anosimcommand.cpp

   1 /*
   2  *  anosimcommand.cpp
   3  *  mothur
   4  *
   5  *  Created by westcott on 2/14/11.
   6  *  Copyright 2011 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "anosimcommand.h"
  11 #include "inputdata.h"
  12 #include "readphylipvector.h"
  13
  14 //**********************************************************************************************************************
  15 vector<string> AnosimCommand::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, "AnosimCommand", "getValidParameters");
  23                 exit(1);
  24         }
  25 }
  26 //**********************************************************************************************************************
  27 AnosimCommand::AnosimCommand(){
  28         try {
  29                 abort = true; calledHelp = true;
  30                 vector<string> tempOutNames;
  31                 outputTypes["anosim"] = tempOutNames;
  32         }
  33         catch(exception& e) {
  34                 m->errorOut(e, "AnosimCommand", "AnosimCommand");
  35                 exit(1);
  36         }
  37 }
  38 //**********************************************************************************************************************
  39 vector<string> AnosimCommand::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, "AnosimCommand", "getRequiredParameters");
  47                 exit(1);
  48         }
  49 }
  50 //**********************************************************************************************************************
  51 vector<string> AnosimCommand::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, "AnosimCommand", "getRequiredFiles");
  59                 exit(1);
  60         }
  61 }
  62 //**********************************************************************************************************************
  63
  64 AnosimCommand::AnosimCommand(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[] =  {"outputdir","iters","phylip","design", "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["anosim"] = 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         }
 139         catch(exception& e) {
 140                 m->errorOut(e, "AnosimCommand", "AnosimCommand");
 141                 exit(1);
 142         }
 143 }
 144
 145 //**********************************************************************************************************************
 146
 147 void AnosimCommand::help(){
 148         try {
 149                 m->mothurOut("Referenced: Clarke, K. R. (1993). Non-parametric multivariate analysis of changes in community structure.   _Australian Journal of Ecology_ 18, 117-143.\n");
 150                 m->mothurOut("The anosim command outputs a .anosim file. \n");
 151                 m->mothurOut("The anosim command parameters are phylip, iters, and alpha.  The phylip and design parameters are required.\n");
 152                 m->mothurOut("The design parameter allows you to assign your samples to groups when you are running anosim. It is required. \n");
 153                 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");
 154                 m->mothurOut("The iters parameter allows you to set number of randomization for the P value.  The default is 1000. \n");
 155                 m->mothurOut("The anosim command should be in the following format: anosim(phylip=file.dist, design=file.design).\n");
 156                 m->mothurOut("Note: No spaces between parameter labels (i.e. iters), '=' and parameters (i.e. 1000).\n\n");
 157
 158         }
 159         catch(exception& e) {
 160                 m->errorOut(e, "AnosimCommand", "help");
 161                 exit(1);
 162         }
 163 }
 164
 165 //**********************************************************************************************************************
 166
 167 AnosimCommand::~AnosimCommand(){}
 168
 169 //**********************************************************************************************************************
 170
 171 int AnosimCommand::execute(){
 172         try {
 173
 174                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 175
 176                 //read design file
 177                 designMap = new GroupMap(designFileName);
 178                 designMap->readDesignMap();
 179
 180                 if (outputDir == "") { outputDir = m->hasPath(phylipFileName); }
 181
 182                 //read in distance matrix and square it
 183                 ReadPhylipVector readMatrix(phylipFileName);
 184                 vector<string> sampleNames = readMatrix.read(distanceMatrix);
 185
 186                 for(int i=0;i<distanceMatrix.size();i++){
 187                         for(int j=0;j<i;j++){
 188                                 distanceMatrix[i][j] *= distanceMatrix[i][j];
 189                         }
 190                 }
 191
 192                 //link designMap to rows/columns in distance matrix
 193                 map<string, vector<int> > origGroupSampleMap;
 194                 for(int i=0;i<sampleNames.size();i++){
 195                         origGroupSampleMap[designMap->getGroup(sampleNames[i])].push_back(i);
 196                 }
 197                 int numGroups = origGroupSampleMap.size();
 198
 199                 //create a new filename
 200                 ofstream ANOSIMFile;
 201                 string ANOSIMFileName = outputDir + m->getRootName(m->getSimpleName(phylipFileName))  + "anosim";
 202                 m->openOutputFile(ANOSIMFileName, ANOSIMFile);
 203                 outputNames.push_back(ANOSIMFileName); outputTypes["anosim"].push_back(ANOSIMFileName);
 204                 m->mothurOut("\ncomparison\tR-value\tP-value\n");
 205                 ANOSIMFile << "comparison\tR-value\tP-value\n";
 206
 207
 208                 double fullANOSIMPValue = runANOSIM(ANOSIMFile, distanceMatrix, origGroupSampleMap, experimentwiseAlpha);
 209
 210
 211                 if(fullANOSIMPValue <= experimentwiseAlpha && numGroups > 2){
 212
 213                         int numCombos = numGroups * (numGroups-1) / 2;
 214                         double pairwiseAlpha = experimentwiseAlpha / (double) numCombos;
 215
 216                         for(map<string, vector<int> >::iterator itA=origGroupSampleMap.begin();itA!=origGroupSampleMap.end();itA++){
 217                                 map<string, vector<int> >::iterator itB = itA;
 218                                 itB++;
 219                                 for(itB;itB!=origGroupSampleMap.end();itB++){
 220
 221                                         map<string, vector<int> > subGroupSampleMap;
 222
 223                                         subGroupSampleMap[itA->first] = itA->second;    string groupA = itA->first;
 224                                         subGroupSampleMap[itB->first] = itB->second;    string groupB = itB->first;
 225
 226                                         vector<int> subIndices;
 227                                         for(map<string, vector<int> >::iterator it=subGroupSampleMap.begin();it!=subGroupSampleMap.end();it++){
 228                                                 subIndices.insert(subIndices.end(), it->second.begin(), it->second.end());
 229                                         }
 230                                         int subNumSamples = subIndices.size();
 231
 232                                         sort(subIndices.begin(), subIndices.end());
 233
 234                                         vector<vector<double> > subDistMatrix(distanceMatrix.size());
 235                                         for(int i=0;i<distanceMatrix.size();i++){
 236                                                 subDistMatrix[i].assign(distanceMatrix.size(), -1);
 237                                         }
 238
 239                                         for(int i=0;i<subNumSamples;i++){
 240                                                 for(int j=0;j<i;j++){
 241                                                         subDistMatrix[subIndices[i]][subIndices[j]] = distanceMatrix[subIndices[i]][subIndices[j]];
 242                                                 }
 243                                         }
 244
 245                                         runANOSIM(ANOSIMFile, subDistMatrix, subGroupSampleMap, pairwiseAlpha);
 246
 247                                 }
 248                         }
 249
 250                         m->mothurOut("\nExperiment-wise error rate: " + toString(experimentwiseAlpha) + '\n');
 251                         m->mothurOut("Pair-wise error rate (Bonferroni): " + toString(pairwiseAlpha) + '\n');
 252                 }
 253                 else{
 254                         m->mothurOut("\nExperiment-wise error rate: " + toString(experimentwiseAlpha) + '\n');
 255                 }
 256                 m->mothurOut("If you have borderline P-values, you should try increasing the number of iterations\n");
 257                 ANOSIMFile.close();
 258
 259
 260                 delete designMap;
 261
 262                 m->mothurOutEndLine();
 263                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 264                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 265                 m->mothurOutEndLine();
 266
 267                 return 0;
 268         }
 269         catch(exception& e) {
 270                 m->errorOut(e, "AnosimCommand", "execute");
 271                 exit(1);
 272         }
 273 }
 274 //**********************************************************************************************************************
 275
 276 double AnosimCommand::runANOSIM(ofstream& ANOSIMFile, vector<vector<double> > dMatrix, map<string, vector<int> > groupSampleMap, double alpha) {
 277         try {
 278
 279
 280                 vector<vector<double> > rankMatrix = convertToRanks(dMatrix);
 281                 double RValue = calcR(rankMatrix, groupSampleMap);
 282
 283                 int pCount = 0;
 284                 for(int i=0;i<iters;i++){
 285                         map<string, vector<int> > randGroupSampleMap = getRandomizedGroups(groupSampleMap);
 286                         double RValueRand = calcR(rankMatrix, randGroupSampleMap);
 287                         if(RValue <= RValueRand){       pCount++;       }
 288                 }
 289
 290                 double pValue = (double)pCount / (double) iters;
 291                 string pString = "";
 292                 if(pValue < 1/(double)iters){   pString = '<' + toString(1/(double)iters);      }
 293                 else                                            {       pString = toString(pValue);                                     }
 294
 295
 296                 map<string, vector<int> >::iterator it=groupSampleMap.begin();
 297                 m->mothurOut(it->first);
 298                 ANOSIMFile << it->first;
 299                 it++;
 300                 for(it;it!=groupSampleMap.end();it++){
 301                         m->mothurOut('-' + it->first);
 302                         ANOSIMFile << '-' << it->first;
 303
 304                 }
 305                 m->mothurOut('\t' + toString(RValue) + '\t' + pString);
 306                 ANOSIMFile << '\t' << RValue << '\t' << pString;
 307
 308                 if(pValue < alpha){
 309                         ANOSIMFile << "*";
 310                         m->mothurOut("*");
 311                 }
 312                 ANOSIMFile << endl;
 313                 m->mothurOutEndLine();
 314
 315                 return pValue;
 316         }
 317         catch(exception& e) {
 318                 m->errorOut(e, "AnosimCommand", "calcAnisom");
 319                 exit(1);
 320         }
 321 }
 322
 323 //**********************************************************************************************************************
 324
 325 double AnosimCommand::calcR(vector<vector<double> > rankMatrix, map<string, vector<int> > groupSampleMap){
 326         try {
 327
 328                 int numSamples = 0;
 329                 for(map<string, vector<int> >::iterator it=groupSampleMap.begin();it!=groupSampleMap.end();it++){
 330                         numSamples += it->second.size();
 331                 }
 332
 333
 334                 double within = 0.0;
 335                 int numWithinComps = 0;
 336
 337                 for(map<string, vector<int> >::iterator it=groupSampleMap.begin();it!=groupSampleMap.end();it++){
 338                         vector<int> indices = it->second;
 339                         for(int i=0;i<indices.size();i++){
 340                                 for(int j=0;j<i;j++){
 341                                         if(indices[i] > indices[j])     {       within += rankMatrix[indices[i]][indices[j]];   }
 342                                         else                                            {       within += rankMatrix[indices[j]][indices[i]];   }
 343                                         numWithinComps++;
 344                                 }
 345                         }
 346                 }
 347
 348                 within /= (float) numWithinComps;
 349
 350                 double between = 0.0;
 351                 int numBetweenComps = 0;
 352
 353                 map<string, vector<int> >::iterator itB;
 354
 355                 for(map<string, vector<int> >::iterator itA=groupSampleMap.begin();itA!=groupSampleMap.end();itA++){
 356
 357                         for(int i=0;i<itA->second.size();i++){
 358                                 int A = itA->second[i];
 359                                 map<string, vector<int> >::iterator itB = itA;
 360                                 itB++;
 361                                 for(itB;itB!=groupSampleMap.end();itB++){
 362                                         for(int j=0;j<itB->second.size();j++){
 363                                                 int B = itB->second[j];
 364                                                 if(A>B) {       between += rankMatrix[A][B];    }
 365                                                 else    {       between += rankMatrix[B][A];    }
 366                                                 numBetweenComps++;
 367                                         }
 368                                 }
 369
 370                         }
 371                 }
 372
 373
 374                 between /= (float) numBetweenComps;
 375
 376                 double Rvalue = (between - within)/(numSamples * (numSamples-1) / 4.0);
 377
 378                 return Rvalue;
 379         }
 380         catch(exception& e) {
 381                 m->errorOut(e, "AnosimCommand", "calcWithinBetween");
 382                 exit(1);
 383         }
 384 }
 385
 386 //**********************************************************************************************************************
 387
 388 vector<vector<double> > AnosimCommand::convertToRanks(vector<vector<double> > dist) {
 389         try {
 390                 vector<seqDist> cells;
 391                 vector<vector<double> > ranks = dist;
 392
 393                 for (int i = 0; i < dist.size(); i++) {
 394                         for (int j = 0; j < i; j++) {
 395                                 if(dist[i][j] != -1){
 396                                         seqDist member(i, j, dist[i][j]);
 397                                         cells.push_back(member);
 398                                 }
 399                         }
 400                 }
 401
 402
 403                 //sort distances
 404                 sort(cells.begin(), cells.end(), compareSequenceDistance);
 405
 406                 //find ranks of distances
 407                 int index = 0;
 408                 int indexSum = 0;
 409                 for(int i=0;i<cells.size()-1;i++){
 410
 411                         index = i;
 412                         indexSum = i + 1;
 413                         while(dist[cells[index].seq1][cells[index].seq2] == dist[cells[index+1].seq1][cells[index+1].seq2]){
 414                                 index++;
 415                                 indexSum += index + 1;
 416                         }
 417
 418                         if(index == i){
 419                                 ranks[cells[i].seq1][cells[i].seq2] = i+1;
 420                         }
 421                         else{
 422                                 double aveIndex = (double)indexSum / (double)(index - i + 1);
 423                                 for(int j=i;j<=index;j++){
 424                                         ranks[cells[j].seq1][cells[j].seq2] = aveIndex;
 425                                 }
 426                                 i = index;
 427                         }
 428                 }
 429
 430                 if(indexSum == cells.size() - 1){
 431                         ranks[cells[cells.size()-1].seq1][cells[cells.size()-1].seq2] = indexSum + 1;
 432                 }
 433
 434                 return ranks;
 435         }
 436         catch(exception& e) {
 437                 m->errorOut(e, "AnosimCommand", "convertToRanks");
 438                 exit(1);
 439         }
 440 }
 441
 442 //**********************************************************************************************************************
 443
 444 map<string, vector<int> > AnosimCommand::getRandomizedGroups(map<string, vector<int> > origMapping){
 445         try{
 446                 vector<int> sampleIndices;
 447                 vector<int> samplesPerGroup;
 448
 449                 map<string, vector<int> >::iterator it;
 450                 for(it=origMapping.begin();it!=origMapping.end();it++){
 451                         vector<int> indices = it->second;
 452                         samplesPerGroup.push_back(indices.size());
 453                         sampleIndices.insert(sampleIndices.end(), indices.begin(), indices.end());
 454                 }
 455
 456                 random_shuffle(sampleIndices.begin(), sampleIndices.end());
 457
 458                 int index = 0;
 459                 map<string, vector<int> > randomizedGroups = origMapping;
 460                 for(it=randomizedGroups.begin();it!=randomizedGroups.end();it++){
 461                         for(int i=0;i<it->second.size();i++){
 462                                 it->second[i] = sampleIndices[index++];
 463                         }
 464                 }
 465
 466                 return randomizedGroups;
 467         }
 468         catch (exception& e) {
 469                 m->errorOut(e, "AnosimCommand", "randomizeGroups");
 470                 exit(1);
 471         }
 472 }
 473
 474 //**********************************************************************************************************************
 475
 476
 477