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