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                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  70
  71                 else {
  72                         string temp;
  73
  74                         vector<string> myArray = setParameters();
  75
  76                         OptionParser parser(option);
  77                         map<string,string> parameters = parser.getParameters();
  78
  79                         ValidParameters validParameter;
  80                         map<string,string>::iterator it;
  81
  82                         //check to make sure all parameters are valid for command
  83                         for (it = parameters.begin(); it != parameters.end(); it++) {
  84                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  85                         }
  86
  87                         //initialize outputTypes
  88                         vector<string> tempOutNames;
  89                         outputTypes["sensspec"] = tempOutNames;
  90
  91                         //if the user changes the input directory command factory will send this info to us in the output parameter
  92                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  93                         if (inputDir == "not found"){   inputDir = "";          }
  94                         else {
  95                                 string path;
  96                                 it = parameters.find("list");
  97                                 //user has given a template file
  98                                 if(it != parameters.end()){
  99                                         path = m->hasPath(it->second);
 100                                         //if the user has not given a path then, add inputdir. else leave path alone.
 101                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 102                                 }
 103
 104                                 it = parameters.find("phylip");
 105                                 //user has given a template file
 106                                 if(it != parameters.end()){
 107                                         path = m->hasPath(it->second);
 108                                         //if the user has not given a path then, add inputdir. else leave path alone.
 109                                         if (path == "") {       parameters["phylip"] = inputDir + it->second;           }
 110                                 }
 111
 112                                 it = parameters.find("column");
 113                                 //user has given a template file
 114                                 if(it != parameters.end()){
 115                                         path = m->hasPath(it->second);
 116                                         //if the user has not given a path then, add inputdir. else leave path alone.
 117                                         if (path == "") {       parameters["column"] = inputDir + it->second;           }
 118                                 }
 119
 120                                 //it = parameters.find("name");
 121                                 //user has given a template file
 122                                 //if(it != parameters.end()){
 123                                         //path = m->hasPath(it->second);
 124                                         //if the user has not given a path then, add inputdir. else leave path alone.
 125                                         //if (path == "") {     parameters["name"] = inputDir + it->second;             }
 126                                 //}
 127
 128                         }
 129                         //check for required parameters
 130                         listFile = validParameter.validFile(parameters, "list", true);
 131                         if (listFile == "not found") {
 132                                 listFile = m->getListFile();
 133                                 if (listFile != "") { m->mothurOut("Using " + listFile + " as input file for the list parameter."); m->mothurOutEndLine(); }
 134                                 else {  m->mothurOut("You have no current list file and the list parameter is required."); m->mothurOutEndLine(); abort = true; }
 135                         }
 136                         else if (listFile == "not open") { abort = true; }
 137
 138                         phylipfile = validParameter.validFile(parameters, "phylip", true);
 139                         if (phylipfile == "not found") { phylipfile = "";  }
 140                         else if (phylipfile == "not open") { abort = true; }
 141                         else { distFile = phylipfile; format = "phylip";   }
 142
 143                         columnfile = validParameter.validFile(parameters, "column", true);
 144                         if (columnfile == "not found") { columnfile = ""; }
 145                         else if (columnfile == "not open") { abort = true; }
 146                         else { distFile = columnfile; format = "column";   }
 147
 148                         if ((phylipfile == "") && (columnfile == "")) { //is there are current file available for either of these?
 149                                 //give priority to column, then phylip
 150                                 columnfile = m->getColumnFile();
 151                                 if (columnfile != "") {  distFile = columnfile; format = "column";  m->mothurOut("Using " + columnfile + " as input file for the column parameter."); m->mothurOutEndLine(); }
 152                                 else {
 153                                         phylipfile = m->getPhylipFile();
 154                                         if (phylipfile != "") {  distFile = phylipfile; format = "phylip"; m->mothurOut("Using " + phylipfile + " as input file for the phylip parameter."); m->mothurOutEndLine(); }
 155                                         else {
 156                                                 m->mothurOut("No valid current files. You must provide a phylip or column file."); m->mothurOutEndLine();
 157                                                 abort = true;
 158                                         }
 159                                 }
 160                         }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; }
 161
 162
 163                         //if the user changes the output directory command factory will send this info to us in the output parameter
 164                         outputDir = validParameter.validFile(parameters, "outputdir", false);
 165                         if (outputDir == "not found"){
 166                                 outputDir = "";
 167                                 outputDir += m->hasPath(listFile); //if user entered a file with a path then preserve it
 168                         }
 169
 170                         //check for optional parameter and set defaults
 171                         // ...at some point should added some additional type checking...
 172                         temp = validParameter.validFile(parameters, "hard", false);
 173                         if (temp == "not found"){       hard = 0;       }
 174                         else if(!m->isTrue(temp))       {       hard = 0;       }
 175                         else if(m->isTrue(temp))        {       hard = 1;       }
 176
 177 //                      temp = validParameter.validFile(parameters, "name", true);
 178 //                      if (temp == "not found")        {       nameFile = "";          }
 179 //                      else if(temp == "not open")     {       abort = true;           }
 180 //                      else                                            {       nameFile = temp;        }
 181 //                      cout << "name:\t" << nameFile << endl;
 182
 183                         temp = validParameter.validFile(parameters, "cutoff", false);           if (temp == "not found") { temp = "-1.00"; }
 184                         convert(temp, cutoff);
 185 //                      cout << cutoff << endl;
 186
 187                         temp = validParameter.validFile(parameters, "precision", false);        if (temp == "not found") { temp = "100"; }
 188                         convert(temp, precision);
 189 //                      cout << precision << endl;
 190
 191                         lineLabel = validParameter.validFile(parameters, "label", false);       if (lineLabel == "not found") { lineLabel = ""; }
 192
 193                         sensSpecFileName = outputDir + m->getRootName(m->getSimpleName(listFile)) + ".sensspec";
 194                 }
 195         }
 196         catch(exception& e) {
 197                 m->errorOut(e, "SensSpecCommand", "SensSpecCommand");
 198                 exit(1);
 199         }
 200 }
 201 //***************************************************************************************************************
 202
 203 int SensSpecCommand::execute(){
 204         try{
 205                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 206
 207                 setUpOutput();
 208                 outputNames.push_back(sensSpecFileName); outputTypes["sensspec"].push_back(sensSpecFileName);
 209                 if(format == "phylip")          {       processPhylip();        }
 210                 else if(format == "column")     {       processColumn();        }
 211
 212                 m->mothurOutEndLine();
 213                 m->mothurOut("Output File Name: "); m->mothurOutEndLine();
 214                 m->mothurOut(sensSpecFileName); m->mothurOutEndLine();
 215                 m->mothurOutEndLine();
 216
 217
 218                 return 0;
 219         }
 220         catch(exception& e) {
 221                 m->errorOut(e, "SensSpecCommand", "execute");
 222                 exit(1);
 223         }
 224 }
 225
 226 //***************************************************************************************************************
 227
 228 void SensSpecCommand::processPhylip(){
 229         try{
 230                 //probably need some checking to confirm that the names in the distance matrix are the same as those in the list file
 231
 232                 ifstream inputListFile;
 233                 m->openInputFile(listFile, inputListFile);
 234
 235                 string origCutoff = "";
 236                 bool getCutoff = 0;
 237                 if(cutoff == -1.00)     {       getCutoff = 1;                                                                                                                  }
 238                 else                            {       origCutoff = toString(cutoff);  cutoff += (0.49 / double(precision));   }
 239
 240                 string label;
 241                 int numOTUs;
 242
 243                 map<string, int> seqMap;
 244                 string seqList;
 245
 246                 while(inputListFile){
 247                         inputListFile >> label >> numOTUs;
 248                         for(int i=0;i<numOTUs;i++){
 249                                 inputListFile >> seqList;
 250                                 int seqListLength = seqList.length();
 251                                 string seqName = "";
 252                                 for(int j=0;j<seqListLength;j++){
 253
 254                                         if(seqList[j] == ','){
 255                                                 seqMap[seqName] = i;
 256                                                 seqName = "";
 257                                         }
 258                                         else{
 259                                                 seqName += seqList[j];
 260                                         }
 261
 262                                 }
 263                                 seqMap[seqName] = i;
 264                         }
 265                         m->gobble(inputListFile);
 266
 267                         int lNumSeqs = seqMap.size();
 268                         int pNumSeqs = 0;
 269
 270                         ifstream phylipFile;
 271                         m->openInputFile(distFile, phylipFile);
 272                         phylipFile >> pNumSeqs;
 273                         if(pNumSeqs != lNumSeqs){       cout << "numSeq mismatch!" << endl;     }
 274
 275                         string seqName;
 276                         double distance;
 277                         vector<int> otuIndices(lNumSeqs, -1);
 278
 279                         truePositives = 0;
 280                         falsePositives = 0;
 281                         trueNegatives = 0;
 282                         falseNegatives = 0;
 283
 284                         if(getCutoff == 1){
 285                                 if(label != "unique"){
 286                                         origCutoff = label;
 287                                         convert(label, cutoff);
 288                                         if(hard == 0){  cutoff += (0.49 / double(precision));   }
 289                                 }
 290                                 else{
 291                                         origCutoff = "unique";
 292                                         cutoff = 0.0000;
 293                                 }
 294                         }
 295
 296                         cout << label << endl;
 297
 298                         for(int i=0;i<lNumSeqs;i++){
 299                                 phylipFile >> seqName;
 300                                 otuIndices[i] = seqMap[seqName];
 301
 302                                 for(int j=0;j<i;j++){
 303                                         phylipFile >> distance;
 304
 305                                         if(distance <= cutoff){
 306                                                 if(otuIndices[i] == otuIndices[j])      {       truePositives++;        }
 307                                                 else                                                            {       falseNegatives++;       }
 308                                         }
 309                                         else{
 310                                                 if(otuIndices[i] == otuIndices[j])      {       falsePositives++;       }
 311                                                 else                                                            {       trueNegatives++;        }
 312                                         }
 313                                 }
 314                         }
 315                         phylipFile.close();
 316
 317                         outputStatistics(label, origCutoff);
 318                 }
 319                 inputListFile.close();
 320
 321         }
 322         catch(exception& e) {
 323                 m->errorOut(e, "SensSpecCommand", "processPhylip");
 324                 exit(1);
 325         }
 326 }
 327
 328 //***************************************************************************************************************
 329
 330 void SensSpecCommand::processColumn(){
 331         try{
 332                 ifstream inputListFile;
 333                 m->openInputFile(listFile, inputListFile);
 334
 335                 string origCutoff = "";
 336                 bool getCutoff = 0;
 337                 if(cutoff == -1.00)     {       getCutoff = 1;                                                                                                                  }
 338                 else                            {       origCutoff = toString(cutoff);  cutoff += (0.49 / double(precision));   }
 339
 340                 set<string> seqPairSet;
 341
 342                 string label, seqList;
 343                 int numOTUs;
 344                 int numSeqs;
 345
 346                 while(inputListFile){
 347                         numSeqs = 0;
 348
 349                         inputListFile >> label >> numOTUs;
 350                         for(int i=0;i<numOTUs;i++){
 351
 352                                 vector<string> seqNameVector;
 353
 354                                 inputListFile >> seqList;
 355                                 int seqListLength = seqList.length();
 356                                 string seqName = "";
 357                                 for(int j=0;j<seqListLength;j++){
 358
 359                                         if(seqList[j] == ','){
 360                                                 seqNameVector.push_back(seqName);
 361                                                 seqName = "";
 362                                         }
 363                                         else{
 364                                                 seqName += seqList[j];
 365                                         }
 366                                 }
 367                                 seqNameVector.push_back(seqName);
 368
 369                                 numSeqs += seqNameVector.size();
 370
 371                                 int numSeqsInOTU = seqNameVector.size();
 372                                 for(int j=0;j<numSeqsInOTU;j++){
 373                                         string seqPairString = "";
 374                                         for(int k=0;k<j;k++){
 375                                                 if(seqNameVector[j] < seqNameVector[k]) {       seqPairString = seqNameVector[j] + '\t' + seqNameVector[k];     }
 376                                                 else                                                                    {       seqPairString = seqNameVector[k] + '\t' + seqNameVector[j];     }
 377                                                 seqPairSet.insert(seqPairString);
 378                                         }
 379                                 }
 380                         }
 381                         m->gobble(inputListFile);
 382
 383                         int numDists = (numSeqs * (numSeqs-1) / 2);
 384
 385                         ifstream columnFile;
 386                         m->openInputFile(distFile, columnFile);
 387                         string seqNameA, seqNameB, seqPairString;
 388                         double distance;
 389
 390                         truePositives = 0;
 391                         falsePositives = 0;
 392                         trueNegatives = numDists;
 393                         falseNegatives = 0;
 394
 395                         if(getCutoff == 1){
 396                                 if(label != "unique"){
 397                                         origCutoff = label;
 398                                         convert(label, cutoff);
 399                                         if(hard == 0){  cutoff += (0.49 / double(precision));   }
 400                                 }
 401                                 else{
 402                                         origCutoff = "unique";
 403                                         cutoff = 0.0000;
 404                                 }
 405                         }
 406
 407                         cout << label << endl;
 408
 409                         while(columnFile){
 410                                 columnFile >> seqNameA >> seqNameB >> distance;
 411                                 if(seqNameA < seqNameB) {       seqPairString = seqNameA + '\t' + seqNameB;     }
 412                                 else                                    {       seqPairString = seqNameB + '\t' + seqNameA;     }
 413
 414                                 set<string>::iterator it = seqPairSet.find(seqPairString);
 415
 416                                 if(distance <= cutoff){
 417                                         if(it != seqPairSet.end()){
 418                                                 truePositives++;
 419                                                 seqPairSet.erase(it);
 420                                         }
 421                                         else{
 422                                                 falseNegatives++;
 423                                         }
 424                                         trueNegatives--;
 425                                 }
 426                                 else if(it != seqPairSet.end()){
 427                                         falsePositives++;
 428                                         trueNegatives--;
 429                                         seqPairSet.erase(it);
 430                                 }
 431
 432                                 m->gobble(columnFile);
 433                         }
 434                         falsePositives += seqPairSet.size();
 435
 436                         outputStatistics(label, origCutoff);
 437                 }
 438         }
 439         catch(exception& e) {
 440                 m->errorOut(e, "SensSpecCommand", "processColumn");
 441                 exit(1);
 442         }
 443 }
 444
 445 //***************************************************************************************************************
 446
 447 void SensSpecCommand::setUpOutput(){
 448         try{
 449                 ofstream sensSpecFile;
 450                 m->openOutputFile(sensSpecFileName, sensSpecFile);
 451
 452                 sensSpecFile << "label\tcutoff\ttp\ttn\tfp\tfn\tsensitivity\tspecificity\tppv\tnpv\tfdr\taccuracy\tmcc\tf1score\n";
 453
 454                 sensSpecFile.close();
 455         }
 456         catch(exception& e) {
 457                 m->errorOut(e, "SensSpecCommand", "setUpOutput");
 458                 exit(1);
 459         }
 460 }
 461
 462 //***************************************************************************************************************
 463
 464 void SensSpecCommand::outputStatistics(string label, string cutoff){
 465         try{
 466                 double tp = (double) truePositives;
 467                 double fp = (double) falsePositives;
 468                 double tn = (double) trueNegatives;
 469                 double fn = (double) falseNegatives;
 470
 471                 double p = tp + fn;
 472                 double n = fp + tn;
 473                 double pPrime = tp + fp;
 474                 double nPrime = tn + fn;
 475
 476                 double sensitivity = tp / p;
 477                 double specificity = tn / n;
 478                 double positivePredictiveValue = tp / pPrime;
 479                 double negativePredictiveValue = tn / nPrime;
 480                 double falseDiscoveryRate = fp / pPrime;
 481
 482                 double accuracy = (tp + tn) / (p + n);
 483                 double matthewsCorrCoef = (tp * tn - fp * fn) / sqrt(p * n * pPrime * nPrime);  if(p == 0 || n == 0){   matthewsCorrCoef = 0;   }
 484                 double f1Score = 2.0 * tp / (p + pPrime);
 485
 486
 487                 if(p == 0)                      {       sensitivity = 0;        matthewsCorrCoef = 0;   }
 488                 if(n == 0)                      {       specificity = 0;        matthewsCorrCoef = 0;   }
 489                 if(p + n == 0)          {       accuracy = 0;                                                           }
 490                 if(p + pPrime == 0)     {       f1Score = 0;                                                            }
 491                 if(pPrime == 0)         {       positivePredictiveValue = 0;    falseDiscoveryRate = 0; matthewsCorrCoef = 0;   }
 492                 if(nPrime == 0)         {       negativePredictiveValue = 0;    matthewsCorrCoef = 0;                                                   }
 493
 494                 ofstream sensSpecFile;
 495                 m->openOutputFileAppend(sensSpecFileName, sensSpecFile);
 496
 497                 sensSpecFile << label << '\t' << cutoff << '\t';
 498                 sensSpecFile << truePositives << '\t' << trueNegatives << '\t' << falsePositives << '\t' << falseNegatives << '\t';
 499                 sensSpecFile << setprecision(4);
 500                 sensSpecFile << sensitivity << '\t' << specificity << '\t' << positivePredictiveValue << '\t' << negativePredictiveValue << '\t';
 501                 sensSpecFile << falseDiscoveryRate << '\t' << accuracy << '\t' << matthewsCorrCoef << '\t' << f1Score << endl;
 502
 503                 sensSpecFile.close();
 504         }
 505         catch(exception& e) {
 506                 m->errorOut(e, "SensSpecCommand", "outputStatistics");
 507                 exit(1);
 508         }
 509 }
 510
 511 //***************************************************************************************************************
 512
 513
 514