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                         else { m->setListFile(listFile); }
 138
 139                         phylipfile = validParameter.validFile(parameters, "phylip", true);
 140                         if (phylipfile == "not found") { phylipfile = "";  }
 141                         else if (phylipfile == "not open") { abort = true; }
 142                         else { distFile = phylipfile; format = "phylip"; m->setPhylipFile(phylipfile);  }
 143
 144                         columnfile = validParameter.validFile(parameters, "column", true);
 145                         if (columnfile == "not found") { columnfile = ""; }
 146                         else if (columnfile == "not open") { abort = true; }
 147                         else { distFile = columnfile; format = "column";   m->setColumnFile(columnfile); }
 148
 149                         if ((phylipfile == "") && (columnfile == "")) { //is there are current file available for either of these?
 150                                 //give priority to column, then phylip
 151                                 columnfile = m->getColumnFile();
 152                                 if (columnfile != "") {  distFile = columnfile; format = "column";  m->mothurOut("Using " + columnfile + " as input file for the column parameter."); m->mothurOutEndLine(); }
 153                                 else {
 154                                         phylipfile = m->getPhylipFile();
 155                                         if (phylipfile != "") {  distFile = phylipfile; format = "phylip"; m->mothurOut("Using " + phylipfile + " as input file for the phylip parameter."); m->mothurOutEndLine(); }
 156                                         else {
 157                                                 m->mothurOut("No valid current files. You must provide a phylip or column file."); m->mothurOutEndLine();
 158                                                 abort = true;
 159                                         }
 160                                 }
 161                         }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; }
 162
 163
 164                         //if the user changes the output directory command factory will send this info to us in the output parameter
 165                         outputDir = validParameter.validFile(parameters, "outputdir", false);
 166                         if (outputDir == "not found"){
 167                                 outputDir = "";
 168                                 outputDir += m->hasPath(listFile); //if user entered a file with a path then preserve it
 169                         }
 170
 171                         //check for optional parameter and set defaults
 172                         // ...at some point should added some additional type checking...
 173                         temp = validParameter.validFile(parameters, "hard", false);
 174                         if (temp == "not found"){       hard = 0;       }
 175                         else if(!m->isTrue(temp))       {       hard = 0;       }
 176                         else if(m->isTrue(temp))        {       hard = 1;       }
 177
 178 //                      temp = validParameter.validFile(parameters, "name", true);
 179 //                      if (temp == "not found")        {       nameFile = "";          }
 180 //                      else if(temp == "not open")     {       abort = true;           }
 181 //                      else                                            {       nameFile = temp;        }
 182 //                      cout << "name:\t" << nameFile << endl;
 183
 184                         temp = validParameter.validFile(parameters, "cutoff", false);           if (temp == "not found") { temp = "-1.00"; }
 185                         convert(temp, cutoff);
 186 //                      cout << cutoff << endl;
 187
 188                         temp = validParameter.validFile(parameters, "precision", false);        if (temp == "not found") { temp = "100"; }
 189                         convert(temp, precision);
 190 //                      cout << precision << endl;
 191
 192                         lineLabel = validParameter.validFile(parameters, "label", false);       if (lineLabel == "not found") { lineLabel = ""; }
 193
 194                         sensSpecFileName = outputDir + m->getRootName(m->getSimpleName(listFile)) + ".sensspec";
 195                 }
 196         }
 197         catch(exception& e) {
 198                 m->errorOut(e, "SensSpecCommand", "SensSpecCommand");
 199                 exit(1);
 200         }
 201 }
 202 //***************************************************************************************************************
 203
 204 int SensSpecCommand::execute(){
 205         try{
 206                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 207
 208                 setUpOutput();
 209                 outputNames.push_back(sensSpecFileName); outputTypes["sensspec"].push_back(sensSpecFileName);
 210                 if(format == "phylip")          {       processPhylip();        }
 211                 else if(format == "column")     {       processColumn();        }
 212
 213                 m->mothurOutEndLine();
 214                 m->mothurOut("Output File Name: "); m->mothurOutEndLine();
 215                 m->mothurOut(sensSpecFileName); m->mothurOutEndLine();
 216                 m->mothurOutEndLine();
 217
 218
 219                 return 0;
 220         }
 221         catch(exception& e) {
 222                 m->errorOut(e, "SensSpecCommand", "execute");
 223                 exit(1);
 224         }
 225 }
 226
 227 //***************************************************************************************************************
 228
 229 void SensSpecCommand::processPhylip(){
 230         try{
 231                 //probably need some checking to confirm that the names in the distance matrix are the same as those in the list file
 232
 233                 ifstream inputListFile;
 234                 m->openInputFile(listFile, inputListFile);
 235
 236                 string origCutoff = "";
 237                 bool getCutoff = 0;
 238                 if(cutoff == -1.00)     {       getCutoff = 1;                                                                                                                  }
 239                 else                            {       origCutoff = toString(cutoff);  cutoff += (0.49 / double(precision));   }
 240
 241                 string label;
 242                 int numOTUs;
 243
 244                 map<string, int> seqMap;
 245                 string seqList;
 246
 247                 while(inputListFile){
 248                         inputListFile >> label >> numOTUs;
 249                         for(int i=0;i<numOTUs;i++){
 250                                 inputListFile >> seqList;
 251                                 int seqListLength = seqList.length();
 252                                 string seqName = "";
 253                                 for(int j=0;j<seqListLength;j++){
 254
 255                                         if(seqList[j] == ','){
 256                                                 seqMap[seqName] = i;
 257                                                 seqName = "";
 258                                         }
 259                                         else{
 260                                                 seqName += seqList[j];
 261                                         }
 262
 263                                 }
 264                                 seqMap[seqName] = i;
 265                         }
 266                         m->gobble(inputListFile);
 267
 268                         int lNumSeqs = seqMap.size();
 269                         int pNumSeqs = 0;
 270
 271                         ifstream phylipFile;
 272                         m->openInputFile(distFile, phylipFile);
 273                         phylipFile >> pNumSeqs;
 274                         if(pNumSeqs != lNumSeqs){       cout << "numSeq mismatch!" << endl;     }
 275
 276                         string seqName;
 277                         double distance;
 278                         vector<int> otuIndices(lNumSeqs, -1);
 279
 280                         truePositives = 0;
 281                         falsePositives = 0;
 282                         trueNegatives = 0;
 283                         falseNegatives = 0;
 284
 285                         if(getCutoff == 1){
 286                                 if(label != "unique"){
 287                                         origCutoff = label;
 288                                         convert(label, cutoff);
 289                                         if(hard == 0){  cutoff += (0.49 / double(precision));   }
 290                                 }
 291                                 else{
 292                                         origCutoff = "unique";
 293                                         cutoff = 0.0000;
 294                                 }
 295                         }
 296
 297                         cout << label << endl;
 298
 299                         for(int i=0;i<lNumSeqs;i++){
 300                                 phylipFile >> seqName;
 301                                 otuIndices[i] = seqMap[seqName];
 302
 303                                 for(int j=0;j<i;j++){
 304                                         phylipFile >> distance;
 305
 306                                         if(distance <= cutoff){
 307                                                 if(otuIndices[i] == otuIndices[j])      {       truePositives++;        }
 308                                                 else                                                            {       falseNegatives++;       }
 309                                         }
 310                                         else{
 311                                                 if(otuIndices[i] == otuIndices[j])      {       falsePositives++;       }
 312                                                 else                                                            {       trueNegatives++;        }
 313                                         }
 314                                 }
 315                         }
 316                         phylipFile.close();
 317
 318                         outputStatistics(label, origCutoff);
 319                 }
 320                 inputListFile.close();
 321
 322         }
 323         catch(exception& e) {
 324                 m->errorOut(e, "SensSpecCommand", "processPhylip");
 325                 exit(1);
 326         }
 327 }
 328
 329 //***************************************************************************************************************
 330
 331 void SensSpecCommand::processColumn(){
 332         try{
 333                 ifstream inputListFile;
 334                 m->openInputFile(listFile, inputListFile);
 335
 336                 string origCutoff = "";
 337                 bool getCutoff = 0;
 338                 if(cutoff == -1.00)     {       getCutoff = 1;                                                                                                                  }
 339                 else                            {       origCutoff = toString(cutoff);  cutoff += (0.49 / double(precision));   }
 340
 341                 set<string> seqPairSet;
 342
 343                 string label, seqList;
 344                 int numOTUs;
 345                 int numSeqs;
 346
 347                 while(inputListFile){
 348                         numSeqs = 0;
 349
 350                         inputListFile >> label >> numOTUs;
 351                         for(int i=0;i<numOTUs;i++){
 352
 353                                 vector<string> seqNameVector;
 354
 355                                 inputListFile >> seqList;
 356                                 int seqListLength = seqList.length();
 357                                 string seqName = "";
 358                                 for(int j=0;j<seqListLength;j++){
 359
 360                                         if(seqList[j] == ','){
 361                                                 seqNameVector.push_back(seqName);
 362                                                 seqName = "";
 363                                         }
 364                                         else{
 365                                                 seqName += seqList[j];
 366                                         }
 367                                 }
 368                                 seqNameVector.push_back(seqName);
 369
 370                                 numSeqs += seqNameVector.size();
 371
 372                                 int numSeqsInOTU = seqNameVector.size();
 373                                 for(int j=0;j<numSeqsInOTU;j++){
 374                                         string seqPairString = "";
 375                                         for(int k=0;k<j;k++){
 376                                                 if(seqNameVector[j] < seqNameVector[k]) {       seqPairString = seqNameVector[j] + '\t' + seqNameVector[k];     }
 377                                                 else                                                                    {       seqPairString = seqNameVector[k] + '\t' + seqNameVector[j];     }
 378                                                 seqPairSet.insert(seqPairString);
 379                                         }
 380                                 }
 381                         }
 382                         m->gobble(inputListFile);
 383
 384                         int numDists = (numSeqs * (numSeqs-1) / 2);
 385
 386                         ifstream columnFile;
 387                         m->openInputFile(distFile, columnFile);
 388                         string seqNameA, seqNameB, seqPairString;
 389                         double distance;
 390
 391                         truePositives = 0;
 392                         falsePositives = 0;
 393                         trueNegatives = numDists;
 394                         falseNegatives = 0;
 395
 396                         if(getCutoff == 1){
 397                                 if(label != "unique"){
 398                                         origCutoff = label;
 399                                         convert(label, cutoff);
 400                                         if(hard == 0){  cutoff += (0.49 / double(precision));   }
 401                                 }
 402                                 else{
 403                                         origCutoff = "unique";
 404                                         cutoff = 0.0000;
 405                                 }
 406                         }
 407
 408                         cout << label << endl;
 409
 410                         while(columnFile){
 411                                 columnFile >> seqNameA >> seqNameB >> distance;
 412                                 if(seqNameA < seqNameB) {       seqPairString = seqNameA + '\t' + seqNameB;     }
 413                                 else                                    {       seqPairString = seqNameB + '\t' + seqNameA;     }
 414
 415                                 set<string>::iterator it = seqPairSet.find(seqPairString);
 416
 417                                 if(distance <= cutoff){
 418                                         if(it != seqPairSet.end()){
 419                                                 truePositives++;
 420                                                 seqPairSet.erase(it);
 421                                         }
 422                                         else{
 423                                                 falseNegatives++;
 424                                         }
 425                                         trueNegatives--;
 426                                 }
 427                                 else if(it != seqPairSet.end()){
 428                                         falsePositives++;
 429                                         trueNegatives--;
 430                                         seqPairSet.erase(it);
 431                                 }
 432
 433                                 m->gobble(columnFile);
 434                         }
 435                         falsePositives += seqPairSet.size();
 436
 437                         outputStatistics(label, origCutoff);
 438                 }
 439         }
 440         catch(exception& e) {
 441                 m->errorOut(e, "SensSpecCommand", "processColumn");
 442                 exit(1);
 443         }
 444 }
 445
 446 //***************************************************************************************************************
 447
 448 void SensSpecCommand::setUpOutput(){
 449         try{
 450                 ofstream sensSpecFile;
 451                 m->openOutputFile(sensSpecFileName, sensSpecFile);
 452
 453                 sensSpecFile << "label\tcutoff\ttp\ttn\tfp\tfn\tsensitivity\tspecificity\tppv\tnpv\tfdr\taccuracy\tmcc\tf1score\n";
 454
 455                 sensSpecFile.close();
 456         }
 457         catch(exception& e) {
 458                 m->errorOut(e, "SensSpecCommand", "setUpOutput");
 459                 exit(1);
 460         }
 461 }
 462
 463 //***************************************************************************************************************
 464
 465 void SensSpecCommand::outputStatistics(string label, string cutoff){
 466         try{
 467                 double tp = (double) truePositives;
 468                 double fp = (double) falsePositives;
 469                 double tn = (double) trueNegatives;
 470                 double fn = (double) falseNegatives;
 471
 472                 double p = tp + fn;
 473                 double n = fp + tn;
 474                 double pPrime = tp + fp;
 475                 double nPrime = tn + fn;
 476
 477                 double sensitivity = tp / p;
 478                 double specificity = tn / n;
 479                 double positivePredictiveValue = tp / pPrime;
 480                 double negativePredictiveValue = tn / nPrime;
 481                 double falseDiscoveryRate = fp / pPrime;
 482
 483                 double accuracy = (tp + tn) / (p + n);
 484                 double matthewsCorrCoef = (tp * tn - fp * fn) / sqrt(p * n * pPrime * nPrime);  if(p == 0 || n == 0){   matthewsCorrCoef = 0;   }
 485                 double f1Score = 2.0 * tp / (p + pPrime);
 486
 487
 488                 if(p == 0)                      {       sensitivity = 0;        matthewsCorrCoef = 0;   }
 489                 if(n == 0)                      {       specificity = 0;        matthewsCorrCoef = 0;   }
 490                 if(p + n == 0)          {       accuracy = 0;                                                           }
 491                 if(p + pPrime == 0)     {       f1Score = 0;                                                            }
 492                 if(pPrime == 0)         {       positivePredictiveValue = 0;    falseDiscoveryRate = 0; matthewsCorrCoef = 0;   }
 493                 if(nPrime == 0)         {       negativePredictiveValue = 0;    matthewsCorrCoef = 0;                                                   }
 494
 495                 ofstream sensSpecFile;
 496                 m->openOutputFileAppend(sensSpecFileName, sensSpecFile);
 497
 498                 sensSpecFile << label << '\t' << cutoff << '\t';
 499                 sensSpecFile << truePositives << '\t' << trueNegatives << '\t' << falsePositives << '\t' << falseNegatives << '\t';
 500                 sensSpecFile << setprecision(4);
 501                 sensSpecFile << sensitivity << '\t' << specificity << '\t' << positivePredictiveValue << '\t' << negativePredictiveValue << '\t';
 502                 sensSpecFile << falseDiscoveryRate << '\t' << accuracy << '\t' << matthewsCorrCoef << '\t' << f1Score << endl;
 503
 504                 sensSpecFile.close();
 505         }
 506         catch(exception& e) {
 507                 m->errorOut(e, "SensSpecCommand", "outputStatistics");
 508                 exit(1);
 509         }
 510 }
 511
 512 //***************************************************************************************************************
 513
 514
 515