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