sensspeccommand.cpp

   1 /*
   2  *  sensspeccommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by Pat Schloss on 7/6/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "sensspeccommand.h"
  11
  12 //**********************************************************************************************************************
  13 vector<string> SensSpecCommand::setParameters(){
  14         try {
  15                 CommandParameter plist("list", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(plist);
  16                 CommandParameter pphylip("phylip", "InputTypes", "", "", "PhylipColumn", "PhylipColumn", "none",false,false); parameters.push_back(pphylip);
  17                 //CommandParameter pname("name", "InputTypes", "", "", "none", "none", "ColumnName",false,false); parameters.push_back(pname);
  18                 CommandParameter pcolumn("column", "InputTypes", "", "", "PhylipColumn", "PhylipColumn", "none",false,false); parameters.push_back(pcolumn);
  19                 CommandParameter plabel("label", "String", "", "", "", "", "",false,false); parameters.push_back(plabel);
  20                 CommandParameter pcutoff("cutoff", "Number", "", "-1.00", "", "", "",false,false); parameters.push_back(pcutoff);
  21                 CommandParameter pprecision("precision", "Number", "", "100", "", "", "",false,false); parameters.push_back(pprecision);
  22                 CommandParameter phard("hard", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(phard);
  23                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  24                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  25
  26                 vector<string> myArray;
  27                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  28                 return myArray;
  29         }
  30         catch(exception& e) {
  31                 m->errorOut(e, "SensSpecCommand", "setParameters");
  32                 exit(1);
  33         }
  34 }
  35 //**********************************************************************************************************************
  36 string SensSpecCommand::getHelpString(){
  37         try {
  38                 string helpString = "";
  39                 helpString += "The sens.spec command....\n";
  40                 return helpString;
  41         }
  42         catch(exception& e) {
  43                 m->errorOut(e, "SensSpecCommand", "getHelpString");
  44                 exit(1);
  45         }
  46 }
  47 //**********************************************************************************************************************
  48 SensSpecCommand::SensSpecCommand(){
  49         try {
  50                 abort = true; calledHelp = true;
  51                 setParameters();
  52                 vector<string> tempOutNames;
  53                 outputTypes["sensspec"] = tempOutNames;
  54         }
  55         catch(exception& e) {
  56                 m->errorOut(e, "SensSpecCommand", "SensSpecCommand");
  57                 exit(1);
  58         }
  59 }
  60 //***************************************************************************************************************
  61
  62 SensSpecCommand::SensSpecCommand(string option)  {
  63         try {
  64
  65                 abort = false; calledHelp = false;
  66
  67                 //allow user to run help
  68                 if(option == "help") { help(); abort = true; calledHelp = true; }
  69
  70                 else {
  71                         string temp;
  72
  73                         vector<string> myArray = setParameters();
  74
  75                         OptionParser parser(option);
  76                         map<string,string> parameters = parser.getParameters();
  77
  78                         ValidParameters validParameter;
  79                         map<string,string>::iterator it;
  80
  81                         //check to make sure all parameters are valid for command
  82                         for (it = parameters.begin(); it != parameters.end(); it++) {
  83                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  84                         }
  85
  86                         //initialize outputTypes
  87                         vector<string> tempOutNames;
  88                         outputTypes["sensspec"] = tempOutNames;
  89
  90                         //if the user changes the input directory command factory will send this info to us in the output parameter
  91                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  92                         if (inputDir == "not found"){   inputDir = "";          }
  93                         else {
  94                                 string path;
  95                                 it = parameters.find("list");
  96                                 //user has given a template file
  97                                 if(it != parameters.end()){
  98                                         path = m->hasPath(it->second);
  99                                         //if the user has not given a path then, add inputdir. else leave path alone.
 100                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 101                                 }
 102
 103                                 it = parameters.find("phylip");
 104                                 //user has given a template file
 105                                 if(it != parameters.end()){
 106                                         path = m->hasPath(it->second);
 107                                         //if the user has not given a path then, add inputdir. else leave path alone.
 108                                         if (path == "") {       parameters["phylip"] = inputDir + it->second;           }
 109                                 }
 110
 111                                 it = parameters.find("column");
 112                                 //user has given a template file
 113                                 if(it != parameters.end()){
 114                                         path = m->hasPath(it->second);
 115                                         //if the user has not given a path then, add inputdir. else leave path alone.
 116                                         if (path == "") {       parameters["column"] = inputDir + it->second;           }
 117                                 }
 118
 119                                 //it = parameters.find("name");
 120                                 //user has given a template file
 121                                 //if(it != parameters.end()){
 122                                         //path = m->hasPath(it->second);
 123                                         //if the user has not given a path then, add inputdir. else leave path alone.
 124                                         //if (path == "") {     parameters["name"] = inputDir + it->second;             }
 125                                 //}
 126
 127                         }
 128                         //check for required parameters
 129                         listFile = validParameter.validFile(parameters, "list", true);
 130                         if (listFile == "not found") {
 131                                 listFile = m->getListFile();
 132                                 if (listFile != "") { m->mothurOut("Using " + listFile + " as input file for the list parameter."); m->mothurOutEndLine(); }
 133                                 else {  m->mothurOut("You have no current list file and the list parameter is required."); m->mothurOutEndLine(); abort = true; }
 134                         }
 135                         else if (listFile == "not open") { abort = true; }
 136
 137                         phylipfile = validParameter.validFile(parameters, "phylip", true);
 138                         if (phylipfile == "not found") { phylipfile = "";  }
 139                         else if (phylipfile == "not open") { abort = true; }
 140                         else { distFile = phylipfile; format = "phylip";   }
 141
 142                         columnfile = validParameter.validFile(parameters, "column", true);
 143                         if (columnfile == "not found") { columnfile = ""; }
 144                         else if (columnfile == "not open") { abort = true; }
 145                         else { distFile = columnfile; format = "column";   }
 146
 147                         if ((phylipfile == "") && (columnfile == "")) { //is there are current file available for either of these?
 148                                 //give priority to column, then phylip
 149                                 columnfile = m->getColumnFile();
 150                                 if (columnfile != "") {  distFile = columnfile; format = "column";  m->mothurOut("Using " + columnfile + " as input file for the column parameter."); m->mothurOutEndLine(); }
 151                                 else {
 152                                         phylipfile = m->getPhylipFile();
 153                                         if (phylipfile != "") {  distFile = phylipfile; format = "phylip"; m->mothurOut("Using " + phylipfile + " as input file for the phylip parameter."); m->mothurOutEndLine(); }
 154                                         else {
 155                                                 m->mothurOut("No valid current files. You must provide a phylip or column file."); m->mothurOutEndLine();
 156                                                 abort = true;
 157                                         }
 158                                 }
 159                         }else if ((phylipfile != "") && (columnfile != "")) { m->mothurOut("When executing a sens.spec command you must enter ONLY ONE of the following: phylip or column."); m->mothurOutEndLine(); abort = true; }
 160
 161
 162                         //if the user changes the output directory command factory will send this info to us in the output parameter
 163                         outputDir = validParameter.validFile(parameters, "outputdir", false);
 164                         if (outputDir == "not found"){
 165                                 outputDir = "";
 166                                 outputDir += m->hasPath(listFile); //if user entered a file with a path then preserve it
 167                         }
 168
 169                         //check for optional parameter and set defaults
 170                         // ...at some point should added some additional type checking...
 171                         temp = validParameter.validFile(parameters, "hard", false);
 172                         if (temp == "not found"){       hard = 0;       }
 173                         else if(!m->isTrue(temp))       {       hard = 0;       }
 174                         else if(m->isTrue(temp))        {       hard = 1;       }
 175
 176 //                      temp = validParameter.validFile(parameters, "name", true);
 177 //                      if (temp == "not found")        {       nameFile = "";          }
 178 //                      else if(temp == "not open")     {       abort = true;           }
 179 //                      else                                            {       nameFile = temp;        }
 180 //                      cout << "name:\t" << nameFile << endl;
 181
 182                         temp = validParameter.validFile(parameters, "cutoff", false);           if (temp == "not found") { temp = "-1.00"; }
 183                         convert(temp, cutoff);
 184 //                      cout << cutoff << endl;
 185
 186                         temp = validParameter.validFile(parameters, "precision", false);        if (temp == "not found") { temp = "100"; }
 187                         convert(temp, precision);
 188 //                      cout << precision << endl;
 189
 190                         lineLabel = validParameter.validFile(parameters, "label", false);       if (lineLabel == "not found") { lineLabel = ""; }
 191
 192                         sensSpecFileName = listFile.substr(0,listFile.find_last_of('.')) + ".sensspec";
 193                 }
 194         }
 195         catch(exception& e) {
 196                 m->errorOut(e, "SensSpecCommand", "SensSpecCommand");
 197                 exit(1);
 198         }
 199 }
 200 //***************************************************************************************************************
 201
 202 int SensSpecCommand::execute(){
 203         try{
 204                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 205
 206                 setUpOutput();
 207                 outputNames.push_back(sensSpecFileName); outputTypes["sensspec"].push_back(sensSpecFileName);
 208                 if(format == "phylip")          {       processPhylip();        }
 209                 else if(format == "column")     {       processColumn();        }
 210
 211
 212                 return 0;
 213         }
 214         catch(exception& e) {
 215                 m->errorOut(e, "SensSpecCommand", "execute");
 216                 exit(1);
 217         }
 218 }
 219
 220 //***************************************************************************************************************
 221
 222 void SensSpecCommand::processPhylip(){
 223         try{
 224                 //probably need some checking to confirm that the names in the distance matrix are the same as those in the list file
 225
 226                 ifstream inputListFile;
 227                 m->openInputFile(listFile, inputListFile);
 228
 229                 string origCutoff = "";
 230                 bool getCutoff = 0;
 231                 if(cutoff == -1.00)     {       getCutoff = 1;                                                                                                                  }
 232                 else                            {       origCutoff = toString(cutoff);  cutoff += (0.49 / double(precision));   }
 233
 234                 string label;
 235                 int numOTUs;
 236
 237                 map<string, int> seqMap;
 238                 string seqList;
 239
 240                 while(inputListFile){
 241                         inputListFile >> label >> numOTUs;
 242                         for(int i=0;i<numOTUs;i++){
 243                                 inputListFile >> seqList;
 244                                 int seqListLength = seqList.length();
 245                                 string seqName = "";
 246                                 for(int j=0;j<seqListLength;j++){
 247
 248                                         if(seqList[j] == ','){
 249                                                 seqMap[seqName] = i;
 250                                                 seqName = "";
 251                                         }
 252                                         else{
 253                                                 seqName += seqList[j];
 254                                         }
 255
 256                                 }
 257                                 seqMap[seqName] = i;
 258                         }
 259                         m->gobble(inputListFile);
 260
 261                         int lNumSeqs = seqMap.size();
 262                         int pNumSeqs = 0;
 263
 264                         ifstream phylipFile;
 265                         m->openInputFile(distFile, phylipFile);
 266                         phylipFile >> pNumSeqs;
 267                         if(pNumSeqs != lNumSeqs){       cout << "numSeq mismatch!" << endl;     }
 268
 269                         string seqName;
 270                         double distance;
 271                         vector<int> otuIndices(lNumSeqs, -1);
 272
 273                         truePositives = 0;
 274                         falsePositives = 0;
 275                         trueNegatives = 0;
 276                         falseNegatives = 0;
 277
 278                         if(getCutoff == 1){
 279                                 if(label != "unique"){
 280                                         origCutoff = label;
 281                                         convert(label, cutoff);
 282                                         if(hard == 0){  cutoff += (0.49 / double(precision));   }
 283                                 }
 284                                 else{
 285                                         origCutoff = "unique";
 286                                         cutoff = 0.0000;
 287                                 }
 288                         }
 289
 290                         cout << label << endl;
 291
 292                         for(int i=0;i<lNumSeqs;i++){
 293                                 phylipFile >> seqName;
 294                                 otuIndices[i] = seqMap[seqName];
 295
 296                                 for(int j=0;j<i;j++){
 297                                         phylipFile >> distance;
 298
 299                                         if(distance <= cutoff){
 300                                                 if(otuIndices[i] == otuIndices[j])      {       truePositives++;        }
 301                                                 else                                                            {       falseNegatives++;       }
 302                                         }
 303                                         else{
 304                                                 if(otuIndices[i] == otuIndices[j])      {       falsePositives++;       }
 305                                                 else                                                            {       trueNegatives++;        }
 306                                         }
 307                                 }
 308                         }
 309                         phylipFile.close();
 310
 311                         outputStatistics(label, origCutoff);
 312                 }
 313                 inputListFile.close();
 314
 315         }
 316         catch(exception& e) {
 317                 m->errorOut(e, "SensSpecCommand", "processPhylip");
 318                 exit(1);
 319         }
 320 }
 321
 322 //***************************************************************************************************************
 323
 324 void SensSpecCommand::processColumn(){
 325         try{
 326                 ifstream inputListFile;
 327                 m->openInputFile(listFile, inputListFile);
 328
 329                 string origCutoff = "";
 330                 bool getCutoff = 0;
 331                 if(cutoff == -1.00)     {       getCutoff = 1;                                                                                                                  }
 332                 else                            {       origCutoff = toString(cutoff);  cutoff += (0.49 / double(precision));   }
 333
 334                 set<string> seqPairSet;
 335
 336                 string label, seqList;
 337                 int numOTUs;
 338                 int numSeqs;
 339
 340                 while(inputListFile){
 341                         numSeqs = 0;
 342
 343                         inputListFile >> label >> numOTUs;
 344                         for(int i=0;i<numOTUs;i++){
 345
 346                                 vector<string> seqNameVector;
 347
 348                                 inputListFile >> seqList;
 349                                 int seqListLength = seqList.length();
 350                                 string seqName = "";
 351                                 for(int j=0;j<seqListLength;j++){
 352
 353                                         if(seqList[j] == ','){
 354                                                 seqNameVector.push_back(seqName);
 355                                                 seqName = "";
 356                                         }
 357                                         else{
 358                                                 seqName += seqList[j];
 359                                         }
 360                                 }
 361                                 seqNameVector.push_back(seqName);
 362
 363                                 numSeqs += seqNameVector.size();
 364
 365                                 int numSeqsInOTU = seqNameVector.size();
 366                                 for(int j=0;j<numSeqsInOTU;j++){
 367                                         string seqPairString = "";
 368                                         for(int k=0;k<j;k++){
 369                                                 if(seqNameVector[j] < seqNameVector[k]) {       seqPairString = seqNameVector[j] + '\t' + seqNameVector[k];     }
 370                                                 else                                                                    {       seqPairString = seqNameVector[k] + '\t' + seqNameVector[j];     }
 371                                                 seqPairSet.insert(seqPairString);
 372                                         }
 373                                 }
 374                         }
 375                         m->gobble(inputListFile);
 376
 377                         int numDists = (numSeqs * (numSeqs-1) / 2);
 378
 379                         ifstream columnFile;
 380                         m->openInputFile(distFile, columnFile);
 381                         string seqNameA, seqNameB, seqPairString;
 382                         double distance;
 383
 384                         truePositives = 0;
 385                         falsePositives = 0;
 386                         trueNegatives = numDists;
 387                         falseNegatives = 0;
 388
 389                         if(getCutoff == 1){
 390                                 if(label != "unique"){
 391                                         origCutoff = label;
 392                                         convert(label, cutoff);
 393                                         if(hard == 0){  cutoff += (0.49 / double(precision));   }
 394                                 }
 395                                 else{
 396                                         origCutoff = "unique";
 397                                         cutoff = 0.0000;
 398                                 }
 399                         }
 400
 401                         cout << label << endl;
 402
 403                         while(columnFile){
 404                                 columnFile >> seqNameA >> seqNameB >> distance;
 405                                 if(seqNameA < seqNameB) {       seqPairString = seqNameA + '\t' + seqNameB;     }
 406                                 else                                    {       seqPairString = seqNameB + '\t' + seqNameA;     }
 407
 408                                 set<string>::iterator it = seqPairSet.find(seqPairString);
 409
 410                                 if(distance <= cutoff){
 411                                         if(it != seqPairSet.end()){
 412                                                 truePositives++;
 413                                                 seqPairSet.erase(it);
 414                                         }
 415                                         else{
 416                                                 falseNegatives++;
 417                                         }
 418                                         trueNegatives--;
 419                                 }
 420                                 else if(it != seqPairSet.end()){
 421                                         falsePositives++;
 422                                         trueNegatives--;
 423                                         seqPairSet.erase(it);
 424                                 }
 425
 426                                 m->gobble(columnFile);
 427                         }
 428                         falsePositives += seqPairSet.size();
 429
 430                         outputStatistics(label, origCutoff);
 431                 }
 432         }
 433         catch(exception& e) {
 434                 m->errorOut(e, "SensSpecCommand", "processColumn");
 435                 exit(1);
 436         }
 437 }
 438
 439 //***************************************************************************************************************
 440
 441 void SensSpecCommand::setUpOutput(){
 442         try{
 443                 ofstream sensSpecFile;
 444                 m->openOutputFile(sensSpecFileName, sensSpecFile);
 445
 446                 sensSpecFile << "label\tcutoff\ttp\ttn\tfp\tfn\tsensitivity\tspecificity\tppv\tnpv\tfdr\taccuracy\tmcc\tf1score\n";
 447
 448                 sensSpecFile.close();
 449         }
 450         catch(exception& e) {
 451                 m->errorOut(e, "SensSpecCommand", "setUpOutput");
 452                 exit(1);
 453         }
 454 }
 455
 456 //***************************************************************************************************************
 457
 458 void SensSpecCommand::outputStatistics(string label, string cutoff){
 459         try{
 460                 double tp = (double) truePositives;
 461                 double fp = (double) falsePositives;
 462                 double tn = (double) trueNegatives;
 463                 double fn = (double) falseNegatives;
 464
 465                 double p = tp + fn;
 466                 double n = fp + tn;
 467                 double pPrime = tp + fp;
 468                 double nPrime = tn + fn;
 469
 470                 double sensitivity = tp / p;
 471                 double specificity = tn / n;
 472                 double positivePredictiveValue = tp / pPrime;
 473                 double negativePredictiveValue = tn / nPrime;
 474                 double falseDiscoveryRate = fp / pPrime;
 475
 476                 double accuracy = (tp + tn) / (p + n);
 477                 double matthewsCorrCoef = (tp * tn - fp * fn) / sqrt(p * n * pPrime * nPrime);  if(p == 0 || n == 0){   matthewsCorrCoef = 0;   }
 478                 double f1Score = 2.0 * tp / (p + pPrime);
 479
 480
 481                 if(p == 0)                      {       sensitivity = 0;        matthewsCorrCoef = 0;   }
 482                 if(n == 0)                      {       specificity = 0;        matthewsCorrCoef = 0;   }
 483                 if(p + n == 0)          {       accuracy = 0;                                                           }
 484                 if(p + pPrime == 0)     {       f1Score = 0;                                                            }
 485                 if(pPrime == 0)         {       positivePredictiveValue = 0;    falseDiscoveryRate = 0; matthewsCorrCoef = 0;   }
 486                 if(nPrime == 0)         {       negativePredictiveValue = 0;    matthewsCorrCoef = 0;                                                   }
 487
 488                 ofstream sensSpecFile;
 489                 m->openOutputFileAppend(sensSpecFileName, sensSpecFile);
 490
 491                 sensSpecFile << label << '\t' << cutoff << '\t';
 492                 sensSpecFile << truePositives << '\t' << trueNegatives << '\t' << falsePositives << '\t' << falseNegatives << '\t';
 493                 sensSpecFile << setprecision(4);
 494                 sensSpecFile << sensitivity << '\t' << specificity << '\t' << positivePredictiveValue << '\t' << negativePredictiveValue << '\t';
 495                 sensSpecFile << falseDiscoveryRate << '\t' << accuracy << '\t' << matthewsCorrCoef << '\t' << f1Score << endl;
 496
 497                 sensSpecFile.close();
 498         }
 499         catch(exception& e) {
 500                 m->errorOut(e, "SensSpecCommand", "outputStatistics");
 501                 exit(1);
 502         }
 503 }
 504
 505 //***************************************************************************************************************
 506
 507
 508