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 string AnosimCommand::getOutputFileNameTag(string type, string inputName=""){
  54         try {
  55         string outputFileName = "";
  56                 map<string, vector<string> >::iterator it;
  57
  58         //is this a type this command creates
  59         it = outputTypes.find(type);
  60         if (it == outputTypes.end()) {  m->mothurOut("[ERROR]: this command doesn't create a " + type + " output file.\n"); }
  61         else {
  62             if (type == "anosim") {  outputFileName =  "anosim"; }
  63             else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true;  }
  64         }
  65         return outputFileName;
  66         }
  67         catch(exception& e) {
  68                 m->errorOut(e, "AnosimCommand", "getOutputFileNameTag");
  69                 exit(1);
  70         }
  71 }
  72 //**********************************************************************************************************************
  73 AnosimCommand::AnosimCommand(){
  74         try {
  75                 abort = true; calledHelp = true;
  76                 setParameters();
  77                 vector<string> tempOutNames;
  78                 outputTypes["anosim"] = tempOutNames;
  79         }
  80         catch(exception& e) {
  81                 m->errorOut(e, "AnosimCommand", "AnosimCommand");
  82                 exit(1);
  83         }
  84 }
  85 //**********************************************************************************************************************
  86
  87 AnosimCommand::AnosimCommand(string option) {
  88         try {
  89                 abort = false; calledHelp = false;
  90
  91                 //allow user to run help
  92                 if(option == "help") { help(); abort = true; calledHelp = true; }
  93                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  94
  95                 else {
  96                         vector<string> myArray = setParameters();
  97
  98                         OptionParser parser(option);
  99                         map<string,string> parameters = parser.getParameters();
 100
 101                         ValidParameters validParameter;
 102
 103                         //check to make sure all parameters are valid for command
 104                         map<string,string>::iterator it;
 105                         for (it = parameters.begin(); it != parameters.end(); it++) {
 106                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 107                         }
 108
 109                         //initialize outputTypes
 110                         vector<string> tempOutNames;
 111                         outputTypes["anosim"] = tempOutNames;
 112
 113                         //if the user changes the output directory command factory will send this info to us in the output parameter
 114                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
 115
 116                         //if the user changes the input directory command factory will send this info to us in the output parameter
 117                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 118                         if (inputDir == "not found"){   inputDir = "";          }
 119                         else {
 120                                 string path;
 121                                 it = parameters.find("design");
 122                                 //user has given a template file
 123                                 if(it != parameters.end()){
 124                                         path = m->hasPath(it->second);
 125                                         //if the user has not given a path then, add inputdir. else leave path alone.
 126                                         if (path == "") {       parameters["design"] = inputDir + it->second;           }
 127                                 }
 128
 129                                 it = parameters.find("phylip");
 130                                 //user has given a template file
 131                                 if(it != parameters.end()){
 132                                         path = m->hasPath(it->second);
 133                                         //if the user has not given a path then, add inputdir. else leave path alone.
 134                                         if (path == "") {       parameters["phylip"] = inputDir + it->second;           }
 135                                 }
 136                         }
 137
 138                         phylipFileName = validParameter.validFile(parameters, "phylip", true);
 139                         if (phylipFileName == "not open") { phylipFileName = ""; abort = true; }
 140                         else if (phylipFileName == "not found") {
 141                                 //if there is a current phylip file, use it
 142                                 phylipFileName = m->getPhylipFile();
 143                                 if (phylipFileName != "") { m->mothurOut("Using " + phylipFileName + " as input file for the phylip parameter."); m->mothurOutEndLine(); }
 144                                 else {  m->mothurOut("You have no current phylip file and the phylip parameter is required."); m->mothurOutEndLine(); abort = true; }
 145
 146                         }else { m->setPhylipFile(phylipFileName); }
 147
 148                         //check for required parameters
 149                         designFileName = validParameter.validFile(parameters, "design", true);
 150                         if (designFileName == "not open") { designFileName = ""; abort = true; }
 151                         else if (designFileName == "not found") {
 152                                 //if there is a current design file, use it
 153                                 designFileName = m->getDesignFile();
 154                                 if (designFileName != "") { m->mothurOut("Using " + designFileName + " as input file for the design parameter."); m->mothurOutEndLine(); }
 155                                 else {  m->mothurOut("You have no current design file and the design parameter is required."); m->mothurOutEndLine(); abort = true; }
 156                         }else { m->setDesignFile(designFileName); }
 157
 158                         string temp = validParameter.validFile(parameters, "iters", false);
 159                         if (temp == "not found") { temp = "1000"; }
 160                         m->mothurConvert(temp, iters);
 161
 162                         temp = validParameter.validFile(parameters, "alpha", false);
 163                         if (temp == "not found") { temp = "0.05"; }
 164                         m->mothurConvert(temp, experimentwiseAlpha);
 165                 }
 166
 167         }
 168         catch(exception& e) {
 169                 m->errorOut(e, "AnosimCommand", "AnosimCommand");
 170                 exit(1);
 171         }
 172 }
 173 //**********************************************************************************************************************
 174 int AnosimCommand::execute(){
 175         try {
 176
 177                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 178
 179                 //read design file
 180                 designMap = new GroupMap(designFileName);
 181                 designMap->readDesignMap();
 182
 183                 if (outputDir == "") { outputDir = m->hasPath(phylipFileName); }
 184
 185                 //read in distance matrix and square it
 186                 ReadPhylipVector readMatrix(phylipFileName);
 187                 vector<string> sampleNames = readMatrix.read(distanceMatrix);
 188
 189                 for(int i=0;i<distanceMatrix.size();i++){
 190                         for(int j=0;j<i;j++){
 191                                 distanceMatrix[i][j] *= distanceMatrix[i][j];
 192                         }
 193                 }
 194
 195                 //link designMap to rows/columns in distance matrix
 196                 map<string, vector<int> > origGroupSampleMap;
 197                 for(int i=0;i<sampleNames.size();i++){
 198                         string group = designMap->getGroup(sampleNames[i]);
 199
 200                         if (group == "not found") {
 201                                 m->mothurOut("[ERROR]: " + sampleNames[i] + " is not in your design file, please correct."); m->mothurOutEndLine(); m->control_pressed = true;
 202                         }else { origGroupSampleMap[group].push_back(i); }
 203                 }
 204                 int numGroups = origGroupSampleMap.size();
 205
 206                 if (m->control_pressed) { delete designMap; return 0; }
 207
 208                 //create a new filename
 209                 ofstream ANOSIMFile;
 210                 string ANOSIMFileName = outputDir + m->getRootName(m->getSimpleName(phylipFileName)) + getOutputFileNameTag("anosim");
 211                 m->openOutputFile(ANOSIMFileName, ANOSIMFile);
 212                 outputNames.push_back(ANOSIMFileName); outputTypes["anosim"].push_back(ANOSIMFileName);
 213                 m->mothurOut("\ncomparison\tR-value\tP-value\n");
 214                 ANOSIMFile << "comparison\tR-value\tP-value\n";
 215
 216
 217                 double fullANOSIMPValue = runANOSIM(ANOSIMFile, distanceMatrix, origGroupSampleMap, experimentwiseAlpha);
 218
 219
 220                 if(fullANOSIMPValue <= experimentwiseAlpha && numGroups > 2){
 221
 222                         int numCombos = numGroups * (numGroups-1) / 2;
 223                         double pairwiseAlpha = experimentwiseAlpha / (double) numCombos;
 224
 225                         for(map<string, vector<int> >::iterator itA=origGroupSampleMap.begin();itA!=origGroupSampleMap.end();itA++){
 226                                 map<string, vector<int> >::iterator itB = itA;
 227                                 itB++;
 228                                 for(itB;itB!=origGroupSampleMap.end();itB++){
 229
 230                                         map<string, vector<int> > subGroupSampleMap;
 231
 232                                         subGroupSampleMap[itA->first] = itA->second;    string groupA = itA->first;
 233                                         subGroupSampleMap[itB->first] = itB->second;    string groupB = itB->first;
 234
 235                                         vector<int> subIndices;
 236                                         for(map<string, vector<int> >::iterator it=subGroupSampleMap.begin();it!=subGroupSampleMap.end();it++){
 237                                                 subIndices.insert(subIndices.end(), it->second.begin(), it->second.end());
 238                                         }
 239                                         int subNumSamples = subIndices.size();
 240
 241                                         sort(subIndices.begin(), subIndices.end());
 242
 243                                         vector<vector<double> > subDistMatrix(distanceMatrix.size());
 244                                         for(int i=0;i<distanceMatrix.size();i++){
 245                                                 subDistMatrix[i].assign(distanceMatrix.size(), -1);
 246                                         }
 247
 248                                         for(int i=0;i<subNumSamples;i++){
 249                                                 for(int j=0;j<i;j++){
 250                                                         subDistMatrix[subIndices[i]][subIndices[j]] = distanceMatrix[subIndices[i]][subIndices[j]];
 251                                                 }
 252                                         }
 253
 254                                         runANOSIM(ANOSIMFile, subDistMatrix, subGroupSampleMap, pairwiseAlpha);
 255
 256                                 }
 257                         }
 258
 259                         m->mothurOut("\nExperiment-wise error rate: " + toString(experimentwiseAlpha) + '\n');
 260                         m->mothurOut("Pair-wise error rate (Bonferroni): " + toString(pairwiseAlpha) + '\n');
 261                 }
 262                 else{
 263                         m->mothurOut("\nExperiment-wise error rate: " + toString(experimentwiseAlpha) + '\n');
 264                 }
 265                 m->mothurOut("If you have borderline P-values, you should try increasing the number of iterations\n");
 266                 ANOSIMFile.close();
 267
 268
 269                 delete designMap;
 270
 271                 m->mothurOutEndLine();
 272                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 273                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 274                 m->mothurOutEndLine();
 275
 276                 return 0;
 277         }
 278         catch(exception& e) {
 279                 m->errorOut(e, "AnosimCommand", "execute");
 280                 exit(1);
 281         }
 282 }
 283 //**********************************************************************************************************************
 284
 285 double AnosimCommand::runANOSIM(ofstream& ANOSIMFile, vector<vector<double> > dMatrix, map<string, vector<int> > groupSampleMap, double alpha) {
 286         try {
 287
 288
 289                 vector<vector<double> > rankMatrix = convertToRanks(dMatrix);
 290                 double RValue = calcR(rankMatrix, groupSampleMap);
 291
 292                 int pCount = 0;
 293                 for(int i=0;i<iters;i++){
 294                         map<string, vector<int> > randGroupSampleMap = getRandomizedGroups(groupSampleMap);
 295                         double RValueRand = calcR(rankMatrix, randGroupSampleMap);
 296                         if(RValue <= RValueRand){       pCount++;       }
 297                 }
 298
 299                 double pValue = (double)pCount / (double) iters;
 300                 string pString = "";
 301                 if(pValue < 1/(double)iters){   pString = '<' + toString(1/(double)iters);      }
 302                 else                                            {       pString = toString(pValue);                                     }
 303
 304
 305                 map<string, vector<int> >::iterator it=groupSampleMap.begin();
 306                 m->mothurOut(it->first);
 307                 ANOSIMFile << it->first;
 308                 it++;
 309                 for(it;it!=groupSampleMap.end();it++){
 310                         m->mothurOut('-' + it->first);
 311                         ANOSIMFile << '-' << it->first;
 312
 313                 }
 314                 m->mothurOut('\t' + toString(RValue) + '\t' + pString);
 315                 ANOSIMFile << '\t' << RValue << '\t' << pString;
 316
 317                 if(pValue < alpha){
 318                         ANOSIMFile << "*";
 319                         m->mothurOut("*");
 320                 }
 321                 ANOSIMFile << endl;
 322                 m->mothurOutEndLine();
 323
 324                 return pValue;
 325         }
 326         catch(exception& e) {
 327                 m->errorOut(e, "AnosimCommand", "calcAnisom");
 328                 exit(1);
 329         }
 330 }
 331
 332 //**********************************************************************************************************************
 333
 334 double AnosimCommand::calcR(vector<vector<double> > rankMatrix, map<string, vector<int> > groupSampleMap){
 335         try {
 336
 337                 int numSamples = 0;
 338                 for(map<string, vector<int> >::iterator it=groupSampleMap.begin();it!=groupSampleMap.end();it++){
 339                         numSamples += it->second.size();
 340                 }
 341
 342
 343                 double within = 0.0;
 344                 int numWithinComps = 0;
 345
 346                 for(map<string, vector<int> >::iterator it=groupSampleMap.begin();it!=groupSampleMap.end();it++){
 347                         vector<int> indices = it->second;
 348                         for(int i=0;i<indices.size();i++){
 349                                 for(int j=0;j<i;j++){
 350                                         if(indices[i] > indices[j])     {       within += rankMatrix[indices[i]][indices[j]];   }
 351                                         else                                            {       within += rankMatrix[indices[j]][indices[i]];   }
 352                                         numWithinComps++;
 353                                 }
 354                         }
 355                 }
 356
 357                 within /= (float) numWithinComps;
 358
 359                 double between = 0.0;
 360                 int numBetweenComps = 0;
 361
 362                 map<string, vector<int> >::iterator itB;
 363
 364                 for(map<string, vector<int> >::iterator itA=groupSampleMap.begin();itA!=groupSampleMap.end();itA++){
 365
 366                         for(int i=0;i<itA->second.size();i++){
 367                                 int A = itA->second[i];
 368                                 map<string, vector<int> >::iterator itB = itA;
 369                                 itB++;
 370                                 for(itB;itB!=groupSampleMap.end();itB++){
 371                                         for(int j=0;j<itB->second.size();j++){
 372                                                 int B = itB->second[j];
 373                                                 if(A>B) {       between += rankMatrix[A][B];    }
 374                                                 else    {       between += rankMatrix[B][A];    }
 375                                                 numBetweenComps++;
 376                                         }
 377                                 }
 378
 379                         }
 380                 }
 381
 382
 383                 between /= (float) numBetweenComps;
 384
 385                 double Rvalue = (between - within)/(numSamples * (numSamples-1) / 4.0);
 386
 387                 return Rvalue;
 388         }
 389         catch(exception& e) {
 390                 m->errorOut(e, "AnosimCommand", "calcWithinBetween");
 391                 exit(1);
 392         }
 393 }
 394
 395 //**********************************************************************************************************************
 396
 397 vector<vector<double> > AnosimCommand::convertToRanks(vector<vector<double> > dist) {
 398         try {
 399                 vector<seqDist> cells;
 400                 vector<vector<double> > ranks = dist;
 401
 402                 for (int i = 0; i < dist.size(); i++) {
 403                         for (int j = 0; j < i; j++) {
 404                                 if(dist[i][j] != -1){
 405                                         seqDist member(i, j, dist[i][j]);
 406                                         cells.push_back(member);
 407                                 }
 408                         }
 409                 }
 410
 411
 412                 //sort distances
 413                 sort(cells.begin(), cells.end(), compareSequenceDistance);
 414
 415                 //find ranks of distances
 416                 int index = 0;
 417                 int indexSum = 0;
 418                 for(int i=0;i<cells.size()-1;i++){
 419
 420                         index = i;
 421                         indexSum = i + 1;
 422                         while(dist[cells[index].seq1][cells[index].seq2] == dist[cells[index+1].seq1][cells[index+1].seq2]){
 423                                 index++;
 424                                 indexSum += index + 1;
 425                         }
 426
 427                         if(index == i){
 428                                 ranks[cells[i].seq1][cells[i].seq2] = i+1;
 429                         }
 430                         else{
 431                                 double aveIndex = (double)indexSum / (double)(index - i + 1);
 432                                 for(int j=i;j<=index;j++){
 433                                         ranks[cells[j].seq1][cells[j].seq2] = aveIndex;
 434                                 }
 435                                 i = index;
 436                         }
 437                 }
 438
 439                 if(indexSum == cells.size() - 1){
 440                         ranks[cells[cells.size()-1].seq1][cells[cells.size()-1].seq2] = indexSum + 1;
 441                 }
 442
 443                 return ranks;
 444         }
 445         catch(exception& e) {
 446                 m->errorOut(e, "AnosimCommand", "convertToRanks");
 447                 exit(1);
 448         }
 449 }
 450
 451 //**********************************************************************************************************************
 452
 453 map<string, vector<int> > AnosimCommand::getRandomizedGroups(map<string, vector<int> > origMapping){
 454         try{
 455                 vector<int> sampleIndices;
 456                 vector<int> samplesPerGroup;
 457
 458                 map<string, vector<int> >::iterator it;
 459                 for(it=origMapping.begin();it!=origMapping.end();it++){
 460                         vector<int> indices = it->second;
 461                         samplesPerGroup.push_back(indices.size());
 462                         sampleIndices.insert(sampleIndices.end(), indices.begin(), indices.end());
 463                 }
 464
 465                 random_shuffle(sampleIndices.begin(), sampleIndices.end());
 466
 467                 int index = 0;
 468                 map<string, vector<int> > randomizedGroups = origMapping;
 469                 for(it=randomizedGroups.begin();it!=randomizedGroups.end();it++){
 470                         for(int i=0;i<it->second.size();i++){
 471                                 it->second[i] = sampleIndices[index++];
 472                         }
 473                 }
 474
 475                 return randomizedGroups;
 476         }
 477         catch (exception& e) {
 478                 m->errorOut(e, "AnosimCommand", "randomizeGroups");
 479                 exit(1);
 480         }
 481 }
 482
 483 //**********************************************************************************************************************
 484
 485
 486