aligncommand.cpp

   1 /*
   2  *  aligncommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by Sarah Westcott on 5/15/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  *      This version of nast does everything I think that the greengenes nast server does and then some.  I have added the
   9  *      feature of allowing users to define their database, kmer size for searching, alignment penalty values and alignment
  10  *      method.  This latter feature is perhaps most significant.  nastPlus enables a user to use either a Needleman-Wunsch
  11  *      (non-affine gap penalty) or Gotoh (affine gap penalty) pairwise alignment algorithm.  This is significant because it
  12  *      allows for a global alignment and not the local alignment provided by bLAst.  Furthermore, it has the potential to
  13  *      provide a better alignment because of the banding method employed by blast (I'm not sure about this).
  14  *
  15  */
  16
  17 #include "aligncommand.h"
  18 #include "sequence.hpp"
  19
  20 #include "gotohoverlap.hpp"
  21 #include "needlemanoverlap.hpp"
  22 #include "blastalign.hpp"
  23 #include "noalign.hpp"
  24
  25 #include "kmerdb.hpp"
  26 #include "suffixdb.hpp"
  27 #include "blastdb.hpp"
  28
  29 #include "nast.hpp"
  30 #include "nastreport.hpp"
  31
  32
  33 //**********************************************************************************************************************
  34
  35 AlignCommand::AlignCommand(string option){
  36         try {
  37                 //              globaldata = GlobalData::getInstance();
  38                 abort = false;
  39
  40                 //allow user to run help
  41                 if(option == "help") { help(); abort = true; }
  42
  43                 else {
  44
  45                         //valid paramters for this command
  46                         string AlignArray[] =  {"template","candidate","search","ksize","align","match","mismatch","gapopen","gapextend", "processors"};
  47                         vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
  48
  49                         OptionParser parser(option);
  50                         map<string, string> parameters = parser.getParameters();
  51
  52                         ValidParameters validParameter;
  53
  54                         //check to make sure all parameters are valid for command
  55                         for (map<string, string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
  56                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  57                         }
  58
  59                         //check for required parameters
  60                         templateFileName = validParameter.validFile(parameters, "template", true);
  61                         if (templateFileName == "not found") {
  62                                 mothurOut("template is a required parameter for the align.seqs command.");
  63                                 mothurOutEndLine();
  64                                 abort = true;
  65                         }
  66                         else if (templateFileName == "not open") { abort = true; }
  67
  68                         candidateFileName = validParameter.validFile(parameters, "candidate", true);
  69                         if (candidateFileName == "not found") {
  70                                 mothurOut("candidate is a required parameter for the align.seqs command.");
  71                                 mothurOutEndLine();
  72                                 abort = true;
  73                         }
  74                         else if (candidateFileName == "not open") { abort = true; }
  75
  76                         //check for optional parameter and set defaults
  77                         // ...at some point should added some additional type checking...
  78                         string temp;
  79                         temp = validParameter.validFile(parameters, "ksize", false);            if (temp == "not found"){       temp = "8";                             }
  80                         convert(temp, kmerSize);
  81
  82                         temp = validParameter.validFile(parameters, "match", false);            if (temp == "not found"){       temp = "1.0";                   }
  83                         convert(temp, match);
  84
  85                         temp = validParameter.validFile(parameters, "mismatch", false);         if (temp == "not found"){       temp = "-1.0";                  }
  86                         convert(temp, misMatch);
  87
  88                         temp = validParameter.validFile(parameters, "gapopen", false);          if (temp == "not found"){       temp = "-2.0";                  }
  89                         convert(temp, gapOpen);
  90
  91                         temp = validParameter.validFile(parameters, "gapextend", false);        if (temp == "not found"){       temp = "-1.0";                  }
  92                         convert(temp, gapExtend);
  93
  94                         temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found"){       temp = "1";                             }
  95                         convert(temp, processors);
  96
  97                         search = validParameter.validFile(parameters, "search", false);         if (search == "not found"){     search = "kmer";                }
  98
  99                         align = validParameter.validFile(parameters, "align", false);           if (align == "not found"){      align = "needleman";    }
 100                 }
 101
 102         }
 103         catch(exception& e) {
 104                 errorOut(e, "AlignCommand", "AlignCommand");
 105                 exit(1);
 106         }
 107 }
 108
 109 //**********************************************************************************************************************
 110
 111 AlignCommand::~AlignCommand(){
 112
 113         if (abort == false) {
 114                 for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 115                 delete templateDB;
 116                 delete alignment;
 117         }
 118 }
 119
 120 //**********************************************************************************************************************
 121
 122 void AlignCommand::help(){
 123         try {
 124                 mothurOut("The align.seqs command reads a file containing sequences and creates an alignment file and a report file.\n");
 125                 mothurOut("The align.seqs command parameters are template, candidate, search, ksize, align, match, mismatch, gapopen and gapextend.\n");
 126                 mothurOut("The template and candidate parameters are required.\n");
 127                 mothurOut("The search parameter allows you to specify the method to find most similar template.  Your options are: suffix, kmer and blast. The default is kmer.\n");
 128                 mothurOut("The align parameter allows you to specify the alignment method to use.  Your options are: gotoh, needleman, blast and noalign. The default is needleman.\n");
 129                 mothurOut("The ksize parameter allows you to specify the kmer size for finding most similar template to candidate.  The default is 7.\n");
 130                 mothurOut("The match parameter allows you to specify the bonus for having the same base. The default is 1.0.\n");
 131                 mothurOut("The mistmatch parameter allows you to specify the penalty for having different bases.  The default is -1.0.\n");
 132                 mothurOut("The gapopen parameter allows you to specify the penalty for opening a gap in an alignment. The default is -1.0.\n");
 133                 mothurOut("The gapextend parameter allows you to specify the penalty for extending a gap in an alignment.  The default is -2.0.\n");
 134                 mothurOut("The align.seqs command should be in the following format: \n");
 135                 mothurOut("align.seqs(template=yourTemplateFile, candidate=yourCandidateFile, align=yourAlignmentMethod, search=yourSearchmethod, ksize=yourKmerSize, match=yourMatchBonus, mismatch=yourMismatchpenalty, gapopen=yourGapopenPenalty, gapextend=yourGapExtendPenalty) \n");
 136                 mothurOut("Example align.seqs(candidate=candidate.fasta, template=core.filtered, align=kmer, search=gotoh, ksize=8, match=2.0, mismatch=3.0, gapopen=-2.0, gapextend=-1.0)\n");
 137                 mothurOut("Note: No spaces between parameter labels (i.e. candidate), '=' and parameters (i.e.yourFastaFile).\n\n");
 138         }
 139         catch(exception& e) {
 140                 errorOut(e, "AlignCommand", "help");
 141                 exit(1);
 142         }
 143 }
 144
 145
 146 //**********************************************************************************************************************
 147
 148 int AlignCommand::execute(){
 149         try {
 150                 if (abort == true) {    return 0;       }
 151
 152                 if(search == "kmer")                    {       templateDB = new KmerDB(templateFileName, kmerSize);    }
 153                 else if(search == "suffix")             {       templateDB = new SuffixDB(templateFileName);                    }
 154                 else if(search == "blast")              {       templateDB = new BlastDB(templateFileName, gapOpen, gapExtend, match, misMatch);        }
 155                 else {
 156                         mothurOut(search + " is not a valid search option. I will run the command using kmer, ksize=8.");
 157                         mothurOutEndLine();
 158                         kmerSize = 8;
 159
 160                         templateDB = new KmerDB(templateFileName, kmerSize);
 161                 }
 162
 163                 int longestBase = templateDB->getLongestBase();
 164
 165                 if(align == "gotoh")                    {       alignment = new GotohOverlap(gapOpen, gapExtend, match, misMatch, longestBase);                 }
 166                 else if(align == "needleman")   {       alignment = new NeedlemanOverlap(gapOpen, match, misMatch, longestBase);                                }
 167                 else if(align == "blast")               {       alignment = new BlastAlignment(gapOpen, gapExtend, match, misMatch);            }
 168                 else if(align == "noalign")             {       alignment = new NoAlign();                                                                                                      }
 169                 else {
 170                         mothurOut(align + " is not a valid alignment option. I will run the command using needleman.");
 171                         mothurOutEndLine();
 172                         alignment = new NeedlemanOverlap(gapOpen, match, misMatch, longestBase);
 173                 }
 174                 mothurOut("Aligning sequences...");
 175                 mothurOutEndLine();
 176
 177                 string alignFileName = candidateFileName.substr(0,candidateFileName.find_last_of(".")+1) + "align";
 178                 string reportFileName = candidateFileName.substr(0,candidateFileName.find_last_of(".")+1) + "align.report";
 179
 180                 int numFastaSeqs = 0;
 181                 int start = time(NULL);
 182
 183 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 184                 if(processors == 1){
 185                         ifstream inFASTA;
 186                         openInputFile(candidateFileName, inFASTA);
 187                         numFastaSeqs=count(istreambuf_iterator<char>(inFASTA),istreambuf_iterator<char>(), '>');
 188                         inFASTA.close();
 189
 190                         lines.push_back(new linePair(0, numFastaSeqs));
 191
 192                         driver(lines[0], alignFileName, reportFileName);
 193
 194                 }
 195                 else{
 196                         vector<int> positions;
 197                         processIDS.resize(0);
 198
 199                         ifstream inFASTA;
 200                         openInputFile(candidateFileName, inFASTA);
 201
 202                         string input;
 203                         while(!inFASTA.eof()){
 204                                 input = getline(inFASTA);
 205                                 if (input.length() != 0) {
 206                                         if(input[0] == '>'){    int pos = inFASTA.tellg(); positions.push_back(pos - input.length() - 1);       }
 207                                 }
 208                         }
 209                         inFASTA.close();
 210
 211                         numFastaSeqs = positions.size();
 212
 213                         int numSeqsPerProcessor = numFastaSeqs / processors;
 214
 215                         for (int i = 0; i < processors; i++) {
 216                                 int startPos = positions[ i * numSeqsPerProcessor ];
 217                                 if(i == processors - 1){
 218                                         numSeqsPerProcessor = numFastaSeqs - i * numSeqsPerProcessor;
 219                                 }
 220                                 lines.push_back(new linePair(startPos, numSeqsPerProcessor));
 221                         }
 222                         createProcesses(alignFileName, reportFileName);
 223
 224                         rename((alignFileName + toString(processIDS[0]) + ".temp").c_str(), alignFileName.c_str());
 225                         rename((reportFileName + toString(processIDS[0]) + ".temp").c_str(), reportFileName.c_str());
 226
 227                         for(int i=1;i<processors;i++){
 228                                 appendAlignFiles((alignFileName + toString(processIDS[i]) + ".temp"), alignFileName);
 229                                 remove((alignFileName + toString(processIDS[i]) + ".temp").c_str());
 230
 231                                 appendReportFiles((reportFileName + toString(processIDS[i]) + ".temp"), reportFileName);
 232                                 remove((reportFileName + toString(processIDS[i]) + ".temp").c_str());
 233                         }
 234
 235                 }
 236 #else
 237                 ifstream inFASTA;
 238                 openInputFile(candidateFileName, inFASTA);
 239                 numFastaSeqs=count(istreambuf_iterator<char>(inFASTA),istreambuf_iterator<char>(), '>');
 240                 inFASTA.close();
 241
 242                 lines.push_back(new linePair(0, numFastaSeqs));
 243
 244                 driver(lines[0], alignFileName, reportFileName);
 245 #endif
 246
 247                 mothurOut("It took " + toString(time(NULL) - start) + " secs to align " + toString(numFastaSeqs) + " sequences.");
 248                 mothurOutEndLine();
 249                 mothurOutEndLine();
 250
 251                 return 0;
 252         }
 253         catch(exception& e) {
 254                 errorOut(e, "AlignCommand", "execute");
 255                 exit(1);
 256         }
 257 }
 258
 259 //**********************************************************************************************************************
 260
 261 int AlignCommand::driver(linePair* line, string alignFName, string reportFName){
 262         try {
 263                 ofstream alignmentFile;
 264                 openOutputFile(alignFName, alignmentFile);
 265                 NastReport report(reportFName);
 266
 267                 ifstream inFASTA;
 268                 openInputFile(candidateFileName, inFASTA);
 269
 270                 inFASTA.seekg(line->start);
 271
 272                 for(int i=0;i<line->numSeqs;i++){
 273
 274                         Sequence* candidateSeq = new Sequence(inFASTA);
 275                         report.setCandidate(candidateSeq);
 276
 277                         Sequence temp = templateDB->findClosestSequence(candidateSeq);
 278                         Sequence* templateSeq = &temp;
 279
 280                         report.setTemplate(templateSeq);
 281                         report.setSearchParameters(search, templateDB->getSearchScore());
 282
 283                         Nast nast(alignment, candidateSeq, templateSeq);
 284
 285                         report.setAlignmentParameters(align, alignment);
 286
 287                         report.setNastParameters(nast);
 288
 289                         alignmentFile << '>' << candidateSeq->getName() << '\n' << candidateSeq->getAligned() << endl;
 290
 291                         report.print();
 292
 293                         delete candidateSeq;
 294                 }
 295
 296                 alignmentFile.close();
 297                 inFASTA.close();
 298
 299                 return 1;
 300         }
 301         catch(exception& e) {
 302                 errorOut(e, "AlignCommand", "driver");
 303                 exit(1);
 304         }
 305 }
 306
 307 /**************************************************************************************************/
 308
 309 void AlignCommand::createProcesses(string alignFileName, string reportFileName) {
 310         try {
 311 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 312                 int process = 0;
 313                 //              processIDS.resize(0);
 314
 315                 //loop through and create all the processes you want
 316                 while (process != processors) {
 317                         int pid = fork();
 318
 319                         if (pid > 0) {
 320                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 321                                 process++;
 322                         }else if (pid == 0){
 323                                 driver(lines[process], alignFileName + toString(getpid()) + ".temp", reportFileName + toString(getpid()) + ".temp");
 324                                 exit(0);
 325                         }else { mothurOut("unable to spawn the necessary processes."); mothurOutEndLine(); exit(0); }
 326                 }
 327
 328                 //force parent to wait until all the processes are done
 329                 for (int i=0;i<processors;i++) {
 330                         int temp = processIDS[i];
 331                         wait(&temp);
 332                 }
 333 #endif
 334         }
 335         catch(exception& e) {
 336                 errorOut(e, "AlignCommand", "createProcesses");
 337                 exit(1);
 338         }
 339 }
 340
 341 /**************************************************************************************************/
 342
 343 void AlignCommand::appendAlignFiles(string temp, string filename) {
 344         try{
 345
 346                 ofstream output;
 347                 ifstream input;
 348                 openOutputFileAppend(filename, output);
 349                 openInputFile(temp, input);
 350
 351                 while(char c = input.get()){
 352                         if(input.eof())         {       break;                  }
 353                         else                            {       output << c;    }
 354                 }
 355
 356                 input.close();
 357                 output.close();
 358         }
 359         catch(exception& e) {
 360                 errorOut(e, "AlignCommand", "appendAlignFiles");
 361                 exit(1);
 362         }
 363 }
 364
 365 /**************************************************************************************************/
 366
 367 void AlignCommand::appendReportFiles(string temp, string filename) {
 368         try{
 369
 370                 ofstream output;
 371                 ifstream input;
 372                 openOutputFileAppend(filename, output);
 373                 openInputFile(temp, input);
 374
 375                 while (!input.eof())    {       char c = input.get(); if (c == 10 || c == 13){  break;  }       } // get header line
 376
 377                 while(char c = input.get()){
 378                         if(input.eof())         {       break;                  }
 379                         else                            {       output << c;    }
 380                 }
 381
 382                 input.close();
 383                 output.close();
 384         }
 385         catch(exception& e) {
 386                 errorOut(e, "AlignCommand", "appendReportFiles");
 387                 exit(1);
 388         }
 389 }
 390
 391 //**********************************************************************************************************************