cooccurrencecommand.cpp

   1 /*
   2  *  cooccurrencecommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by kiverson on 1/2/12.
   6  *  Copyright 2012 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "cooccurrencecommand.h"
  11
  12 //**********************************************************************************************************************
  13 vector<string> CooccurrenceCommand::setParameters() {
  14         try {
  15                 CommandParameter pshared("shared", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pshared);
  16                 CommandParameter pmetric("metric", "Multiple", "cscore-checker-combo-vratio", "cscore", "", "", "",false,false); parameters.push_back(pmetric);
  17                 CommandParameter pmatrix("matrixmodel", "Multiple", "sim1-sim2-sim3-sim4-sim5-sim6-sim7-sim8-sim9", "sim2", "", "", "",false,false); parameters.push_back(pmatrix);
  18         CommandParameter pruns("iters", "Number", "", "1000", "", "", "",false,false); parameters.push_back(pruns);
  19                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  20                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  21                 CommandParameter plabel("label", "String", "", "", "", "", "",false,false); parameters.push_back(plabel);
  22         CommandParameter pgroups("groups", "String", "", "", "", "", "",false,false); parameters.push_back(pgroups);
  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, "CooccurrenceCommand", "setParameters");
  30                 exit(1);
  31         }
  32 }
  33 //**********************************************************************************************************************
  34 string CooccurrenceCommand::getHelpString(){
  35         try {
  36                 string helpString = "The cooccurrence command calculates four metrics and tests their significance to assess whether presence-absence patterns are different than what one would expect by chance.";
  37         helpString += "The cooccurrence command parameters are shared, metric, matrixmodel, iters, label and groups.";
  38         helpString += "The matrixmodel parameter options are sim1, sim2, sim3, sim4, sim5, sim6, sim7, sim8 and sim9. Default=sim2";
  39         helpString += "The metric parameter options are cscore, checker, combo and vratio. Default=cscore";
  40         helpString += "The label parameter is used to analyze specific labels in your input.\n";
  41                 helpString += "The groups parameter allows you to specify which of the groups you would like analyzed.\n";
  42         helpString += "The cooccurrence command should be in the following format: \n";
  43                 helpString += "cooccurrence(shared=yourSharedFile) \n";
  44                 helpString += "Example cooccurrence(shared=final.an.shared).\n";
  45                 helpString += "Note: No spaces between parameter labels (i.e. shared), '=' and parameters (i.e.yourShared).\n";
  46                 return helpString;
  47         }
  48         catch(exception& e) {
  49                 m->errorOut(e, "CooccurrenceCommand", "getHelpString");
  50                 exit(1);
  51         }
  52 }
  53 //**********************************************************************************************************************
  54 CooccurrenceCommand::CooccurrenceCommand(){
  55         try {
  56                 abort = true; calledHelp = true;
  57                 setParameters();
  58         vector<string> tempOutNames;
  59                 outputTypes["summary"] = tempOutNames;
  60
  61         }
  62         catch(exception& e) {
  63                 m->errorOut(e, "CooccurrenceCommand", "CooccurrenceCommand");
  64                 exit(1);
  65         }
  66 }
  67 //**********************************************************************************************************************
  68 CooccurrenceCommand::CooccurrenceCommand(string option) {
  69         try {
  70                 abort = false; calledHelp = false;
  71                 allLines = 1;
  72
  73                 //allow user to run help
  74                 if(option == "help") { help(); abort = true; calledHelp = true; }
  75                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  76
  77                 else {
  78                         vector<string> myArray = setParameters();
  79
  80                         OptionParser parser(option);
  81                         map<string,string> parameters = parser.getParameters();
  82                         map<string,string>::iterator it;
  83
  84                         ValidParameters validParameter;
  85
  86                         //check to make sure all parameters are valid for command
  87                         for (it = parameters.begin(); it != parameters.end(); it++) {
  88                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  89                         }
  90
  91
  92                         //if the user changes the input directory command factory will send this info to us in the output parameter
  93                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  94                         if (inputDir == "not found"){   inputDir = "";          }
  95                         else {
  96                                 string path;
  97                                 it = parameters.find("shared");
  98                                 //user has given a template file
  99                                 if(it != parameters.end()){
 100                                         path = m->hasPath(it->second);
 101                                         //if the user has not given a path then, add inputdir. else leave path alone.
 102                                         if (path == "") {       parameters["shared"] = inputDir + it->second;           }
 103                                 }
 104                         }
 105
 106             vector<string> tempOutNames;
 107             outputTypes["summary"] = tempOutNames;
 108
 109                 //check for optional parameter and set defaults
 110                         // ...at some point should added some additional type checking...
 111                         label = validParameter.validFile(parameters, "label", false);
 112                         if (label == "not found") { label = ""; }
 113                         else {
 114                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 115                                 else { allLines = 1;  }
 116                         }
 117
 118                         //get shared file
 119                         sharedfile = validParameter.validFile(parameters, "shared", true);
 120                         if (sharedfile == "not open") { sharedfile = ""; abort = true; }
 121                         else if (sharedfile == "not found") {
 122                                 //if there is a current shared file, use it
 123                                 sharedfile = m->getSharedFile();
 124                                 if (sharedfile != "") { m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); }
 125                                 else {  m->mothurOut("You have no current sharedfile and the shared parameter is required."); m->mothurOutEndLine(); abort = true; }
 126                         }else { m->setSharedFile(sharedfile); }
 127
 128
 129                         //if the user changes the output directory command factory will send this info to us in the output parameter
 130                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = m->hasPath(sharedfile);             }
 131
 132
 133                         metric = validParameter.validFile(parameters, "metric", false);                         if (metric == "not found") { metric = "cscore"; }
 134
 135                         if ((metric != "cscore") && (metric != "checker") && (metric != "combo") && (metric != "vratio")) {
 136                                 m->mothurOut("[ERROR]: " + metric + " is not a valid metric option for the cooccurrence command. Choices are cscore, checker, combo, vratio."); m->mothurOutEndLine(); abort = true;
 137                         }
 138
 139                         matrix = validParameter.validFile(parameters, "matrixmodel", false);                            if (matrix == "not found") { matrix = "sim2"; }
 140
 141                         if ((matrix != "sim1") && (matrix != "sim2") && (matrix != "sim3") && (matrix != "sim4") && (matrix != "sim5" ) && (matrix != "sim6" ) && (matrix != "sim7" ) && (matrix != "sim8" ) && (matrix != "sim9" )) {
 142                                 m->mothurOut("[ERROR]: " + matrix + " is not a valid matrix option for the cooccurrence command. Choices are sim1, sim2, sim3, sim4, sim5, sim6, sim7, sim8, sim9."); m->mothurOutEndLine(); abort = true;
 143                         }
 144
 145             groups = validParameter.validFile(parameters, "groups", false);
 146                         if (groups == "not found") { groups = "";   }
 147                         else {
 148                                 m->splitAtDash(groups, Groups);
 149                         }
 150                         m->setGroups(Groups);
 151
 152             string temp = validParameter.validFile(parameters, "iters", false);                 if (temp == "not found") { temp = "1000"; }
 153                         m->mothurConvert(temp, runs);
 154
 155                 }
 156
 157         }
 158         catch(exception& e) {
 159                 m->errorOut(e, "CooccurrenceCommand", "CooccurrenceCommand");
 160                 exit(1);
 161         }
 162 }
 163 //**********************************************************************************************************************
 164
 165 int CooccurrenceCommand::execute(){
 166         try {
 167
 168                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 169
 170                 InputData* input = new InputData(sharedfile, "sharedfile");
 171                 vector<SharedRAbundVector*> lookup = input->getSharedRAbundVectors();
 172                 string lastLabel = lookup[0]->getLabel();
 173
 174                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 175                 set<string> processedLabels;
 176                 set<string> userLabels = labels;
 177
 178         ofstream out;
 179                 string outputFileName = outputDir + m->getRootName(m->getSimpleName(sharedfile)) + "cooccurence.summary";
 180         m->openOutputFile(outputFileName, out);
 181         outputNames.push_back(outputFileName);  outputTypes["summary"].push_back(outputFileName);
 182         out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
 183         out << "metric\tlabel\tScore\tpValue\n";
 184
 185                 //as long as you are not at the end of the file or done wih the lines you want
 186                 while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 187
 188                         if (m->control_pressed) { for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  } delete input; out.close(); m->mothurRemove(outputFileName); return 0; }
 189
 190                         if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){
 191
 192                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 193
 194                                 getCooccurrence(lookup, out);
 195
 196                                 processedLabels.insert(lookup[0]->getLabel());
 197                                 userLabels.erase(lookup[0]->getLabel());
 198                         }
 199
 200                         if ((m->anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 201                                 string saveLabel = lookup[0]->getLabel();
 202
 203                                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 204                                 lookup = input->getSharedRAbundVectors(lastLabel);
 205                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 206                                 getCooccurrence(lookup, out);
 207
 208                                 processedLabels.insert(lookup[0]->getLabel());
 209                                 userLabels.erase(lookup[0]->getLabel());
 210
 211                                 //restore real lastlabel to save below
 212                                 lookup[0]->setLabel(saveLabel);
 213                         }
 214
 215                         lastLabel = lookup[0]->getLabel();
 216                         //prevent memory leak
 217                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i]; lookup[i] = NULL; }
 218
 219                         if (m->control_pressed) {  outputTypes.clear(); delete input; out.close(); m->mothurRemove(outputFileName); return 0; }
 220
 221                         //get next line to process
 222                         lookup = input->getSharedRAbundVectors();
 223                 }
 224
 225                 if (m->control_pressed) { delete input; out.close(); m->mothurRemove(outputFileName); return 0; }
 226
 227                 //output error messages about any remaining user labels
 228                 set<string>::iterator it;
 229                 bool needToRun = false;
 230                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 231                         m->mothurOut("Your file does not include the label " + *it);
 232                         if (processedLabels.count(lastLabel) != 1) {
 233                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 234                                 needToRun = true;
 235                         }else {
 236                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 237                         }
 238                 }
 239
 240                 //run last label if you need to
 241                 if (needToRun == true)  {
 242                         for (int i = 0; i < lookup.size(); i++) { if (lookup[i] != NULL) { delete lookup[i]; } }
 243                         lookup = input->getSharedRAbundVectors(lastLabel);
 244
 245                         m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 246
 247                         getCooccurrence(lookup, out);
 248
 249                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 250                 }
 251
 252         out.close();
 253
 254                 //reset groups parameter
 255                 delete input;
 256         m->clearGroups();
 257
 258         m->mothurOutEndLine();
 259                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 260                 m->mothurOut(outputFileName); m->mothurOutEndLine();
 261                 m->mothurOutEndLine();
 262
 263                 return 0;
 264         }
 265         catch(exception& e) {
 266                 m->errorOut(e, "CooccurrenceCommand", "execute");
 267                 exit(1);
 268         }
 269 }
 270 //**********************************************************************************************************************
 271
 272 int CooccurrenceCommand::getCooccurrence(vector<SharedRAbundVector*>& thisLookUp, ofstream& out){
 273     try {
 274         int numOTUS = thisLookUp[0]->getNumBins();
 275         vector< vector<int> > co_matrix; co_matrix.resize(thisLookUp[0]->getNumBins());
 276         for (int i = 0; i < thisLookUp[0]->getNumBins(); i++) { co_matrix[i].resize((thisLookUp.size()), 0); }
 277         vector<int> columntotal; columntotal.resize(thisLookUp.size(), 0);
 278         vector<int> rowtotal; rowtotal.resize(numOTUS, 0);
 279
 280         for (int i = 0; i < thisLookUp.size(); i++) { //nrows in the shared file
 281             for (int j = 0; j < thisLookUp[i]->getNumBins(); j++) { //cols of original shared file
 282                 if (m->control_pressed) { return 0; }
 283                 int abund = thisLookUp[i]->getAbundance(j);
 284
 285                 if(abund > 0) {
 286                     co_matrix[j][i] = 1;
 287                     rowtotal[j]++;
 288                     columntotal[i]++;
 289                 }
 290             }
 291         }
 292
 293         //nrows is ncols of inital matrix. All the functions need this value. They assume the transposition has already taken place and nrows and ncols refer to that matrix.
 294         //comatrix and initmatrix are still vectors of vectors of ints as in the original script. The abundancevector is only what was read in ie not a co-occurrence matrix!
 295         int nrows = numOTUS;//rows of inital matrix
 296         int ncols = thisLookUp.size();//groups
 297         double initscore = 0.0;
 298
 299         vector<double> stats;
 300         double probabilityMatrix[ncols * nrows];
 301         vector<vector<int> > nullmatrix(nrows, vector<int>(ncols, 0));
 302
 303         TrialSwap2 trial;
 304
 305         int n = accumulate( columntotal.begin(), columntotal.end(), 0 );
 306
 307         //============================================================
 308
 309         //generate a probability matrix. Only do this once.
 310         float start = 0.0;
 311
 312         if (matrix == "sim1") {
 313             for(int i=0;i<nrows;i++) {
 314                 for(int j=0;j<ncols;j++) {
 315                     probabilityMatrix[ncols * i + j] = start + 1/double(nrows*ncols);
 316                     start = start + 1/double(nrows*ncols);
 317                 }
 318             }
 319         }
 320         else if (matrix == "sim2") {
 321             for(int i=0;i<nrows;i++) {
 322                 start = 0.0;
 323                 for(int j=0;j<ncols;j++) {
 324                     probabilityMatrix[ncols * i + j] = start + 1/double(ncols);
 325                     start = start + 1/double(ncols);
 326                 }
 327             }
 328         }
 329
 330         else if (matrix == "sim3") {
 331             for(int j=0;j<ncols;j++) {
 332                 start = 0.0;
 333                 for(int i=0;i<nrows;i++) {
 334                     probabilityMatrix[ncols * i + j] = start + 1/double(nrows);
 335                     start = start + 1/double(nrows);
 336                 }
 337             }
 338         }
 339
 340         else if (matrix == "sim4") {
 341             for(int i=0;i<nrows;i++) {
 342                 start = 0.0;
 343                 for(int j=0;j<ncols;j++) {
 344                     probabilityMatrix[ncols * i + j] = start + columntotal[j]/double(n);
 345                     start = start + columntotal[j]/double(n);
 346                 }
 347             }
 348         }
 349
 350         else if (matrix == "sim5") {
 351             for(int j=0;j<ncols;j++) {
 352                 start = 0.0;
 353                 for(int i=0;i<nrows;i++) {
 354                     probabilityMatrix[ncols * i + j] = start + rowtotal[i]/double(n);
 355                     start = start + rowtotal[i]/double(n);
 356                 }
 357             }
 358         }
 359
 360         else if (matrix == "sim6") {
 361             for(int i=0;i<nrows;i++) {
 362                 for(int j=0;j<ncols;j++) {
 363                     probabilityMatrix[ncols * i + j] = start + columntotal[j]/double(n*nrows);
 364                     start = start + columntotal[j]/double(n*nrows);
 365                 }
 366             }
 367         }
 368
 369
 370         else if (matrix == "sim7") {
 371             for(int i=0;i<nrows;i++) {
 372                 for(int j=0;j<ncols;j++) {
 373                     probabilityMatrix[ncols * i + j] = start + rowtotal[i]/double(n*ncols);
 374                     start = start + rowtotal[i]/double(n*ncols);
 375                 }
 376             }
 377         }
 378
 379         else if (matrix == "sim8") {
 380             for(int i=0;i<nrows;i++) {
 381                 for(int j=0;j<ncols;j++) {
 382                     probabilityMatrix[ncols * i + j] = start + (rowtotal[i]*columntotal[j])/double(n*n);
 383                     start = start + (rowtotal[i]*columntotal[j])/double(n*n);
 384                 }
 385             }
 386         }
 387         else if (matrix == "sim9") { }
 388         else {
 389             m->mothurOut("[ERROR]: No model selected! \n");
 390             m->control_pressed = true;
 391         }
 392
 393
 394         //co_matrix is the transposed shared file, initmatrix is the original shared file
 395         if (metric == "cscore") { initscore = trial.calc_c_score(co_matrix, rowtotal, ncols, nrows); }
 396         else if (metric == "checker") { initscore = trial.calc_checker(co_matrix, rowtotal, ncols, nrows); }
 397         else if (metric == "vratio") { initscore = trial.calc_vratio(nrows, ncols, rowtotal, columntotal); }
 398         else if (metric == "combo") { initscore = trial.calc_combo(nrows, ncols, co_matrix); }
 399         else { m->mothurOut("[ERROR]: No metric selected!\n"); m->control_pressed = true; return 1; }
 400
 401         m->mothurOut("Initial c score: " + toString(initscore)); m->mothurOutEndLine();
 402
 403         double previous;
 404         double current;
 405         double randnum;
 406         int count;
 407
 408         //burn-in for sim9
 409         if(matrix == "sim9") {
 410             for(int i=0;i<10000;i++) trial.swap_checkerboards (co_matrix, rowtotal, columntotal, ncols, nrows);
 411         }
 412
 413         //populate null matrix from probability matrix, do this a lot.
 414         for(int i=0;i<runs;i++){
 415             nullmatrix.clear();
 416             //zero-fill the null matrix
 417             nullmatrix.assign(nrows, vector<int>(ncols, 0));
 418
 419             if(matrix == "sim1" || matrix == "sim6" || matrix == "sim8" || matrix == "sim7") {
 420                 count = 0;
 421                 while(count < n) {
 422                 nextnum2:
 423                     previous = 0.0;
 424                     randnum = rand() / double(RAND_MAX);
 425                     for(int i=0;i<nrows;i++) {
 426                         for(int j=0;j<ncols;j++) {
 427                             current = probabilityMatrix[ncols * i + j];
 428                             if(randnum <= current && randnum > previous) {
 429                                 nullmatrix[i][j] = 1;
 430                                 count++;
 431                                 if (count > n) break;
 432                                 else
 433                                     goto nextnum2;
 434                             }
 435                             previous = current;
 436                         }
 437                     }
 438                 }
 439             }
 440
 441             else if(matrix == "sim2" || matrix == "sim4") {
 442                 for(int i=0;i<nrows;i++) {
 443                     previous = 0.0;
 444                     count = 0;
 445                     while(count < rowtotal[i]) {
 446                         randnum = rand() / double(RAND_MAX);
 447                         for(int j=0;j<ncols;j++) {
 448                             current = probabilityMatrix[ncols * i + j];
 449                             if(randnum <= current && randnum > previous && nullmatrix[i][j] != 1) {
 450                                 nullmatrix[i][j] = 1;
 451                                 count++;
 452                                 previous = 0.0;
 453                                 break;
 454                             }
 455                             previous = current;
 456                         }
 457                     }
 458                 }
 459             }
 460
 461             else if(matrix == "sim3" || matrix == "sim5") {
 462                 //columns
 463                 for(int j=0;j<ncols;j++) {
 464                     count = 0;
 465                     while(count < columntotal[j]) {
 466                         randnum = rand() / double(RAND_MAX);
 467                         for(int i=0;i<nrows;i++) {
 468                             current = probabilityMatrix[ncols * i + j];
 469                             if(randnum <= current && randnum > previous && nullmatrix[i][j] != 1) {
 470                                 nullmatrix[i][j] = 1;
 471                                 count++;
 472                                 previous = 0.0;
 473                                 break;
 474                             }
 475                             previous = current;
 476                         }
 477                     }
 478                 }
 479             }
 480
 481             //swap_checkerboards takes the original matrix and swaps checkerboards
 482             else if(matrix == "sim9") {
 483                 trial.swap_checkerboards (co_matrix, rowtotal, columntotal, ncols, nrows);
 484             }
 485             else {
 486                 m->mothurOut("[ERROR]: No null model selected!\n\n"); m->control_pressed = true;
 487                 return 1;
 488             }
 489
 490             //run metric on null matrix and add score to the stats vector
 491             if (metric == "cscore"){
 492                 stats.push_back(trial.calc_c_score(nullmatrix, rowtotal, ncols, nrows));
 493             }
 494             else if (metric == "checker") {
 495                 stats.push_back(trial.calc_checker(nullmatrix, rowtotal, ncols, nrows));
 496             }
 497             else if (metric == "vratio") {
 498                 stats.push_back(trial.calc_vratio(nrows, ncols, rowtotal, columntotal));
 499             }
 500             else if (metric == "combo") {
 501                 stats.push_back(trial.calc_combo(nrows, ncols, nullmatrix));
 502             }
 503             else {
 504                 m->mothurOut("[ERROR]: No metric selected!\n\n"); m->control_pressed = true;
 505                 return 1;
 506             }
 507
 508         }
 509
 510
 511
 512         double total = 0.0;
 513         for (int i=0; i<stats.size();i++) { total+=stats[i]; }
 514
 515         double nullMean = double (total/(double)stats.size());
 516
 517         m->mothurOutEndLine(); m->mothurOut("average metric score: " + toString(nullMean)); m->mothurOutEndLine();
 518
 519         double pvalue = 0.0;
 520         if (metric == "cscore" || metric == "checker") { pvalue = trial.calc_pvalue_greaterthan (stats, initscore); }
 521         else{ pvalue = trial.calc_pvalue_lessthan (stats, initscore); }
 522
 523         m->mothurOut("pvalue: " + toString(pvalue)); m->mothurOutEndLine();
 524         out << metric << '\t' << thisLookUp[0]->getLabel() << '\t' << nullMean << '\t' << pvalue << endl;
 525
 526         return 0;
 527     }
 528     catch(exception& e) {
 529         m->errorOut(e, "CooccurrenceCommand", "Cooccurrence");
 530         exit(1);
 531     }
 532 }
 533 //**********************************************************************************************************************
 534
 535