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 string HomovaCommand::getOutputFileNameTag(string type, string inputName=""){
  54         try {
  55         string outputFileName = "";
  56                 map<string, vector<string> >::iterator it;
  57
  58         //is this a type this command creates
  59         it = outputTypes.find(type);
  60         if (it == outputTypes.end()) {  m->mothurOut("[ERROR]: this command doesn't create a " + type + " output file.\n"); }
  61         else {
  62             if (type == "homova")            {   outputFileName =  "homova";   }
  63             else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true;  }
  64         }
  65         return outputFileName;
  66         }
  67         catch(exception& e) {
  68                 m->errorOut(e, "HomovaCommand", "getOutputFileNameTag");
  69                 exit(1);
  70         }
  71 }
  72 //**********************************************************************************************************************
  73 HomovaCommand::HomovaCommand(){
  74         try {
  75                 abort = true; calledHelp = true;
  76                 setParameters();
  77                 vector<string> tempOutNames;
  78                 outputTypes["homova"] = tempOutNames;
  79         }
  80         catch(exception& e) {
  81                 m->errorOut(e, "HomovaCommand", "HomovaCommand");
  82                 exit(1);
  83         }
  84 }
  85 //**********************************************************************************************************************
  86
  87 HomovaCommand::HomovaCommand(string option) {
  88         try {
  89                 abort = false; calledHelp = false;
  90
  91                 //allow user to run help
  92                 if(option == "help") { help(); abort = true; calledHelp = true; }
  93                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  94
  95                 else {
  96                         vector<string> myArray = setParameters();
  97
  98                         OptionParser parser(option);
  99                         map<string,string> parameters = parser.getParameters();
 100
 101                         ValidParameters validParameter;
 102
 103                         //check to make sure all parameters are valid for command
 104                         map<string,string>::iterator it;
 105                         for (it = parameters.begin(); it != parameters.end(); it++) {
 106                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 107                         }
 108
 109                         //initialize outputTypes
 110                         vector<string> tempOutNames;
 111                         outputTypes["homova"] = tempOutNames;
 112
 113                         //if the user changes the output directory command factory will send this info to us in the output parameter
 114                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
 115
 116                         //if the user changes the input directory command factory will send this info to us in the output parameter
 117                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 118                         if (inputDir == "not found"){   inputDir = "";          }
 119                         else {
 120                                 string path;
 121                                 it = parameters.find("design");
 122                                 //user has given a template file
 123                                 if(it != parameters.end()){
 124                                         path = m->hasPath(it->second);
 125                                         //if the user has not given a path then, add inputdir. else leave path alone.
 126                                         if (path == "") {       parameters["design"] = inputDir + it->second;           }
 127                                 }
 128
 129                                 it = parameters.find("phylip");
 130                                 //user has given a template file
 131                                 if(it != parameters.end()){
 132                                         path = m->hasPath(it->second);
 133                                         //if the user has not given a path then, add inputdir. else leave path alone.
 134                                         if (path == "") {       parameters["phylip"] = inputDir + it->second;           }
 135                                 }
 136                         }
 137
 138                         phylipFileName = validParameter.validFile(parameters, "phylip", true);
 139                         if (phylipFileName == "not open") { phylipFileName = ""; abort = true; }
 140                         else if (phylipFileName == "not found") {
 141                                 //if there is a current phylip file, use it
 142                                 phylipFileName = m->getPhylipFile();
 143                                 if (phylipFileName != "") { m->mothurOut("Using " + phylipFileName + " as input file for the phylip parameter."); m->mothurOutEndLine(); }
 144                                 else {  m->mothurOut("You have no current phylip file and the phylip parameter is required."); m->mothurOutEndLine(); abort = true; }
 145
 146                         }else { m->setPhylipFile(phylipFileName); }
 147
 148                         //check for required parameters
 149                         designFileName = validParameter.validFile(parameters, "design", true);
 150                         if (designFileName == "not open") { abort = true; }
 151                         else if (designFileName == "not found") {
 152                                 //if there is a current design file, use it
 153                                 designFileName = m->getDesignFile();
 154                                 if (designFileName != "") { m->mothurOut("Using " + designFileName + " as input file for the design parameter."); m->mothurOutEndLine(); }
 155                                 else {  m->mothurOut("You have no current design file and the design parameter is required."); m->mothurOutEndLine(); abort = true; }
 156                         }else { m->setDesignFile(designFileName); }
 157
 158                         string temp = validParameter.validFile(parameters, "iters", false);
 159                         if (temp == "not found") { temp = "1000"; }
 160                         m->mothurConvert(temp, iters);
 161
 162                         temp = validParameter.validFile(parameters, "alpha", false);
 163                         if (temp == "not found") { temp = "0.05"; }
 164                         m->mothurConvert(temp, experimentwiseAlpha);
 165                 }
 166
 167         }
 168         catch(exception& e) {
 169                 m->errorOut(e, "HomovaCommand", "HomovaCommand");
 170                 exit(1);
 171         }
 172 }
 173 //**********************************************************************************************************************
 174
 175 int HomovaCommand::execute(){
 176         try {
 177
 178                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 179
 180                 //read design file
 181                 designMap = new GroupMap(designFileName);
 182                 designMap->readDesignMap();
 183
 184                 if (outputDir == "") { outputDir = m->hasPath(phylipFileName); }
 185
 186                 //read in distance matrix and square it
 187                 ReadPhylipVector readMatrix(phylipFileName);
 188                 vector<string> sampleNames = readMatrix.read(distanceMatrix);
 189
 190                 for(int i=0;i<distanceMatrix.size();i++){
 191                         for(int j=0;j<i;j++){
 192                                 distanceMatrix[i][j] *= distanceMatrix[i][j];
 193                         }
 194                 }
 195
 196                 //link designMap to rows/columns in distance matrix
 197                 map<string, vector<int> > origGroupSampleMap;
 198                 for(int i=0;i<sampleNames.size();i++){
 199                         string group = designMap->getGroup(sampleNames[i]);
 200
 201                         if (group == "not found") {
 202                                 m->mothurOut("[ERROR]: " + sampleNames[i] + " is not in your design file, please correct."); m->mothurOutEndLine(); m->control_pressed = true;
 203                         }else { origGroupSampleMap[group].push_back(i); }
 204                 }
 205                 int numGroups = origGroupSampleMap.size();
 206
 207                 if (m->control_pressed) { delete designMap; return 0; }
 208
 209                 //create a new filename
 210                 ofstream HOMOVAFile;
 211                 string HOMOVAFileName = outputDir + m->getRootName(m->getSimpleName(phylipFileName)) + getOutputFileNameTag("homova");
 212                 m->openOutputFile(HOMOVAFileName, HOMOVAFile);
 213                 outputNames.push_back(HOMOVAFileName); outputTypes["homova"].push_back(HOMOVAFileName);
 214
 215                 HOMOVAFile << "HOMOVA\tBValue\tP-value\tSSwithin/(Ni-1)_values" << endl;
 216                 m->mothurOut("HOMOVA\tBValue\tP-value\tSSwithin/(Ni-1)_values\n");
 217
 218                 double fullHOMOVAPValue = runHOMOVA(HOMOVAFile, origGroupSampleMap, experimentwiseAlpha);
 219
 220                 if(fullHOMOVAPValue <= experimentwiseAlpha && numGroups > 2){
 221
 222                         int numCombos = numGroups * (numGroups-1) / 2;
 223                         double pairwiseAlpha = experimentwiseAlpha / (double) numCombos;
 224
 225                         map<string, vector<int> >::iterator itA;
 226                         map<string, vector<int> >::iterator itB;
 227
 228                         for(itA=origGroupSampleMap.begin();itA!=origGroupSampleMap.end();itA++){
 229                                 itB = itA;itB++;
 230                                 for(;itB!=origGroupSampleMap.end();itB++){
 231                                         map<string, vector<int> > pairwiseGroupSampleMap;
 232                                         pairwiseGroupSampleMap[itA->first] = itA->second;
 233                                         pairwiseGroupSampleMap[itB->first] = itB->second;
 234
 235                                         runHOMOVA(HOMOVAFile, pairwiseGroupSampleMap, pairwiseAlpha);
 236                                 }
 237                         }
 238                         HOMOVAFile << endl;
 239                         m->mothurOutEndLine();
 240
 241                         m->mothurOut("Experiment-wise error rate: " + toString(experimentwiseAlpha) + '\n');
 242                         m->mothurOut("Pair-wise error rate (Bonferroni): " + toString(pairwiseAlpha) + '\n');
 243                 }
 244                 else{
 245                         m->mothurOut("Experiment-wise error rate: " + toString(experimentwiseAlpha) + '\n');
 246                 }
 247
 248                 m->mothurOut("If you have borderline P-values, you should try increasing the number of iterations\n");
 249
 250                 delete designMap;
 251
 252                 m->mothurOutEndLine();
 253                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 254                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 255                 m->mothurOutEndLine();
 256
 257                 return 0;
 258         }
 259         catch(exception& e) {
 260                 m->errorOut(e, "HomovaCommand", "execute");
 261                 exit(1);
 262         }
 263 }
 264
 265 //**********************************************************************************************************************
 266
 267 double HomovaCommand::runHOMOVA(ofstream& HOMOVAFile, map<string, vector<int> > groupSampleMap, double alpha){
 268         try {
 269                 map<string, vector<int> >::iterator it;
 270                 int numGroups = groupSampleMap.size();
 271
 272                 vector<double> ssWithinOrigVector;
 273                 double bValueOrig = calcBValue(groupSampleMap, ssWithinOrigVector);
 274
 275                 double counter = 0;
 276                 for(int i=0;i<iters;i++){
 277                         vector<double> ssWithinRandVector;
 278                         map<string, vector<int> > randomizedGroup = getRandomizedGroups(groupSampleMap);
 279                         double bValueRand = calcBValue(randomizedGroup, ssWithinRandVector);
 280                         if(bValueRand > bValueOrig){    counter++;      }
 281                 }
 282
 283                 double pValue = (double) counter / (double) iters;
 284                 string pString = "";
 285                 if(pValue < 1/(double)iters){   pString = '<' + toString(1/(double)iters);      }
 286                 else                                            {       pString = toString(pValue);                                     }
 287
 288
 289                 //print homova table
 290                 it = groupSampleMap.begin();
 291                 HOMOVAFile << it->first;
 292                 m->mothurOut(it->first);
 293                 it++;
 294                 for(;it!=groupSampleMap.end();it++){
 295                         HOMOVAFile << '-' << it->first;
 296                         m->mothurOut('-' + it->first);
 297                 }
 298
 299                 HOMOVAFile << '\t' << bValueOrig << '\t' << pString;
 300                 m->mothurOut('\t' + toString(bValueOrig) + '\t' + pString);
 301
 302                 if(pValue < alpha){
 303                         HOMOVAFile << "*";
 304                         m->mothurOut("*");
 305                 }
 306
 307                 for(int i=0;i<numGroups;i++){
 308                         HOMOVAFile << '\t' << ssWithinOrigVector[i];
 309                         m->mothurOut('\t' + toString(ssWithinOrigVector[i]));
 310                 }
 311                 HOMOVAFile << endl;
 312                 m->mothurOutEndLine();
 313
 314                 return pValue;
 315         }
 316         catch(exception& e) {
 317                 m->errorOut(e, "HomovaCommand", "runHOMOVA");
 318                 exit(1);
 319         }
 320 }
 321
 322 //**********************************************************************************************************************
 323
 324 double HomovaCommand::calcSigleSSWithin(vector<int> sampleIndices) {
 325         try {
 326                 double ssWithin = 0.0;
 327                 int numSamplesInGroup = sampleIndices.size();
 328
 329                 for(int i=0;i<numSamplesInGroup;i++){
 330                         int row = sampleIndices[i];
 331
 332                         for(int j=0;j<numSamplesInGroup;j++){
 333                                 int col = sampleIndices[j];
 334
 335                                 if(col < row){
 336                                         ssWithin += distanceMatrix[row][col];
 337                                 }
 338
 339                         }
 340                 }
 341
 342                 ssWithin /= numSamplesInGroup;
 343                 return ssWithin;
 344         }
 345         catch(exception& e) {
 346                 m->errorOut(e, "HomovaCommand", "calcSigleSSWithin");
 347                 exit(1);
 348         }
 349 }
 350
 351 //**********************************************************************************************************************
 352
 353 double HomovaCommand::calcBValue(map<string, vector<int> > groupSampleMap, vector<double>& ssWithinVector) {
 354         try {
 355
 356                 map<string, vector<int> >::iterator it;
 357
 358                 double numGroups = (double)groupSampleMap.size();
 359                 ssWithinVector.resize(numGroups, 0);
 360
 361                 double totalNumSamples = 0;
 362                 double ssWithinFull;
 363                 double secondTermSum = 0;
 364                 double inverseOneMinusSum = 0;
 365                 int index = 0;
 366
 367                 ssWithinVector.resize(numGroups, 0);
 368                 for(it = groupSampleMap.begin();it!=groupSampleMap.end();it++){
 369                         int numSamplesInGroup = it->second.size();
 370                         totalNumSamples += numSamplesInGroup;
 371
 372                         ssWithinVector[index] = calcSigleSSWithin(it->second);
 373                         ssWithinFull += ssWithinVector[index];
 374
 375                         secondTermSum += (numSamplesInGroup - 1) * log(ssWithinVector[index] / (double)(numSamplesInGroup - 1));
 376                         inverseOneMinusSum += 1.0 / (double)(numSamplesInGroup - 1);
 377
 378                         ssWithinVector[index] /= (double)(numSamplesInGroup - 1); //this line is only for output purposes to scale SSw by the number of samples in the group
 379                         index++;
 380                 }
 381
 382                 double B = (totalNumSamples - numGroups) * log(ssWithinFull/(totalNumSamples-numGroups)) - secondTermSum;
 383                 double denomintor = 1 + 1.0/(3.0 * (numGroups - 1.0)) * (inverseOneMinusSum - 1.0 / (double) (totalNumSamples - numGroups));
 384                 B /= denomintor;
 385
 386                 return B;
 387
 388         }
 389         catch(exception& e) {
 390                 m->errorOut(e, "HomovaCommand", "calcBValue");
 391                 exit(1);
 392         }
 393 }
 394
 395 //**********************************************************************************************************************
 396
 397 map<string, vector<int> > HomovaCommand::getRandomizedGroups(map<string, vector<int> > origMapping){
 398         try{
 399                 vector<int> sampleIndices;
 400                 vector<int> samplesPerGroup;
 401
 402                 map<string, vector<int> >::iterator it;
 403                 for(it=origMapping.begin();it!=origMapping.end();it++){
 404                         vector<int> indices = it->second;
 405                         samplesPerGroup.push_back(indices.size());
 406                         sampleIndices.insert(sampleIndices.end(), indices.begin(), indices.end());
 407                 }
 408
 409                 random_shuffle(sampleIndices.begin(), sampleIndices.end());
 410
 411                 int index = 0;
 412                 map<string, vector<int> > randomizedGroups = origMapping;
 413                 for(it=randomizedGroups.begin();it!=randomizedGroups.end();it++){
 414                         for(int i=0;i<it->second.size();i++){
 415                                 it->second[i] = sampleIndices[index++];
 416                         }
 417                 }
 418
 419                 return randomizedGroups;
 420         }
 421         catch (exception& e) {
 422                 m->errorOut(e, "AmovaCommand", "randomizeGroups");
 423                 exit(1);
 424         }
 425 }
 426
 427 //**********************************************************************************************************************
 428
 429