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