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