amovacommand.cpp

   1 /*
   2  *  amovacommand.cpp
   3  *  mothur
   4  *
   5  *  Created by westcott on 2/7/11.
   6  *  Copyright 2011 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "amovacommand.h"
  11 #include "readphylipvector.h"
  12 #include "groupmap.h"
  13
  14 //**********************************************************************************************************************
  15 vector<string> AmovaCommand::getValidParameters(){
  16         try {
  17                 string Array[] =  {"outputdir","iters","phylip","design","alpha", "inputdir"};
  18                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  19                 return myArray;
  20         }
  21         catch(exception& e) {
  22                 m->errorOut(e, "AmovaCommand", "getValidParameters");
  23                 exit(1);
  24         }
  25 }
  26 //**********************************************************************************************************************
  27 AmovaCommand::AmovaCommand(){
  28         try {
  29                 abort = true; calledHelp = true;
  30                 vector<string> tempOutNames;
  31                 outputTypes["amova"] = tempOutNames;
  32         }
  33         catch(exception& e) {
  34                 m->errorOut(e, "AmovaCommand", "AmovaCommand");
  35                 exit(1);
  36         }
  37 }
  38 //**********************************************************************************************************************
  39 vector<string> AmovaCommand::getRequiredParameters(){
  40         try {
  41                 string Array[] =  {"design"};
  42                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  43                 return myArray;
  44         }
  45         catch(exception& e) {
  46                 m->errorOut(e, "AmovaCommand", "getRequiredParameters");
  47                 exit(1);
  48         }
  49 }
  50 //**********************************************************************************************************************
  51 vector<string> AmovaCommand::getRequiredFiles(){
  52         try {
  53                 string Array[] =  {};
  54                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  55                 return myArray;
  56         }
  57         catch(exception& e) {
  58                 m->errorOut(e, "AmovaCommand", "getRequiredFiles");
  59                 exit(1);
  60         }
  61 }
  62 //**********************************************************************************************************************
  63
  64 AmovaCommand::AmovaCommand(string option) {
  65         try {
  66                 abort = false; calledHelp = false;
  67
  68                 //allow user to run help
  69                 if(option == "help") { help(); abort = true; calledHelp = true; }
  70
  71                 else {
  72                         //valid paramters for this command
  73                         string AlignArray[] =  {"design","outputdir","iters","phylip","alpha", "inputdir"};
  74                         vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
  75
  76                         OptionParser parser(option);
  77                         map<string,string> parameters = parser.getParameters();
  78
  79                         ValidParameters validParameter;
  80
  81                         //check to make sure all parameters are valid for command
  82                         map<string,string>::iterator it;
  83                         for (it = parameters.begin(); it != parameters.end(); it++) {
  84                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  85                         }
  86
  87                         //initialize outputTypes
  88                         vector<string> tempOutNames;
  89                         outputTypes["amova"] = tempOutNames;
  90
  91                         //if the user changes the output directory command factory will send this info to us in the output parameter
  92                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
  93
  94                         //if the user changes the input directory command factory will send this info to us in the output parameter
  95                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  96                         if (inputDir == "not found"){   inputDir = "";          }
  97                         else {
  98                                 string path;
  99                                 it = parameters.find("design");
 100                                 //user has given a template file
 101                                 if(it != parameters.end()){
 102                                         path = m->hasPath(it->second);
 103                                         //if the user has not given a path then, add inputdir. else leave path alone.
 104                                         if (path == "") {       parameters["design"] = inputDir + it->second;           }
 105                                 }
 106
 107                                 it = parameters.find("phylip");
 108                                 //user has given a template file
 109                                 if(it != parameters.end()){
 110                                         path = m->hasPath(it->second);
 111                                         //if the user has not given a path then, add inputdir. else leave path alone.
 112                                         if (path == "") {       parameters["phylip"] = inputDir + it->second;           }
 113                                 }
 114                         }
 115
 116                         phylipFileName = validParameter.validFile(parameters, "phylip", true);
 117                         if (phylipFileName == "not open") { phylipFileName = ""; abort = true; }
 118                         else if (phylipFileName == "not found") {
 119                                 phylipFileName = "";
 120                         }
 121
 122                         //check for required parameters
 123                         designFileName = validParameter.validFile(parameters, "design", true);
 124                         if (designFileName == "not open") { abort = true; }
 125                         else if (designFileName == "not found") {
 126                                 designFileName = "";
 127                         }
 128
 129                         string temp = validParameter.validFile(parameters, "iters", false);
 130                         if (temp == "not found") { temp = "1000"; }
 131                         convert(temp, iters);
 132
 133                         temp = validParameter.validFile(parameters, "alpha", false);
 134                         if (temp == "not found") { temp = "0.05"; }
 135                         convert(temp, experimentwiseAlpha);
 136                 }
 137         }
 138         catch(exception& e) {
 139                 m->errorOut(e, "AmovaCommand", "AmovaCommand");
 140                 exit(1);
 141         }
 142 }
 143
 144 //**********************************************************************************************************************
 145
 146 void AmovaCommand::help(){
 147         try {
 148                 m->mothurOut("Referenced: Anderson MJ (2001). A new method for non-parametric multivariate analysis of variance. Austral Ecol 26: 32-46.\n");
 149                 m->mothurOut("The amova command outputs a .amova file. \n");
 150                 m->mothurOut("The amova command parameters are phylip, iters, and alpha.  The phylip and design parameters are required.\n");
 151                 m->mothurOut("The design parameter allows you to assign your samples to groups when you are running amova. It is required. \n");
 152                 m->mothurOut("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");
 153                 m->mothurOut("The iters parameter allows you to set number of randomization for the P value.  The default is 1000. \n");
 154                 m->mothurOut("The amova command should be in the following format: amova(phylip=file.dist, design=file.design).\n");
 155                 m->mothurOut("Note: No spaces between parameter labels (i.e. iters), '=' and parameters (i.e. 1000).\n\n");
 156         }
 157         catch(exception& e) {
 158                 m->errorOut(e, "AmovaCommand", "help");
 159                 exit(1);
 160         }
 161 }
 162
 163 //**********************************************************************************************************************
 164
 165 AmovaCommand::~AmovaCommand(){}
 166
 167 //**********************************************************************************************************************
 168
 169 int AmovaCommand::execute(){
 170         try {
 171
 172                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 173
 174                 //read design file
 175                 designMap = new GroupMap(designFileName);
 176                 designMap->readDesignMap();
 177
 178                 if (outputDir == "") { outputDir = m->hasPath(phylipFileName); }
 179
 180                 //read in distance matrix and square it
 181                 ReadPhylipVector readMatrix(phylipFileName);
 182                 vector<string> sampleNames = readMatrix.read(distanceMatrix);
 183
 184                 for(int i=0;i<distanceMatrix.size();i++){
 185                         for(int j=0;j<i;j++){
 186                                 distanceMatrix[i][j] *= distanceMatrix[i][j];
 187                         }
 188                 }
 189
 190                 //link designMap to rows/columns in distance matrix
 191                 map<string, vector<int> > origGroupSampleMap;
 192                 for(int i=0;i<sampleNames.size();i++){
 193                         origGroupSampleMap[designMap->getGroup(sampleNames[i])].push_back(i);
 194                 }
 195                 int numGroups = origGroupSampleMap.size();
 196
 197                 //create a new filename
 198                 ofstream AMOVAFile;
 199                 string AMOVAFileName = outputDir + m->getRootName(m->getSimpleName(phylipFileName))  + "amova";
 200                 m->openOutputFile(AMOVAFileName, AMOVAFile);
 201                 outputNames.push_back(AMOVAFileName); outputTypes["amova"].push_back(AMOVAFileName);
 202
 203                 double fullANOVAPValue = runAMOVA(AMOVAFile, origGroupSampleMap, experimentwiseAlpha);
 204                 if(fullANOVAPValue <= experimentwiseAlpha && numGroups > 2){
 205
 206                         int numCombos = numGroups * (numGroups-1) / 2;
 207                         double pairwiseAlpha = experimentwiseAlpha / (double) numCombos;
 208
 209                         map<string, vector<int> >::iterator itA;
 210                         map<string, vector<int> >::iterator itB;
 211
 212                         for(itA=origGroupSampleMap.begin();itA!=origGroupSampleMap.end();itA++){
 213                                 itB = itA;itB++;
 214                                 for(itB;itB!=origGroupSampleMap.end();itB++){
 215
 216                                         map<string, vector<int> > pairwiseGroupSampleMap;
 217                                         pairwiseGroupSampleMap[itA->first] = itA->second;
 218                                         pairwiseGroupSampleMap[itB->first] = itB->second;
 219
 220                                         runAMOVA(AMOVAFile, pairwiseGroupSampleMap, pairwiseAlpha);
 221                                 }
 222                         }
 223                         m->mothurOut("Experiment-wise error rate: " + toString(experimentwiseAlpha) + '\n');
 224                         m->mothurOut("Pair-wise error rate (Bonferroni): " + toString(pairwiseAlpha) + '\n');
 225                 }
 226                 else{
 227                         m->mothurOut("Experiment-wise error rate: " + toString(experimentwiseAlpha) + '\n');
 228                 }
 229                 m->mothurOut("If you have borderline P-values, you should try increasing the number of iterations\n");
 230                 AMOVAFile.close();
 231
 232                 delete designMap;
 233
 234                 m->mothurOutEndLine();
 235                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 236                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 237                 m->mothurOutEndLine();
 238
 239                 return 0;
 240         }
 241         catch(exception& e) {
 242                 m->errorOut(e, "AmovaCommand", "execute");
 243                 exit(1);
 244         }
 245 }
 246
 247 //**********************************************************************************************************************
 248
 249 double AmovaCommand::runAMOVA(ofstream& AMOVAFile, map<string, vector<int> > groupSampleMap, double alpha) {
 250         try {
 251                 map<string, vector<int> >::iterator it;
 252
 253                 int numGroups = groupSampleMap.size();
 254                 int totalNumSamples = 0;
 255
 256                 for(it = groupSampleMap.begin();it!=groupSampleMap.end();it++){
 257                         totalNumSamples += it->second.size();
 258                 }
 259
 260                 double ssTotalOrig = calcSSTotal(groupSampleMap);
 261                 double ssWithinOrig = calcSSWithin(groupSampleMap);
 262                 double ssAmongOrig = ssTotalOrig - ssWithinOrig;
 263
 264                 double counter = 0;
 265                 for(int i=0;i<iters;i++){
 266                         map<string, vector<int> > randomizedGroup = getRandomizedGroups(groupSampleMap);
 267                         double ssWithinRand = calcSSWithin(randomizedGroup);
 268                         if(ssWithinRand < ssWithinOrig){        counter++;      }
 269                 }
 270
 271                 double pValue = (double)counter / (double) iters;
 272                 string pString = "";
 273                 if(pValue < 1/(double)iters){   pString = '<' + toString(1/(double)iters);      }
 274                 else                                            {       pString = toString(pValue);                                     }
 275
 276
 277                 //print anova table
 278                 it = groupSampleMap.begin();
 279                 AMOVAFile << it->first;
 280                 m->mothurOut(it->first);
 281                 it++;
 282                 for(it;it!=groupSampleMap.end();it++){
 283                         AMOVAFile << '-' << it->first;
 284                         m->mothurOut('-' + it->first);
 285                 }
 286
 287                 AMOVAFile << "\tAmong\tWithin\tTotal" << endl;
 288                 m->mothurOut("\tAmong\tWithin\tTotal\n");
 289
 290                 AMOVAFile << "SS\t" << ssAmongOrig << '\t' << ssWithinOrig << '\t' << ssTotalOrig << endl;
 291                 m->mothurOut("SS\t" + toString(ssAmongOrig) + '\t' + toString(ssWithinOrig) + '\t' + toString(ssTotalOrig) + '\n');
 292
 293                 int dfAmong = numGroups - 1;                            double MSAmong = ssAmongOrig / (double) dfAmong;
 294                 int dfWithin = totalNumSamples - numGroups;     double MSWithin = ssWithinOrig / (double) dfWithin;
 295                 int dfTotal = totalNumSamples - 1;                      double Fs = MSAmong / MSWithin;
 296
 297                 AMOVAFile << "df\t" << dfAmong << '\t' << dfWithin << '\t' << dfTotal << endl;
 298                 m->mothurOut("df\t" + toString(dfAmong) + '\t' + toString(dfWithin) + '\t' + toString(dfTotal) + '\n');
 299
 300                 AMOVAFile << "MS\t" << MSAmong << '\t' << MSWithin << endl << endl;
 301                 m->mothurOut("MS\t" + toString(MSAmong) + '\t' + toString(MSWithin) + "\n\n");
 302
 303                 AMOVAFile << "Fs:\t" << Fs << endl;
 304                 m->mothurOut("Fs:\t" + toString(Fs) + '\n');
 305
 306                 AMOVAFile << "p-value: " << pString;
 307                 m->mothurOut("p-value: " + pString);
 308
 309                 if(pValue < alpha){
 310                         AMOVAFile << "*";
 311                         m->mothurOut("*");
 312                 }
 313                 AMOVAFile << endl << endl;
 314                 m->mothurOutEndLine();m->mothurOutEndLine();
 315
 316                 return pValue;
 317         }
 318         catch(exception& e) {
 319                 m->errorOut(e, "AmovaCommand", "runAMOVA");
 320                 exit(1);
 321         }
 322 }
 323
 324 //**********************************************************************************************************************
 325
 326 map<string, vector<int> > AmovaCommand::getRandomizedGroups(map<string, vector<int> > origMapping){
 327         try{
 328                 vector<int> sampleIndices;
 329                 vector<int> samplesPerGroup;
 330
 331                 map<string, vector<int> >::iterator it;
 332                 for(it=origMapping.begin();it!=origMapping.end();it++){
 333                         vector<int> indices = it->second;
 334                         samplesPerGroup.push_back(indices.size());
 335                         sampleIndices.insert(sampleIndices.end(), indices.begin(), indices.end());
 336                 }
 337
 338                 random_shuffle(sampleIndices.begin(), sampleIndices.end());
 339
 340                 int index = 0;
 341                 map<string, vector<int> > randomizedGroups = origMapping;
 342                 for(it=randomizedGroups.begin();it!=randomizedGroups.end();it++){
 343                         for(int i=0;i<it->second.size();i++){
 344                                 it->second[i] = sampleIndices[index++];
 345                         }
 346                 }
 347
 348                 return randomizedGroups;
 349         }
 350         catch (exception& e) {
 351                 m->errorOut(e, "AmovaCommand", "getRandomizedGroups");
 352                 exit(1);
 353         }
 354 }
 355
 356 //**********************************************************************************************************************
 357
 358 double AmovaCommand::calcSSTotal(map<string, vector<int> >& groupSampleMap) {
 359         try {
 360
 361                 vector<int> indices;
 362                 map<string, vector<int> >::iterator it;
 363                 for(it=groupSampleMap.begin();it!=groupSampleMap.end();it++){
 364                         indices.insert(indices.end(), it->second.begin(), it->second.end());
 365                 }
 366                 sort(indices.begin(), indices.end());
 367
 368                 int numIndices =indices.size();
 369                 double ssTotal = 0.0;
 370
 371                 for(int i=1;i<numIndices;i++){
 372                         int row = indices[i];
 373
 374                         for(int j=0;j<i;j++){
 375                                 ssTotal += distanceMatrix[row][indices[j]];
 376                         }
 377                 }
 378                 ssTotal /= numIndices;
 379
 380                 return ssTotal;
 381         }
 382         catch(exception& e) {
 383                 m->errorOut(e, "AmovaCommand", "calcSSTotal");
 384                 exit(1);
 385         }
 386 }
 387
 388 //**********************************************************************************************************************
 389
 390 double AmovaCommand::calcSSWithin(map<string, vector<int> >& groupSampleMap) {
 391         try {
 392
 393                 double ssWithin = 0.0;
 394
 395                 map<string, vector<int> >::iterator it;
 396                 for(it=groupSampleMap.begin();it!=groupSampleMap.end();it++){
 397
 398                         double withinGroup = 0;
 399
 400                         vector<int> samples = it->second;
 401
 402                         for(int i=0;i<samples.size();i++){
 403                                 int row = samples[i];
 404
 405                                 for(int j=0;j<samples.size();j++){
 406                                         int col = samples[j];
 407
 408                                         if(col < row){
 409                                                 withinGroup += distanceMatrix[row][col];
 410                                         }
 411
 412                                 }
 413                         }
 414
 415                         ssWithin += withinGroup / samples.size();
 416                 }
 417
 418                 return ssWithin;
 419         }
 420         catch(exception& e) {
 421                 m->errorOut(e, "AmovaCommand", "calcSSWithin");
 422                 exit(1);
 423         }
 424 }
 425
 426 //**********************************************************************************************************************