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::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, "AnosimCommand", "setParameters");
  30                 exit(1);
  31         }
  32 }
  33 //**********************************************************************************************************************
  34 string AnosimCommand::getHelpString(){
  35         try {
  36                 string helpString = "";
  37                 helpString += "Referenced: Clarke, K. R. (1993). Non-parametric multivariate analysis of changes in community structure.   _Australian Journal of Ecology_ 18, 117-143.\n";
  38                 helpString += "The anosim command outputs a .anosim file. \n";
  39                 helpString += "The anosim command parameters are phylip, iters, and alpha.  The phylip and design parameters are required, unless you have valid current files.\n";
  40                 helpString += "The design parameter allows you to assign your samples to groups when you are running anosim. 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 anosim command should be in the following format: anosim(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, "AnosimCommand", "getHelpString");
  49                 exit(1);
  50         }
  51 }
  52
  53 //**********************************************************************************************************************
  54 AnosimCommand::AnosimCommand(){
  55         try {
  56                 abort = true; calledHelp = true;
  57                 setParameters();
  58                 vector<string> tempOutNames;
  59                 outputTypes["anosim"] = tempOutNames;
  60         }
  61         catch(exception& e) {
  62                 m->errorOut(e, "AnosimCommand", "AnosimCommand");
  63                 exit(1);
  64         }
  65 }
  66 //**********************************************************************************************************************
  67
  68 AnosimCommand::AnosimCommand(string option) {
  69         try {
  70                 abort = false; calledHelp = false;
  71
  72                 //allow user to run help
  73                 if(option == "help") { help(); abort = true; calledHelp = true; }
  74                 else if(option == "citation") { citation(); 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["anosim"] = 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, "AnosimCommand", "AnosimCommand");
 151                 exit(1);
 152         }
 153 }
 154 //**********************************************************************************************************************
 155 int AnosimCommand::execute(){
 156         try {
 157
 158                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 159
 160                 //read design file
 161                 designMap = new GroupMap(designFileName);
 162                 designMap->readDesignMap();
 163
 164                 if (outputDir == "") { outputDir = m->hasPath(phylipFileName); }
 165
 166                 //read in distance matrix and square it
 167                 ReadPhylipVector readMatrix(phylipFileName);
 168                 vector<string> sampleNames = readMatrix.read(distanceMatrix);
 169
 170                 for(int i=0;i<distanceMatrix.size();i++){
 171                         for(int j=0;j<i;j++){
 172                                 distanceMatrix[i][j] *= distanceMatrix[i][j];
 173                         }
 174                 }
 175
 176                 //link designMap to rows/columns in distance matrix
 177                 map<string, vector<int> > origGroupSampleMap;
 178                 for(int i=0;i<sampleNames.size();i++){
 179                         origGroupSampleMap[designMap->getGroup(sampleNames[i])].push_back(i);
 180                 }
 181                 int numGroups = origGroupSampleMap.size();
 182
 183                 //create a new filename
 184                 ofstream ANOSIMFile;
 185                 string ANOSIMFileName = outputDir + m->getRootName(m->getSimpleName(phylipFileName))  + "anosim";
 186                 m->openOutputFile(ANOSIMFileName, ANOSIMFile);
 187                 outputNames.push_back(ANOSIMFileName); outputTypes["anosim"].push_back(ANOSIMFileName);
 188                 m->mothurOut("\ncomparison\tR-value\tP-value\n");
 189                 ANOSIMFile << "comparison\tR-value\tP-value\n";
 190
 191
 192                 double fullANOSIMPValue = runANOSIM(ANOSIMFile, distanceMatrix, origGroupSampleMap, experimentwiseAlpha);
 193
 194
 195                 if(fullANOSIMPValue <= experimentwiseAlpha && numGroups > 2){
 196
 197                         int numCombos = numGroups * (numGroups-1) / 2;
 198                         double pairwiseAlpha = experimentwiseAlpha / (double) numCombos;
 199
 200                         for(map<string, vector<int> >::iterator itA=origGroupSampleMap.begin();itA!=origGroupSampleMap.end();itA++){
 201                                 map<string, vector<int> >::iterator itB = itA;
 202                                 itB++;
 203                                 for(itB;itB!=origGroupSampleMap.end();itB++){
 204
 205                                         map<string, vector<int> > subGroupSampleMap;
 206
 207                                         subGroupSampleMap[itA->first] = itA->second;    string groupA = itA->first;
 208                                         subGroupSampleMap[itB->first] = itB->second;    string groupB = itB->first;
 209
 210                                         vector<int> subIndices;
 211                                         for(map<string, vector<int> >::iterator it=subGroupSampleMap.begin();it!=subGroupSampleMap.end();it++){
 212                                                 subIndices.insert(subIndices.end(), it->second.begin(), it->second.end());
 213                                         }
 214                                         int subNumSamples = subIndices.size();
 215
 216                                         sort(subIndices.begin(), subIndices.end());
 217
 218                                         vector<vector<double> > subDistMatrix(distanceMatrix.size());
 219                                         for(int i=0;i<distanceMatrix.size();i++){
 220                                                 subDistMatrix[i].assign(distanceMatrix.size(), -1);
 221                                         }
 222
 223                                         for(int i=0;i<subNumSamples;i++){
 224                                                 for(int j=0;j<i;j++){
 225                                                         subDistMatrix[subIndices[i]][subIndices[j]] = distanceMatrix[subIndices[i]][subIndices[j]];
 226                                                 }
 227                                         }
 228
 229                                         runANOSIM(ANOSIMFile, subDistMatrix, subGroupSampleMap, pairwiseAlpha);
 230
 231                                 }
 232                         }
 233
 234                         m->mothurOut("\nExperiment-wise error rate: " + toString(experimentwiseAlpha) + '\n');
 235                         m->mothurOut("Pair-wise error rate (Bonferroni): " + toString(pairwiseAlpha) + '\n');
 236                 }
 237                 else{
 238                         m->mothurOut("\nExperiment-wise error rate: " + toString(experimentwiseAlpha) + '\n');
 239                 }
 240                 m->mothurOut("If you have borderline P-values, you should try increasing the number of iterations\n");
 241                 ANOSIMFile.close();
 242
 243
 244                 delete designMap;
 245
 246                 m->mothurOutEndLine();
 247                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 248                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 249                 m->mothurOutEndLine();
 250
 251                 return 0;
 252         }
 253         catch(exception& e) {
 254                 m->errorOut(e, "AnosimCommand", "execute");
 255                 exit(1);
 256         }
 257 }
 258 //**********************************************************************************************************************
 259
 260 double AnosimCommand::runANOSIM(ofstream& ANOSIMFile, vector<vector<double> > dMatrix, map<string, vector<int> > groupSampleMap, double alpha) {
 261         try {
 262
 263
 264                 vector<vector<double> > rankMatrix = convertToRanks(dMatrix);
 265                 double RValue = calcR(rankMatrix, groupSampleMap);
 266
 267                 int pCount = 0;
 268                 for(int i=0;i<iters;i++){
 269                         map<string, vector<int> > randGroupSampleMap = getRandomizedGroups(groupSampleMap);
 270                         double RValueRand = calcR(rankMatrix, randGroupSampleMap);
 271                         if(RValue <= RValueRand){       pCount++;       }
 272                 }
 273
 274                 double pValue = (double)pCount / (double) iters;
 275                 string pString = "";
 276                 if(pValue < 1/(double)iters){   pString = '<' + toString(1/(double)iters);      }
 277                 else                                            {       pString = toString(pValue);                                     }
 278
 279
 280                 map<string, vector<int> >::iterator it=groupSampleMap.begin();
 281                 m->mothurOut(it->first);
 282                 ANOSIMFile << it->first;
 283                 it++;
 284                 for(it;it!=groupSampleMap.end();it++){
 285                         m->mothurOut('-' + it->first);
 286                         ANOSIMFile << '-' << it->first;
 287
 288                 }
 289                 m->mothurOut('\t' + toString(RValue) + '\t' + pString);
 290                 ANOSIMFile << '\t' << RValue << '\t' << pString;
 291
 292                 if(pValue < alpha){
 293                         ANOSIMFile << "*";
 294                         m->mothurOut("*");
 295                 }
 296                 ANOSIMFile << endl;
 297                 m->mothurOutEndLine();
 298
 299                 return pValue;
 300         }
 301         catch(exception& e) {
 302                 m->errorOut(e, "AnosimCommand", "calcAnisom");
 303                 exit(1);
 304         }
 305 }
 306
 307 //**********************************************************************************************************************
 308
 309 double AnosimCommand::calcR(vector<vector<double> > rankMatrix, map<string, vector<int> > groupSampleMap){
 310         try {
 311
 312                 int numSamples = 0;
 313                 for(map<string, vector<int> >::iterator it=groupSampleMap.begin();it!=groupSampleMap.end();it++){
 314                         numSamples += it->second.size();
 315                 }
 316
 317
 318                 double within = 0.0;
 319                 int numWithinComps = 0;
 320
 321                 for(map<string, vector<int> >::iterator it=groupSampleMap.begin();it!=groupSampleMap.end();it++){
 322                         vector<int> indices = it->second;
 323                         for(int i=0;i<indices.size();i++){
 324                                 for(int j=0;j<i;j++){
 325                                         if(indices[i] > indices[j])     {       within += rankMatrix[indices[i]][indices[j]];   }
 326                                         else                                            {       within += rankMatrix[indices[j]][indices[i]];   }
 327                                         numWithinComps++;
 328                                 }
 329                         }
 330                 }
 331
 332                 within /= (float) numWithinComps;
 333
 334                 double between = 0.0;
 335                 int numBetweenComps = 0;
 336
 337                 map<string, vector<int> >::iterator itB;
 338
 339                 for(map<string, vector<int> >::iterator itA=groupSampleMap.begin();itA!=groupSampleMap.end();itA++){
 340
 341                         for(int i=0;i<itA->second.size();i++){
 342                                 int A = itA->second[i];
 343                                 map<string, vector<int> >::iterator itB = itA;
 344                                 itB++;
 345                                 for(itB;itB!=groupSampleMap.end();itB++){
 346                                         for(int j=0;j<itB->second.size();j++){
 347                                                 int B = itB->second[j];
 348                                                 if(A>B) {       between += rankMatrix[A][B];    }
 349                                                 else    {       between += rankMatrix[B][A];    }
 350                                                 numBetweenComps++;
 351                                         }
 352                                 }
 353
 354                         }
 355                 }
 356
 357
 358                 between /= (float) numBetweenComps;
 359
 360                 double Rvalue = (between - within)/(numSamples * (numSamples-1) / 4.0);
 361
 362                 return Rvalue;
 363         }
 364         catch(exception& e) {
 365                 m->errorOut(e, "AnosimCommand", "calcWithinBetween");
 366                 exit(1);
 367         }
 368 }
 369
 370 //**********************************************************************************************************************
 371
 372 vector<vector<double> > AnosimCommand::convertToRanks(vector<vector<double> > dist) {
 373         try {
 374                 vector<seqDist> cells;
 375                 vector<vector<double> > ranks = dist;
 376
 377                 for (int i = 0; i < dist.size(); i++) {
 378                         for (int j = 0; j < i; j++) {
 379                                 if(dist[i][j] != -1){
 380                                         seqDist member(i, j, dist[i][j]);
 381                                         cells.push_back(member);
 382                                 }
 383                         }
 384                 }
 385
 386
 387                 //sort distances
 388                 sort(cells.begin(), cells.end(), compareSequenceDistance);
 389
 390                 //find ranks of distances
 391                 int index = 0;
 392                 int indexSum = 0;
 393                 for(int i=0;i<cells.size()-1;i++){
 394
 395                         index = i;
 396                         indexSum = i + 1;
 397                         while(dist[cells[index].seq1][cells[index].seq2] == dist[cells[index+1].seq1][cells[index+1].seq2]){
 398                                 index++;
 399                                 indexSum += index + 1;
 400                         }
 401
 402                         if(index == i){
 403                                 ranks[cells[i].seq1][cells[i].seq2] = i+1;
 404                         }
 405                         else{
 406                                 double aveIndex = (double)indexSum / (double)(index - i + 1);
 407                                 for(int j=i;j<=index;j++){
 408                                         ranks[cells[j].seq1][cells[j].seq2] = aveIndex;
 409                                 }
 410                                 i = index;
 411                         }
 412                 }
 413
 414                 if(indexSum == cells.size() - 1){
 415                         ranks[cells[cells.size()-1].seq1][cells[cells.size()-1].seq2] = indexSum + 1;
 416                 }
 417
 418                 return ranks;
 419         }
 420         catch(exception& e) {
 421                 m->errorOut(e, "AnosimCommand", "convertToRanks");
 422                 exit(1);
 423         }
 424 }
 425
 426 //**********************************************************************************************************************
 427
 428 map<string, vector<int> > AnosimCommand::getRandomizedGroups(map<string, vector<int> > origMapping){
 429         try{
 430                 vector<int> sampleIndices;
 431                 vector<int> samplesPerGroup;
 432
 433                 map<string, vector<int> >::iterator it;
 434                 for(it=origMapping.begin();it!=origMapping.end();it++){
 435                         vector<int> indices = it->second;
 436                         samplesPerGroup.push_back(indices.size());
 437                         sampleIndices.insert(sampleIndices.end(), indices.begin(), indices.end());
 438                 }
 439
 440                 random_shuffle(sampleIndices.begin(), sampleIndices.end());
 441
 442                 int index = 0;
 443                 map<string, vector<int> > randomizedGroups = origMapping;
 444                 for(it=randomizedGroups.begin();it!=randomizedGroups.end();it++){
 445                         for(int i=0;i<it->second.size();i++){
 446                                 it->second[i] = sampleIndices[index++];
 447                         }
 448                 }
 449
 450                 return randomizedGroups;
 451         }
 452         catch (exception& e) {
 453                 m->errorOut(e, "AnosimCommand", "randomizeGroups");
 454                 exit(1);
 455         }
 456 }
 457
 458 //**********************************************************************************************************************
 459
 460
 461