summaryqualcommand.cpp

   1 /*
   2  *  summaryqualcommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 11/28/11.
   6  *  Copyright 2011 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "summaryqualcommand.h"
  11 #include "counttable.h"
  12
  13 //**********************************************************************************************************************
  14 vector<string> SummaryQualCommand::setParameters(){
  15         try {
  16                 CommandParameter pqual("qfile", "InputTypes", "", "", "none", "none", "none","summary",false,true,true); parameters.push_back(pqual);
  17                 CommandParameter pname("name", "InputTypes", "", "", "namecount", "none", "none","",false,false,true); parameters.push_back(pname);
  18         CommandParameter pcount("count", "InputTypes", "", "", "namecount", "none", "none","",false,false,true); parameters.push_back(pcount);
  19                 CommandParameter pprocessors("processors", "Number", "", "1", "", "", "","",false,false,true); parameters.push_back(pprocessors);
  20                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(pinputdir);
  21                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(poutputdir);
  22
  23                 vector<string> myArray;
  24                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  25                 return myArray;
  26         }
  27         catch(exception& e) {
  28                 m->errorOut(e, "SummaryQualCommand", "setParameters");
  29                 exit(1);
  30         }
  31 }
  32 //**********************************************************************************************************************
  33 string SummaryQualCommand::getHelpString(){
  34         try {
  35                 string helpString = "";
  36                 helpString += "The summary.qual command reads a quality file and an optional name or count file, and summarizes the quality information.\n";
  37                 helpString += "The summary.tax command parameters are qfile, name, count and processors. qfile is required, unless you have a valid current quality file.\n";
  38                 helpString += "The name parameter allows you to enter a name file associated with your quality file. \n";
  39         helpString += "The count parameter allows you to enter a count file associated with your quality file. \n";
  40                 helpString += "The summary.qual command should be in the following format: \n";
  41                 helpString += "summary.qual(qfile=yourQualityFile) \n";
  42                 helpString += "Note: No spaces between parameter labels (i.e. qfile), '=' and parameters (i.e.yourQualityFile).\n";
  43                 return helpString;
  44         }
  45         catch(exception& e) {
  46                 m->errorOut(e, "SummaryQualCommand", "getHelpString");
  47                 exit(1);
  48         }
  49 }
  50 //**********************************************************************************************************************
  51 string SummaryQualCommand::getOutputPattern(string type) {
  52     try {
  53         string pattern = "";
  54
  55         if (type == "summary") {  pattern = "[filename],qual.summary"; }
  56         else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true;  }
  57
  58         return pattern;
  59     }
  60     catch(exception& e) {
  61         m->errorOut(e, "SummaryQualCommand", "getOutputPattern");
  62         exit(1);
  63     }
  64 }
  65 //**********************************************************************************************************************
  66 SummaryQualCommand::SummaryQualCommand(){
  67         try {
  68                 abort = true; calledHelp = true;
  69                 setParameters();
  70                 vector<string> tempOutNames;
  71                 outputTypes["summary"] = tempOutNames;
  72         }
  73         catch(exception& e) {
  74                 m->errorOut(e, "SummaryQualCommand", "SummaryQualCommand");
  75                 exit(1);
  76         }
  77 }
  78 //***************************************************************************************************************
  79
  80 SummaryQualCommand::SummaryQualCommand(string option)  {
  81         try {
  82                 abort = false; calledHelp = false;
  83
  84                 //allow user to run help
  85                 if(option == "help") { help(); abort = true; calledHelp = true; }
  86                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  87
  88                 else {
  89                         vector<string> myArray = setParameters();
  90
  91                         OptionParser parser(option);
  92                         map<string,string> parameters = parser.getParameters();
  93
  94                         ValidParameters validParameter;
  95                         map<string,string>::iterator it;
  96
  97                         //check to make sure all parameters are valid for command
  98                         for (it = parameters.begin(); it != parameters.end(); it++) {
  99                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 100                         }
 101
 102                         //if the user changes the input directory command factory will send this info to us in the output parameter
 103                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 104                         if (inputDir == "not found"){   inputDir = "";          }
 105                         else {
 106                                 string path;
 107                                 it = parameters.find("qfile");
 108                                 //user has given a template file
 109                                 if(it != parameters.end()){
 110                                         path = m->hasPath(it->second);
 111                                         //if the user has not given a path then, add inputdir. else leave path alone.
 112                                         if (path == "") {       parameters["qfile"] = inputDir + it->second;            }
 113                                 }
 114
 115                                 it = parameters.find("name");
 116                                 //user has given a template file
 117                                 if(it != parameters.end()){
 118                                         path = m->hasPath(it->second);
 119                                         //if the user has not given a path then, add inputdir. else leave path alone.
 120                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 121                                 }
 122
 123                 it = parameters.find("count");
 124                                 //user has given a template file
 125                                 if(it != parameters.end()){
 126                                         path = m->hasPath(it->second);
 127                                         //if the user has not given a path then, add inputdir. else leave path alone.
 128                                         if (path == "") {       parameters["count"] = inputDir + it->second;            }
 129                                 }
 130                         }
 131
 132                         //initialize outputTypes
 133                         vector<string> tempOutNames;
 134                         outputTypes["summary"] = tempOutNames;
 135
 136                         //check for required parameters
 137                         qualfile = validParameter.validFile(parameters, "qfile", true);
 138                         if (qualfile == "not open") { qualfile = ""; abort = true; }
 139                         else if (qualfile == "not found") {
 140                                 qualfile = m->getQualFile();
 141                                 if (qualfile != "") { m->mothurOut("Using " + qualfile + " as input file for the qfile parameter."); m->mothurOutEndLine(); }
 142                                 else {  m->mothurOut("You have no current quality file and the qfile parameter is required."); m->mothurOutEndLine(); abort = true; }
 143                         }else { m->setQualFile(qualfile); }
 144
 145                         namefile = validParameter.validFile(parameters, "name", true);
 146                         if (namefile == "not open") { namefile = ""; abort = true; }
 147                         else if (namefile == "not found") { namefile = "";  }
 148                         else { m->setNameFile(namefile); }
 149
 150             countfile = validParameter.validFile(parameters, "count", true);
 151                         if (countfile == "not open") { abort = true; countfile = ""; }
 152                         else if (countfile == "not found") { countfile = ""; }
 153                         else { m->setCountTableFile(countfile); }
 154
 155             if ((countfile != "") && (namefile != "")) { m->mothurOut("You must enter ONLY ONE of the following: count or name."); m->mothurOutEndLine(); abort = true; }
 156
 157                         //if the user changes the output directory command factory will send this info to us in the output parameter
 158                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 159                                 outputDir = "";
 160                                 outputDir += m->hasPath(qualfile); //if user entered a file with a path then preserve it
 161                         }
 162
 163                         string temp = validParameter.validFile(parameters, "processors", false);        if (temp == "not found"){       temp = m->getProcessors();      }
 164                         m->setProcessors(temp);
 165                         m->mothurConvert(temp, processors);
 166
 167
 168                         if (countfile == "") {
 169                 if (namefile == "") {
 170                     vector<string> files; files.push_back(qualfile);
 171                     parser.getNameFile(files);
 172                 }
 173             }
 174                 }
 175         }
 176         catch(exception& e) {
 177                 m->errorOut(e, "SummaryQualCommand", "SummaryQualCommand");
 178                 exit(1);
 179         }
 180 }
 181 //***************************************************************************************************************
 182 int SummaryQualCommand::execute(){
 183         try{
 184
 185                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 186
 187                 int start = time(NULL);
 188                 int numSeqs = 0;
 189
 190                 vector<int> position;
 191                 vector<int> averageQ;
 192                 vector< vector<int> > scores;
 193
 194                 if (m->control_pressed) { return 0; }
 195
 196                 if (namefile != "") { nameMap = m->readNames(namefile); }
 197                 else if (countfile != "") {
 198             CountTable ct;
 199             ct.readTable(countfile, false, false);
 200             nameMap = ct.getNameMap();
 201         }
 202
 203                 vector<unsigned long long> positions;
 204 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 205                 positions = m->divideFile(qualfile, processors);
 206                 for (int i = 0; i < (positions.size()-1); i++) {        lines.push_back(linePair(positions[i], positions[(i+1)]));      }
 207 #else
 208                 if (processors == 1) {
 209                         lines.push_back(linePair(0, 1000));
 210                 }else {
 211                         positions = m->setFilePosFasta(qualfile, numSeqs);
 212             if (positions.size() < processors) { processors = positions.size(); }
 213
 214                         //figure out how many sequences you have to process
 215                         int numSeqsPerProcessor = numSeqs / processors;
 216                         for (int i = 0; i < processors; i++) {
 217                                 int startIndex =  i * numSeqsPerProcessor;
 218                                 if(i == (processors - 1)){      numSeqsPerProcessor = numSeqs - i * numSeqsPerProcessor;        }
 219                                 lines.push_back(linePair(positions[startIndex], numSeqsPerProcessor));
 220                         }
 221                 }
 222 #endif
 223
 224
 225                 if(processors == 1){ numSeqs = driverCreateSummary(position, averageQ, scores, qualfile, lines[0]);  }
 226                 else{  numSeqs = createProcessesCreateSummary(position, averageQ, scores, qualfile);  }
 227
 228                 if (m->control_pressed) {  return 0; }
 229
 230                 //print summary file
 231         map<string, string> variables;
 232                 variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(qualfile));
 233                 string summaryFile = getOutputFileName("summary",variables);
 234                 printQual(summaryFile, position, averageQ, scores);
 235
 236                 if (m->control_pressed) {  m->mothurRemove(summaryFile); return 0; }
 237
 238                 //output results to screen
 239                 cout.setf(ios::fixed, ios::floatfield); cout.setf(ios::showpoint);
 240                 m->mothurOutEndLine();
 241                 m->mothurOut("Position\tNumSeqs\tAverageQ"); m->mothurOutEndLine();
 242                 for (int i = 0; i < position.size(); i+=100) {
 243                         float average = averageQ[i] / (float) position[i];
 244                         cout << i << '\t' << position[i] << '\t' << average;
 245                         m->mothurOutJustToLog(toString(i) + "\t" + toString(position[i]) + "\t" + toString(average)); m->mothurOutEndLine();
 246                 }
 247
 248                 m->mothurOutEndLine();
 249                 m->mothurOut("It took " + toString(time(NULL) - start) + " secs to create the summary file for " + toString(numSeqs) + " sequences."); m->mothurOutEndLine(); m->mothurOutEndLine();
 250                 m->mothurOutEndLine();
 251                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 252                 m->mothurOut(summaryFile); m->mothurOutEndLine();       outputNames.push_back(summaryFile); outputTypes["summary"].push_back(summaryFile);
 253                 m->mothurOutEndLine();
 254
 255                 return 0;
 256         }
 257         catch(exception& e) {
 258                 m->errorOut(e, "SummaryQualCommand", "execute");
 259                 exit(1);
 260         }
 261 }
 262 /**************************************************************************************/
 263 int SummaryQualCommand::driverCreateSummary(vector<int>& position, vector<int>& averageQ, vector< vector<int> >& scores, string filename, linePair filePos) {
 264         try {
 265                 ifstream in;
 266                 m->openInputFile(filename, in);
 267
 268                 in.seekg(filePos.start);
 269
 270                 bool done = false;
 271                 int count = 0;
 272
 273                 while (!done) {
 274
 275                         if (m->control_pressed) { in.close(); return 1; }
 276
 277                         QualityScores current(in); m->gobble(in);
 278
 279                         if (current.getName() != "") {
 280
 281                                 int num = 1;
 282                                 if ((namefile != "") || (countfile != "")) {
 283                                         //make sure this sequence is in the namefile, else error
 284                                         map<string, int>::iterator it = nameMap.find(current.getName());
 285
 286                                         if (it == nameMap.end()) { m->mothurOut("[ERROR]: " + current.getName() + " is not in your namefile, please correct."); m->mothurOutEndLine(); m->control_pressed = true; }
 287                                         else { num = it->second; }
 288                                 }
 289
 290                                 vector<int> thisScores = current.getQualityScores();
 291
 292                                 //resize to num of positions setting number of seqs with that size to 1
 293                                 if (position.size() < thisScores.size()) { position.resize(thisScores.size(), 0); }
 294                                 if (averageQ.size() < thisScores.size()) { averageQ.resize(thisScores.size(), 0); }
 295                                 if (scores.size() < thisScores.size()) {
 296                                         scores.resize(thisScores.size());
 297                                         for (int i = 0; i < scores.size(); i++) { scores[i].resize(41, 0); }
 298                                 }
 299
 300                                 //increase counts of number of seqs with this position
 301                                 //average is really the total, we will average in execute
 302                                 for (int i = 0; i < thisScores.size(); i++) {
 303                                         position[i] += num;
 304                                         averageQ[i] += (thisScores[i] * num); //weighting for namesfile
 305                                         if (thisScores[i] > 40) { m->mothurOut("[ERROR]: " + current.getName() + " has a quality scores of " + toString(thisScores[i]) + ", expecting values to be less than 40."); m->mothurOutEndLine(); m->control_pressed = true; }
 306                                         else { scores[i][thisScores[i]] += num; }
 307                                 }
 308
 309                                 count += num;
 310                         }
 311
 312 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 313                         unsigned long long pos = in.tellg();
 314                         if ((pos == -1) || (pos >= filePos.end)) { break; }
 315 #else
 316                         if (in.eof()) { break; }
 317 #endif
 318                 }
 319
 320                 in.close();
 321
 322                 return count;
 323         }
 324         catch(exception& e) {
 325                 m->errorOut(e, "SummaryQualCommand", "driverCreateSummary");
 326                 exit(1);
 327         }
 328 }
 329 /**************************************************************************************************/
 330 int SummaryQualCommand::createProcessesCreateSummary(vector<int>& position, vector<int>& averageQ, vector< vector<int> >& scores, string filename) {
 331         try {
 332                 int process = 1;
 333                 int numSeqs = 0;
 334                 processIDS.clear();
 335
 336 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 337
 338                 //loop through and create all the processes you want
 339                 while (process != processors) {
 340                         int pid = fork();
 341
 342                         if (pid > 0) {
 343                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 344                                 process++;
 345                         }else if (pid == 0){
 346                                 numSeqs = driverCreateSummary(position, averageQ, scores, qualfile, lines[process]);
 347
 348                                 //pass numSeqs to parent
 349                                 ofstream out;
 350                                 string tempFile = qualfile + toString(getpid()) + ".num.temp";
 351                                 m->openOutputFile(tempFile, out);
 352
 353                                 out << numSeqs << endl;
 354                                 out << position.size() << endl;
 355                                 for (int k = 0; k < position.size(); k++)                       {               out << position[k] << '\t'; }  out << endl;
 356                                 for (int k = 0; k < averageQ.size(); k++)                       {               out << averageQ[k] << '\t'; }  out << endl;
 357                                 for (int k = 0; k < scores.size(); k++) {
 358                                         for (int j = 0; j < 41; j++) {
 359                                                 out << scores[k][j] << '\t';
 360                                         }
 361                                         out << endl;
 362                                 }
 363                                 out << endl;
 364
 365                                 out.close();
 366
 367                                 exit(0);
 368                         }else {
 369                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 370                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 371                                 exit(0);
 372                         }
 373                 }
 374
 375                 //do your part
 376                 numSeqs = driverCreateSummary(position, averageQ, scores, qualfile, lines[0]);
 377
 378                 //force parent to wait until all the processes are done
 379                 for (int i=0;i<processIDS.size();i++) {
 380                         int temp = processIDS[i];
 381                         wait(&temp);
 382                 }
 383
 384                 //parent reads in and combine Filter info
 385                 for (int i = 0; i < processIDS.size(); i++) {
 386                         string tempFilename = qualfile + toString(processIDS[i]) + ".num.temp";
 387                         ifstream in;
 388                         m->openInputFile(tempFilename, in);
 389
 390                         int temp, tempNum;
 391                         in >> tempNum; m->gobble(in); numSeqs += tempNum;
 392                         in >> tempNum; m->gobble(in);
 393
 394                         if (position.size() < tempNum) { position.resize(tempNum, 0); }
 395                         if (averageQ.size() < tempNum) { averageQ.resize(tempNum, 0); }
 396                         if (scores.size() < tempNum) {
 397                                 scores.resize(tempNum);
 398                                 for (int i = 0; i < scores.size(); i++) { scores[i].resize(41, 0); }
 399                         }
 400
 401                         for (int k = 0; k < tempNum; k++)                       {               in >> temp; position[k] += temp;                        }               m->gobble(in);
 402                         for (int k = 0; k < tempNum; k++)                       {               in >> temp; averageQ[k] += temp;                }               m->gobble(in);
 403                         for (int k = 0; k < tempNum; k++)                       {
 404                                 for (int j = 0; j < 41; j++) {
 405                                         in >> temp; scores[k][j] += temp;
 406                                         m->gobble(in);
 407                                 }
 408                         }
 409
 410                         in.close();
 411                         m->mothurRemove(tempFilename);
 412                 }
 413
 414 #else
 415                 //////////////////////////////////////////////////////////////////////////////////////////////////////
 416                 //Windows version shared memory, so be careful when passing variables through the seqSumQualData struct.
 417                 //Above fork() will clone, so memory is separate, but that's not the case with windows,
 418                 //Taking advantage of shared memory to pass results vectors.
 419                 //////////////////////////////////////////////////////////////////////////////////////////////////////
 420
 421                 vector<seqSumQualData*> pDataArray;
 422                 DWORD   dwThreadIdArray[processors];
 423                 HANDLE  hThreadArray[processors];
 424
 425         bool hasNameMap = false;
 426         if ((namefile !="") || (countfile != "")) { hasNameMap = true; }
 427
 428                 //Create processor worker threads.
 429                 for( int i=0; i<processors; i++ ){
 430
 431                         // Allocate memory for thread data.
 432                         seqSumQualData* tempSum = new seqSumQualData(filename, m, lines[i].start, lines[i].end, hasNameMap, nameMap);
 433                         pDataArray.push_back(tempSum);
 434                         processIDS.push_back(i);
 435
 436                         hThreadArray[i] = CreateThread(NULL, 0, MySeqSumQualThreadFunction, pDataArray[i], 0, &dwThreadIdArray[i]);
 437                 }
 438
 439                 //Wait until all threads have terminated.
 440                 WaitForMultipleObjects(processors, hThreadArray, TRUE, INFINITE);
 441
 442                 //Close all thread handles and free memory allocations.
 443                 for(int i=0; i < pDataArray.size(); i++){
 444                         numSeqs += pDataArray[i]->numSeqs;
 445             if (pDataArray[i]->count != pDataArray[i]->end) {
 446                 m->mothurOut("[ERROR]: process " + toString(i) + " only processed " + toString(pDataArray[i]->count) + " of " + toString(pDataArray[i]->end) + " sequences assigned to it, quitting. \n"); m->control_pressed = true;
 447             }
 448             int tempNum = pDataArray[i]->position.size();
 449             if (position.size() < tempNum) { position.resize(tempNum, 0); }
 450                         if (averageQ.size() < tempNum) { averageQ.resize(tempNum, 0); }
 451                         if (scores.size() < tempNum) {
 452                                 scores.resize(tempNum);
 453                                 for (int i = 0; i < scores.size(); i++) { scores[i].resize(41, 0); }
 454                         }
 455
 456             for (int k = 0; k < tempNum; k++)                   {                position[k]    +=  pDataArray[i]->position[k];         }
 457                         for (int k = 0; k < tempNum; k++)                       {                averageQ[k]    +=  pDataArray[i]->averageQ[k];         }
 458                         for (int k = 0; k < tempNum; k++)                       {       for (int j = 0; j < 41; j++) {  scores[k][j] += pDataArray[i]->scores[k][j];   }        }
 459
 460                         CloseHandle(hThreadArray[i]);
 461                         delete pDataArray[i];
 462                 }
 463 #endif
 464                 return numSeqs;
 465         }
 466         catch(exception& e) {
 467                 m->errorOut(e, "SummaryQualCommand", "createProcessesCreateSummary");
 468                 exit(1);
 469         }
 470 }
 471 /**************************************************************************************************/
 472 int SummaryQualCommand::printQual(string sumFile, vector<int>& position, vector<int>& averageQ, vector< vector<int> >& scores) {
 473         try {
 474                 ofstream out;
 475                 m->openOutputFile(sumFile, out);
 476                 out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
 477                 outputNames.push_back(sumFile); outputTypes["summary"].push_back(sumFile);
 478
 479                 //print headings
 480                 out << "Position\tnumSeqs\tAverageQ\t";
 481                 for (int i = 0; i < 41; i++) { out << "q" << i << '\t'; }
 482                 out << endl;
 483
 484                 for (int i = 0; i < position.size(); i++) {
 485
 486                         if (m->control_pressed) { out.close(); return 0; }
 487
 488                         double average = averageQ[i] / (float) position[i];
 489                         out << i << '\t' << position[i] << '\t' << average << '\t';
 490
 491                         for (int j = 0; j < 41; j++) {
 492                                 out << scores[i][j] << '\t';
 493                         }
 494                         out << endl;
 495                 }
 496
 497                 out.close();
 498
 499                 return 0;
 500         }
 501         catch(exception& e) {
 502                 m->errorOut(e, "SummaryQualCommand", "printQual");
 503                 exit(1);
 504         }
 505 }
 506
 507 /**************************************************************************************/
 508
 509