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