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                         }else { m->setPhylipFile(phylipFileName); }
 128
 129                         //check for required parameters
 130                         designFileName = validParameter.validFile(parameters, "design", true);
 131                         if (designFileName == "not open") { designFileName = ""; 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                         }else { m->setDesignFile(designFileName); }
 138
 139                         string temp = validParameter.validFile(parameters, "iters", false);
 140                         if (temp == "not found") { temp = "1000"; }
 141                         m->mothurConvert(temp, iters);
 142
 143                         temp = validParameter.validFile(parameters, "alpha", false);
 144                         if (temp == "not found") { temp = "0.05"; }
 145                         m->mothurConvert(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                         string group = designMap->getGroup(sampleNames[i]);
 180
 181                         if (group == "not found") {
 182                                 m->mothurOut("[ERROR]: " + sampleNames[i] + " is not in your design file, please correct."); m->mothurOutEndLine(); m->control_pressed = true;
 183                         }else { origGroupSampleMap[group].push_back(i); }
 184                 }
 185                 int numGroups = origGroupSampleMap.size();
 186
 187                 if (m->control_pressed) { delete designMap; return 0; }
 188
 189                 //create a new filename
 190                 ofstream ANOSIMFile;
 191                 string ANOSIMFileName = outputDir + m->getRootName(m->getSimpleName(phylipFileName))  + "anosim";
 192                 m->openOutputFile(ANOSIMFileName, ANOSIMFile);
 193                 outputNames.push_back(ANOSIMFileName); outputTypes["anosim"].push_back(ANOSIMFileName);
 194                 m->mothurOut("\ncomparison\tR-value\tP-value\n");
 195                 ANOSIMFile << "comparison\tR-value\tP-value\n";
 196
 197
 198                 double fullANOSIMPValue = runANOSIM(ANOSIMFile, distanceMatrix, origGroupSampleMap, experimentwiseAlpha);
 199
 200
 201                 if(fullANOSIMPValue <= experimentwiseAlpha && numGroups > 2){
 202
 203                         int numCombos = numGroups * (numGroups-1) / 2;
 204                         double pairwiseAlpha = experimentwiseAlpha / (double) numCombos;
 205
 206                         for(map<string, vector<int> >::iterator itA=origGroupSampleMap.begin();itA!=origGroupSampleMap.end();itA++){
 207                                 map<string, vector<int> >::iterator itB = itA;
 208                                 itB++;
 209                                 for(itB;itB!=origGroupSampleMap.end();itB++){
 210
 211                                         map<string, vector<int> > subGroupSampleMap;
 212
 213                                         subGroupSampleMap[itA->first] = itA->second;    string groupA = itA->first;
 214                                         subGroupSampleMap[itB->first] = itB->second;    string groupB = itB->first;
 215
 216                                         vector<int> subIndices;
 217                                         for(map<string, vector<int> >::iterator it=subGroupSampleMap.begin();it!=subGroupSampleMap.end();it++){
 218                                                 subIndices.insert(subIndices.end(), it->second.begin(), it->second.end());
 219                                         }
 220                                         int subNumSamples = subIndices.size();
 221
 222                                         sort(subIndices.begin(), subIndices.end());
 223
 224                                         vector<vector<double> > subDistMatrix(distanceMatrix.size());
 225                                         for(int i=0;i<distanceMatrix.size();i++){
 226                                                 subDistMatrix[i].assign(distanceMatrix.size(), -1);
 227                                         }
 228
 229                                         for(int i=0;i<subNumSamples;i++){
 230                                                 for(int j=0;j<i;j++){
 231                                                         subDistMatrix[subIndices[i]][subIndices[j]] = distanceMatrix[subIndices[i]][subIndices[j]];
 232                                                 }
 233                                         }
 234
 235                                         runANOSIM(ANOSIMFile, subDistMatrix, subGroupSampleMap, pairwiseAlpha);
 236
 237                                 }
 238                         }
 239
 240                         m->mothurOut("\nExperiment-wise error rate: " + toString(experimentwiseAlpha) + '\n');
 241                         m->mothurOut("Pair-wise error rate (Bonferroni): " + toString(pairwiseAlpha) + '\n');
 242                 }
 243                 else{
 244                         m->mothurOut("\nExperiment-wise error rate: " + toString(experimentwiseAlpha) + '\n');
 245                 }
 246                 m->mothurOut("If you have borderline P-values, you should try increasing the number of iterations\n");
 247                 ANOSIMFile.close();
 248
 249
 250                 delete designMap;
 251
 252                 m->mothurOutEndLine();
 253                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 254                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 255                 m->mothurOutEndLine();
 256
 257                 return 0;
 258         }
 259         catch(exception& e) {
 260                 m->errorOut(e, "AnosimCommand", "execute");
 261                 exit(1);
 262         }
 263 }
 264 //**********************************************************************************************************************
 265
 266 double AnosimCommand::runANOSIM(ofstream& ANOSIMFile, vector<vector<double> > dMatrix, map<string, vector<int> > groupSampleMap, double alpha) {
 267         try {
 268
 269
 270                 vector<vector<double> > rankMatrix = convertToRanks(dMatrix);
 271                 double RValue = calcR(rankMatrix, groupSampleMap);
 272
 273                 int pCount = 0;
 274                 for(int i=0;i<iters;i++){
 275                         map<string, vector<int> > randGroupSampleMap = getRandomizedGroups(groupSampleMap);
 276                         double RValueRand = calcR(rankMatrix, randGroupSampleMap);
 277                         if(RValue <= RValueRand){       pCount++;       }
 278                 }
 279
 280                 double pValue = (double)pCount / (double) iters;
 281                 string pString = "";
 282                 if(pValue < 1/(double)iters){   pString = '<' + toString(1/(double)iters);      }
 283                 else                                            {       pString = toString(pValue);                                     }
 284
 285
 286                 map<string, vector<int> >::iterator it=groupSampleMap.begin();
 287                 m->mothurOut(it->first);
 288                 ANOSIMFile << it->first;
 289                 it++;
 290                 for(it;it!=groupSampleMap.end();it++){
 291                         m->mothurOut('-' + it->first);
 292                         ANOSIMFile << '-' << it->first;
 293
 294                 }
 295                 m->mothurOut('\t' + toString(RValue) + '\t' + pString);
 296                 ANOSIMFile << '\t' << RValue << '\t' << pString;
 297
 298                 if(pValue < alpha){
 299                         ANOSIMFile << "*";
 300                         m->mothurOut("*");
 301                 }
 302                 ANOSIMFile << endl;
 303                 m->mothurOutEndLine();
 304
 305                 return pValue;
 306         }
 307         catch(exception& e) {
 308                 m->errorOut(e, "AnosimCommand", "calcAnisom");
 309                 exit(1);
 310         }
 311 }
 312
 313 //**********************************************************************************************************************
 314
 315 double AnosimCommand::calcR(vector<vector<double> > rankMatrix, map<string, vector<int> > groupSampleMap){
 316         try {
 317
 318                 int numSamples = 0;
 319                 for(map<string, vector<int> >::iterator it=groupSampleMap.begin();it!=groupSampleMap.end();it++){
 320                         numSamples += it->second.size();
 321                 }
 322
 323
 324                 double within = 0.0;
 325                 int numWithinComps = 0;
 326
 327                 for(map<string, vector<int> >::iterator it=groupSampleMap.begin();it!=groupSampleMap.end();it++){
 328                         vector<int> indices = it->second;
 329                         for(int i=0;i<indices.size();i++){
 330                                 for(int j=0;j<i;j++){
 331                                         if(indices[i] > indices[j])     {       within += rankMatrix[indices[i]][indices[j]];   }
 332                                         else                                            {       within += rankMatrix[indices[j]][indices[i]];   }
 333                                         numWithinComps++;
 334                                 }
 335                         }
 336                 }
 337
 338                 within /= (float) numWithinComps;
 339
 340                 double between = 0.0;
 341                 int numBetweenComps = 0;
 342
 343                 map<string, vector<int> >::iterator itB;
 344
 345                 for(map<string, vector<int> >::iterator itA=groupSampleMap.begin();itA!=groupSampleMap.end();itA++){
 346
 347                         for(int i=0;i<itA->second.size();i++){
 348                                 int A = itA->second[i];
 349                                 map<string, vector<int> >::iterator itB = itA;
 350                                 itB++;
 351                                 for(itB;itB!=groupSampleMap.end();itB++){
 352                                         for(int j=0;j<itB->second.size();j++){
 353                                                 int B = itB->second[j];
 354                                                 if(A>B) {       between += rankMatrix[A][B];    }
 355                                                 else    {       between += rankMatrix[B][A];    }
 356                                                 numBetweenComps++;
 357                                         }
 358                                 }
 359
 360                         }
 361                 }
 362
 363
 364                 between /= (float) numBetweenComps;
 365
 366                 double Rvalue = (between - within)/(numSamples * (numSamples-1) / 4.0);
 367
 368                 return Rvalue;
 369         }
 370         catch(exception& e) {
 371                 m->errorOut(e, "AnosimCommand", "calcWithinBetween");
 372                 exit(1);
 373         }
 374 }
 375
 376 //**********************************************************************************************************************
 377
 378 vector<vector<double> > AnosimCommand::convertToRanks(vector<vector<double> > dist) {
 379         try {
 380                 vector<seqDist> cells;
 381                 vector<vector<double> > ranks = dist;
 382
 383                 for (int i = 0; i < dist.size(); i++) {
 384                         for (int j = 0; j < i; j++) {
 385                                 if(dist[i][j] != -1){
 386                                         seqDist member(i, j, dist[i][j]);
 387                                         cells.push_back(member);
 388                                 }
 389                         }
 390                 }
 391
 392
 393                 //sort distances
 394                 sort(cells.begin(), cells.end(), compareSequenceDistance);
 395
 396                 //find ranks of distances
 397                 int index = 0;
 398                 int indexSum = 0;
 399                 for(int i=0;i<cells.size()-1;i++){
 400
 401                         index = i;
 402                         indexSum = i + 1;
 403                         while(dist[cells[index].seq1][cells[index].seq2] == dist[cells[index+1].seq1][cells[index+1].seq2]){
 404                                 index++;
 405                                 indexSum += index + 1;
 406                         }
 407
 408                         if(index == i){
 409                                 ranks[cells[i].seq1][cells[i].seq2] = i+1;
 410                         }
 411                         else{
 412                                 double aveIndex = (double)indexSum / (double)(index - i + 1);
 413                                 for(int j=i;j<=index;j++){
 414                                         ranks[cells[j].seq1][cells[j].seq2] = aveIndex;
 415                                 }
 416                                 i = index;
 417                         }
 418                 }
 419
 420                 if(indexSum == cells.size() - 1){
 421                         ranks[cells[cells.size()-1].seq1][cells[cells.size()-1].seq2] = indexSum + 1;
 422                 }
 423
 424                 return ranks;
 425         }
 426         catch(exception& e) {
 427                 m->errorOut(e, "AnosimCommand", "convertToRanks");
 428                 exit(1);
 429         }
 430 }
 431
 432 //**********************************************************************************************************************
 433
 434 map<string, vector<int> > AnosimCommand::getRandomizedGroups(map<string, vector<int> > origMapping){
 435         try{
 436                 vector<int> sampleIndices;
 437                 vector<int> samplesPerGroup;
 438
 439                 map<string, vector<int> >::iterator it;
 440                 for(it=origMapping.begin();it!=origMapping.end();it++){
 441                         vector<int> indices = it->second;
 442                         samplesPerGroup.push_back(indices.size());
 443                         sampleIndices.insert(sampleIndices.end(), indices.begin(), indices.end());
 444                 }
 445
 446                 random_shuffle(sampleIndices.begin(), sampleIndices.end());
 447
 448                 int index = 0;
 449                 map<string, vector<int> > randomizedGroups = origMapping;
 450                 for(it=randomizedGroups.begin();it!=randomizedGroups.end();it++){
 451                         for(int i=0;i<it->second.size();i++){
 452                                 it->second[i] = sampleIndices[index++];
 453                         }
 454                 }
 455
 456                 return randomizedGroups;
 457         }
 458         catch (exception& e) {
 459                 m->errorOut(e, "AnosimCommand", "randomizeGroups");
 460                 exit(1);
 461         }
 462 }
 463
 464 //**********************************************************************************************************************
 465
 466
 467