                 for(int i=0;i<sampleNames.size();i++){
 180                         origGroupSampleMap[designMap->getGroup(sampleNames[i])].push_back(i);
 181                 }
 182                 int numGroups = origGroupSampleMap.size();
 183
 184                 //create a new filename
 185                 ofstream HOMOVAFile;
 186                 string HOMOVAFileName = outputDir + m->getRootName(m->getSimpleName(phylipFileName))  + "homova";
 187                 m->openOutputFile(HOMOVAFileName, HOMOVAFile);
 188                 outputNames.push_back(HOMOVAFileName); outputTypes["homova"].push_back(HOMOVAFileName);
 189
 190                 HOMOVAFile << "HOMOVA\tBValue\tP-value\tSSwithin/(Ni-1)_values" << endl;
 191                 m->mothurOut("HOMOVA\tBValue\tP-value\tSSwithin/(Ni-1)_values\n");
 192
 193                 double fullHOMOVAPValue = runHOMOVA(HOMOVAFile, origGroupSampleMap, experimentwiseAlpha);
 194
 195                 if(fullHOMOVAPValue <= experimentwiseAlpha && numGroups > 2){
 196
 197                         int numCombos = numGroups * (numGroups-1) / 2;
 198                         double pairwiseAlpha = experimentwiseAlpha / (double) numCombos;
 199
 200                         map<string, vector<int> >::iterator itA;
 201                         map<string, vector<int> >::iterator itB;
 202
 203                         for(itA=origGroupSampleMap.begin();itA!=origGroupSampleMap.end();itA++){
 204                                 itB = itA;itB++;
 205                                 for(;itB!=origGroupSampleMap.end();itB++){
 206                                         map<string, vector<int> > pairwiseGroupSampleMap;
 207                                         pairwiseGroupSampleMap[itA->first] = itA->second;
 208                                         pairwiseGroupSampleMap[itB->first] = itB->second;
 209
 210                                         runHOMOVA(HOMOVAFile, pairwiseGroupSampleMap, pairwiseAlpha);
 211                                 }
 212                         }
 213                         HOMOVAFile << endl;
 214                         m->mothurOutEndLine();
 215
 216                         m->mothurOut("Experiment-wise error rate: " + toString(experimentwiseAlpha) + '\n');
 217                         m->mothurOut("Pair-wise error rate (Bonferroni): " + toString(pairwiseAlpha) + '\n');
 218                 }
 219                 else{
 220                         m->mothurOut("Experiment-wise error rate: " + toString(experimentwiseAlpha) + '\n');
 221                 }
 222
 223                 m->mothurOut("If you have borderline P-values, you should try increasing the number of iterations\n");
 224
 225                 delete designMap;
 226
 227                 m->mothurOutEndLine();
 228                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 229                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 230                 m->mothurOutEndLine();
 231
 232                 return 0;
 233         }
 234         catch(exception& e) {
 235                 m->errorOut(e, "HomovaCommand", "execute");
 236                 exit(1);
 237         }
 238 }
 239
 240 //**********************************************************************************************************************
 241
 242 double HomovaCommand::runHOMOVA(ofstream& HOMOVAFile, map<string, vector<int> > groupSampleMap, double alpha){
 243         try {
 244                 map<string, vector<int> >::iterator it;
 245                 int numGroups = groupSampleMap.size();
 246
 247                 vector<double> ssWithinOrigVector;
 248                 double bValueOrig = calcBValue(groupSampleMap, ssWithinOrigVector);
 249
 250                 double counter = 0;
 251                 for(int i=0;i<iters;i++){
 252                         vector<double> ssWithinRandVector;
 253                         map<string, vector<int> > randomizedGroup = getRandomizedGroups(groupSampleMap);
 254                         double bValueRand = calcBValue(randomizedGroup, ssWithinRandVector);
 255                         if(bValueRand > bValueOrig){    counter++;      }
 256                 }
 257
 258                 double pValue = (double) counter / (double) iters;
 259                 string pString = "";
 260                 if(pValue < 1/(double)iters){   pString = '<' + toString(1/(double)iters);      }
 261                 else                                            {       pString = toString(pValue);                                     }
 262
 263
 264                 //print homova table
 265                 it = groupSampleMap.begin();
 266                 HOMOVAFile << it->first;
 267                 m->mothurOut(it->first);
 268                 it++;
 269                 for(;it!=groupSampleMap.end();it++){
 270                         HOMOVAFile << '-' << it->first;
 271                         m->mothurOut('-' + it->first);
 272                 }
 273
 274                 HOMOVAFile << '\t' << bValueOrig << '\t' << pString;
 275                 m->mothurOut('\t' + toString(bValueOrig) + '\t' + pString);
 276
 277                 if(pValue < alpha){
 278                         HOMOVAFile << "*";
 279                         m->mothurOut("*");
 280                 }
 281
 282                 for(int i=0;i<numGroups;i++){
 283                         HOMOVAFile << '\t' << ssWithinOrigVector[i];
 284                         m->mothurOut('\t' + toString(ssWithinOrigVector[i]));
 285                 }
 286                 HOMOVAFile << endl;
 287                 m->mothurOutEndLine();
 288
 289                 return pValue;
 290         }
 291         catch(exception& e) {
 292                 m->errorOut(e, "HomovaCommand", "runHOMOVA");
 293                 exit(1);
 294         }
 295 }
 296
 297 //**********************************************************************************************************************
 298
 299 double HomovaCommand::calcSigleSSWithin(vector<int> sampleIndices) {
 300         try {
 301                 double ssWithin = 0.0;
 302                 int numSamplesInGroup = sampleIndices.size();
 303
 304                 for(int i=0;i<numSamplesInGroup;i++){
 305                         int row = sampleIndices[i];
 306
 307                         for(int j=0;j<numSamplesInGroup;j++){
 308                                 int col = sampleIndices[j];
 309
 310                                 if(col < row){
 311                                         ssWithin += distanceMatrix[row][col];
 312                                 }
 313
 314                         }
 315                 }
 316
 317                 ssWithin /= numSamplesInGroup;
 318                 return ssWithin;
 319         }
 320         catch(exception& e) {
 321                 m->errorOut(e, "HomovaCommand", "calcSigleSSWithin");
 322                 exit(1);
 323         }
 324 }
 325
 326 //**********************************************************************************************************************
 327
 328 double HomovaCommand::calcBValue(map<string, vector<int> > groupSampleMap, vector<double>& ssWithinVector) {
 329         try {
 330
 331                 map<string, vector<int> >::iterator it;
 332
 333                 double numGroups = (double)groupSampleMap.size();
 334                 ssWithinVector.resize(numGroups, 0);
 335
 336                 double totalNumSamples = 0;
 337                 double ssWithinFull;
 338                 double secondTermSum = 0;
 339                 double inverseOneMinusSum = 0;
 340                 int index = 0;
 341
 342                 ssWithinVector.resize(numGroups, 0);
 343                 for(it = groupSampleMap.begin();it!=groupSampleMap.end();it++){
 344                         int numSamplesInGroup = it->second.size();
 345                         totalNumSamples += numSamplesInGroup;
 346
 347                         ssWithinVector[index] = calcSigleSSWithin(it->second);
 348                         ssWithinFull += ssWithinVector[index];
 349
 350                         secondTermSum += (numSamplesInGroup - 1) * log(ssWithinVector[index] / (double)(numSamplesInGroup - 1));
 351                         inverseOneMinusSum += 1.0 / (double)(numSamplesInGroup - 1);
 352
 353                         ssWithinVector[index] /= (double)(numSamplesInGroup - 1); //this line is only for output purposes to scale SSw by the number of samples in the group
 354                         index++;
 355                 }
 356
 357                 double B = (totalNumSamples - numGroups) * log(ssWithinFull/(totalNumSamples-numGroups)) - secondTermSum;
 358                 double denomintor = 1 + 1.0/(3.0 * (numGroups - 1.0)) * (inverseOneMinusSum - 1.0 / (double) (totalNumSamples - numGroups));
 359                 B /= denomintor;
 360
 361                 return B;
 362
 363         }
 364         catch(exception& e) {
 365                 m->errorOut(e, "HomovaCommand", "calcBValue");
 366                 exit(1);
 367         }
 368 }
 369
 370 //**********************************************************************************************************************
 371
 372 map<string, vector<int> > HomovaCommand::getRandomizedGroups(map<string, vector<int> > origMapping){
 373         try{
 374                 vector<int> sampleIndices;
 375                 vector<int> samplesPerGroup;
 376
 377                 map<string, vector<int> >::iterator it;
 378                 for(it=origMapping.begin();it!=origMapping.end();it++){
 379                         vector<int> indices = it->second;
 380                         samplesPerGroup.push_back(indices.size());
 381                         sampleIndices.insert(sampleIndices.end(), indices.begin(), indices.end());
 382                 }
 383
 384                 random_shuffle(sampleIndices.begin(), sampleIndices.end());
 385
 386                 int index = 0;
 387                 map<string, vector<int> > randomizedGroups = origMapping;
 388                 for(it=randomizedGroups.begin();it!=randomizedGroups.end();it++){
 389                         for(int i=0;i<it->second.size();i++){
 390                                 it->second[i] = sampleIndices[index++];
 391                         }
 392                 }
 393
 394                 return randomizedGroups;
 395         }
 396         catch (exception& e) {
 397                 m->errorOut(e, "AmovaCommand", "randomizeGroups");
 398                 exit(1);
 399         }
 400 }
 401
 402 //**********************************************************************************************************************
 403
 404