aligncommand.h

   1 #ifndef ALIGNCOMMAND_H
   2 #define ALIGNCOMMAND_H
   3
   4 /*
   5  *  aligncommand.h
   6  *  Mothur
   7  *
   8  *  Created by Sarah Westcott on 5/15/09.
   9  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
  10  *
  11  */
  12
  13 #include "mothur.h"
  14 #include "command.hpp"
  15 #include "database.hpp"
  16 #include "alignment.hpp"
  17 #include "alignmentdb.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 class AlignCommand : public Command {
  30
  31 public:
  32         AlignCommand(string);
  33         AlignCommand();
  34         ~AlignCommand();
  35
  36         vector<string> setParameters();
  37         string getCommandName()                 { return "align.seqs";                  }
  38         string getCommandCategory()             { return "Sequence Processing"; }
  39         string getHelpString();
  40         string getCitation() { return "DeSantis TZ, Jr., Hugenholtz P, Keller K, Brodie EL, Larsen N, Piceno YM, Phan R, Andersen GL (2006). NAST: a multiple sequence alignment server for comparative analysis of 16S rRNA genes. Nucleic Acids Res 34: W394-9.\nSchloss PD (2009). A high-throughput DNA sequence aligner for microbial ecology studies. PLoS ONE 4: e8230.\nSchloss PD (2010). The effects of alignment quality, distance calculation method, sequence filtering, and region on the analysis of 16S rRNA gene-based studies. PLoS Comput Biol 6: e1000844.\nhttp://www.mothur.org/wiki/Align.seqs http://www.mothur.org/wiki/Align.seqs"; }
  41         string getDescription()         { return "align sequences"; }
  42
  43         int execute();
  44         void help() { m->mothurOut(getHelpString()); }
  45
  46 private:
  47         struct linePair {
  48                 unsigned long long start;
  49                 unsigned long long end;
  50                 linePair(unsigned long long i, unsigned long long j) : start(i), end(j) {}
  51         };
  52         vector<int> processIDS;   //processid
  53         vector<linePair*> lines;
  54         bool MPIWroteAccnos;
  55
  56         AlignmentDB* templateDB;
  57
  58         int driver(linePair*, string, string, string, string);
  59         int createProcesses(string, string, string, string);
  60         void appendAlignFiles(string, string);
  61         void appendReportFiles(string, string);
  62
  63         #ifdef USE_MPI
  64         int driverMPI(int, int, MPI_File&, MPI_File&, MPI_File&, MPI_File&, vector<unsigned long long>&);
  65         #endif
  66
  67         string candidateFileName, templateFileName, distanceFileName, search, align, outputDir;
  68         float match, misMatch, gapOpen, gapExtend, threshold;
  69         int processors, kmerSize;
  70         vector<string> candidateFileNames;
  71         vector<string> outputNames;
  72
  73         bool abort, flip, calledHelp, save;
  74
  75 };
  76
  77 /**************************************************************************************************/
  78 //custom data structure for threads to use.
  79 // This is passed by void pointer so it can be any data type
  80 // that can be passed using a single void pointer (LPVOID).
  81 struct alignData {
  82         string templateFileName;
  83         string alignFName;
  84         string reportFName;
  85         string accnosFName;
  86         string filename;
  87         string align;
  88         string search;
  89         bool flip;
  90         unsigned long long start;
  91         unsigned long long end;
  92         MothurOut* m;
  93         //AlignmentDB* templateDB;
  94         float match, misMatch, gapOpen, gapExtend, threshold;
  95         int count, kmerSize, threadID;
  96
  97         alignData(){}
  98         alignData(string te, string a, string r, string ac, string f, string al, string se, int ks, MothurOut* mout, unsigned long long st, unsigned long long en, bool fl, float ma, float misMa, float gapO, float gapE, float thr, int tid) {
  99                 templateFileName = te;
 100                 alignFName = a;
 101                 reportFName = r;
 102                 accnosFName = ac;
 103                 filename = f;
 104                 flip = fl;
 105                 m = mout;
 106                 start = st;
 107                 end = en;
 108                 //templateDB = tdb;
 109                 match = ma;
 110                 misMatch = misMa;
 111                 gapOpen = gapO;
 112                 gapExtend = gapE;
 113                 threshold = thr;
 114                 align = al;
 115                 search = se;
 116                 count = 0;
 117                 kmerSize = ks;
 118                 threadID = tid;
 119         }
 120 };
 121
 122 /**************************************************************************************************/
 123 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 124 #else
 125 static DWORD WINAPI MyAlignThreadFunction(LPVOID lpParam){
 126         alignData* pDataArray;
 127         pDataArray = (alignData*)lpParam;
 128
 129         try {
 130                 ofstream alignmentFile;
 131                 pDataArray->m->openOutputFile(pDataArray->alignFName, alignmentFile);
 132
 133                 ofstream accnosFile;
 134                 pDataArray->m->openOutputFile(pDataArray->accnosFName, accnosFile);
 135
 136                 NastReport report(pDataArray->reportFName);
 137
 138                 ifstream inFASTA;
 139                 pDataArray->m->openInputFile(pDataArray->filename, inFASTA);
 140
 141                 //print header if you are process 0
 142                 if ((pDataArray->start == 0) || (pDataArray->start == 1)) {
 143                         inFASTA.seekg(0);
 144                 }else { //this accounts for the difference in line endings.
 145                         inFASTA.seekg(pDataArray->start-1); pDataArray->m->gobble(inFASTA);
 146                 }
 147
 148                 pDataArray->count = pDataArray->end;
 149
 150                 AlignmentDB* templateDB = new AlignmentDB(pDataArray->templateFileName, pDataArray->search, pDataArray->kmerSize, pDataArray->gapOpen, pDataArray->gapExtend, pDataArray->match, pDataArray->misMatch, pDataArray->threadID);
 151
 152                 //moved this into driver to avoid deep copies in windows paralellized version
 153                 Alignment* alignment;
 154                 int longestBase = templateDB->getLongestBase();
 155                 if(pDataArray->align == "gotoh")                        {       alignment = new GotohOverlap(pDataArray->gapOpen, pDataArray->gapExtend, pDataArray->match, pDataArray->misMatch, longestBase);                 }
 156                 else if(pDataArray->align == "needleman")       {       alignment = new NeedlemanOverlap(pDataArray->gapOpen, pDataArray->match, pDataArray->misMatch, longestBase);                            }
 157                 else if(pDataArray->align == "blast")           {       alignment = new BlastAlignment(pDataArray->gapOpen, pDataArray->gapExtend, pDataArray->match, pDataArray->misMatch);            }
 158                 else if(pDataArray->align == "noalign")         {       alignment = new NoAlign();                                                                                                      }
 159                 else {
 160                         pDataArray->m->mothurOut(pDataArray->align + " is not a valid alignment option. I will run the command using needleman.");
 161                         pDataArray->m->mothurOutEndLine();
 162                         alignment = new NeedlemanOverlap(pDataArray->gapOpen, pDataArray->match, pDataArray->misMatch, longestBase);
 163                 }
 164
 165                 int count = 0;
 166                 for(int i = 0; i < pDataArray->end; i++){ //end is the number of sequences to process
 167
 168                         if (pDataArray->m->control_pressed) {  break; }
 169
 170                         Sequence* candidateSeq = new Sequence(inFASTA);  pDataArray->m->gobble(inFASTA);
 171                         report.setCandidate(candidateSeq);
 172
 173                         int origNumBases = candidateSeq->getNumBases();
 174                         string originalUnaligned = candidateSeq->getUnaligned();
 175                         int numBasesNeeded = origNumBases * pDataArray->threshold;
 176
 177                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 178                                 if (candidateSeq->getUnaligned().length() > alignment->getnRows()) {
 179                                         alignment->resize(candidateSeq->getUnaligned().length()+1);
 180                                 }
 181
 182                                 Sequence temp = templateDB->findClosestSequence(candidateSeq);
 183                                 Sequence* templateSeq = &temp;
 184
 185                                 float searchScore = templateDB->getSearchScore();
 186
 187                                 Nast* nast = new Nast(alignment, candidateSeq, templateSeq);
 188
 189                                 Sequence* copy;
 190
 191                                 Nast* nast2;
 192                                 bool needToDeleteCopy = false;  //this is needed in case you have you enter the ifs below
 193                                 //since nast does not make a copy of hte sequence passed, and it is used by the reporter below
 194                                 //you can't delete the copy sequence til after you report, but you may choose not to create it in the first place
 195                                 //so this bool tells you if you need to delete it
 196
 197                                 //if there is a possibility that this sequence should be reversed
 198                                 if (candidateSeq->getNumBases() < numBasesNeeded) {
 199
 200                                         string wasBetter =  "";
 201                                         //if the user wants you to try the reverse
 202                                         if (pDataArray->flip) {
 203
 204                                                 //get reverse compliment
 205                                                 copy = new Sequence(candidateSeq->getName(), originalUnaligned);
 206                                                 copy->reverseComplement();
 207
 208                                                 //rerun alignment
 209                                                 Sequence temp2 = templateDB->findClosestSequence(copy);
 210                                                 Sequence* templateSeq2 = &temp2;
 211
 212                                                 searchScore = templateDB->getSearchScore();
 213
 214                                                 nast2 = new Nast(alignment, copy, templateSeq2);
 215
 216                                                 //check if any better
 217                                                 if (copy->getNumBases() > candidateSeq->getNumBases()) {
 218                                                         candidateSeq->setAligned(copy->getAligned());  //use reverse compliments alignment since its better
 219                                                         templateSeq = templateSeq2;
 220                                                         delete nast;
 221                                                         nast = nast2;
 222                                                         needToDeleteCopy = true;
 223                                                         wasBetter = "\treverse complement produced a better alignment, so mothur used the reverse complement.";
 224                                                 }else{
 225                                                         wasBetter = "\treverse complement did NOT produce a better alignment so it was not used, please check sequence.";
 226                                                         delete nast2;
 227                                                         delete copy;
 228                                                 }
 229                                         }
 230
 231                                         //create accnos file with names
 232                                         accnosFile << candidateSeq->getName() << wasBetter << endl;
 233                                 }
 234
 235                                 report.setTemplate(templateSeq);
 236                                 report.setSearchParameters(pDataArray->search, searchScore);
 237                                 report.setAlignmentParameters(pDataArray->align, alignment);
 238                                 report.setNastParameters(*nast);
 239
 240                                 alignmentFile << '>' << candidateSeq->getName() << '\n' << candidateSeq->getAligned() << endl;
 241
 242                                 report.print();
 243                                 delete nast;
 244                                 if (needToDeleteCopy) {   delete copy;   }
 245
 246                                 count++;
 247                         }
 248                         delete candidateSeq;
 249
 250                         //report progress
 251                         if((count) % 100 == 0){ pDataArray->m->mothurOut(toString(count)); pDataArray->m->mothurOutEndLine();           }
 252
 253                 }
 254                 //report progress
 255                 if((count) % 100 != 0){ pDataArray->m->mothurOut(toString(count)); pDataArray->m->mothurOutEndLine();           }
 256
 257                 delete alignment;
 258                 delete templateDB;
 259                 alignmentFile.close();
 260                 inFASTA.close();
 261                 accnosFile.close();
 262         }
 263         catch(exception& e) {
 264                 pDataArray->m->errorOut(e, "AlignCommand", "MyAlignThreadFunction");
 265                 exit(1);
 266         }
 267 }
 268 #endif
 269
 270
 271
 272 #endif