aligncommand.h

   1 #ifndef ALIGNCOMMAND_H
   2 #define ALIGNCOMMAND_H
   3 /*
   4  *  aligncommand.h
   5  *  Mothur
   6  *
   7  *  Created by Sarah Westcott on 5/15/09.
   8  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   9  *
  10  */
  11
  12 #include "command.hpp"
  13 #include "database.hpp"
  14 #include "alignment.hpp"
  15 #include "alignmentdb.h"
  16 #include "sequence.hpp"
  17
  18 #include "gotohoverlap.hpp"
  19 #include "needlemanoverlap.hpp"
  20 #include "blastalign.hpp"
  21 #include "noalign.hpp"
  22
  23 #include "nast.hpp"
  24 #include "nastreport.hpp"
  25
  26 //test
  27 class AlignCommand : public Command {
  28
  29 public:
  30         AlignCommand(string);
  31         AlignCommand();
  32         ~AlignCommand();
  33
  34         vector<string> setParameters();
  35         string getCommandName()                 { return "align.seqs";                  }
  36         string getCommandCategory()             { return "Sequence Processing"; }
  37
  38         string getHelpString();
  39     string getOutputPattern(string);
  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 appendReportFiles(string, string);
  61
  62         #ifdef USE_MPI
  63         int driverMPI(int, int, MPI_File&, MPI_File&, MPI_File&, MPI_File&, vector<unsigned long long>&);
  64         #endif
  65
  66         string candidateFileName, templateFileName, distanceFileName, search, align, outputDir;
  67         float match, misMatch, gapOpen, gapExtend, threshold;
  68         int processors, kmerSize;
  69         vector<string> candidateFileNames;
  70         vector<string> outputNames;
  71
  72         bool abort, flip, calledHelp, save;
  73
  74 };
  75
  76 /**************************************************************************************************/
  77 //custom data structure for threads to use.
  78 // This is passed by void pointer so it can be any data type
  79 // that can be passed using a single void pointer (LPVOID).
  80 struct alignData {
  81         string templateFileName;
  82         string alignFName;
  83         string reportFName;
  84         string accnosFName;
  85         string filename;
  86         string align;
  87         string search;
  88         bool flip;
  89         unsigned long long start;
  90         unsigned long long end;
  91         MothurOut* m;
  92         //AlignmentDB* templateDB;
  93         float match, misMatch, gapOpen, gapExtend, threshold;
  94         int count, kmerSize, threadID;
  95
  96         alignData(){}
  97         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) {
  98                 templateFileName = te;
  99                 alignFName = a;
 100                 reportFName = r;
 101                 accnosFName = ac;
 102                 filename = f;
 103                 flip = fl;
 104                 m = mout;
 105                 start = st;
 106                 end = en;
 107                 //templateDB = tdb;
 108                 match = ma;
 109                 misMatch = misMa;
 110                 gapOpen = gapO;
 111                 gapExtend = gapE;
 112                 threshold = thr;
 113                 align = al;
 114                 search = se;
 115                 count = 0;
 116                 kmerSize = ks;
 117                 threadID = tid;
 118         }
 119 };
 120
 121 /**************************************************************************************************/
 122 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 123 #else
 124 static DWORD WINAPI MyAlignThreadFunction(LPVOID lpParam){
 125         alignData* pDataArray;
 126         pDataArray = (alignData*)lpParam;
 127
 128         try {
 129                 ofstream alignmentFile;
 130                 pDataArray->m->openOutputFile(pDataArray->alignFName, alignmentFile);
 131
 132                 ofstream accnosFile;
 133                 pDataArray->m->openOutputFile(pDataArray->accnosFName, accnosFile);
 134
 135                 NastReport report(pDataArray->reportFName);
 136
 137                 ifstream inFASTA;
 138                 pDataArray->m->openInputFile(pDataArray->filename, inFASTA);
 139
 140                 //print header if you are process 0
 141                 if ((pDataArray->start == 0) || (pDataArray->start == 1)) {
 142                         inFASTA.seekg(0);
 143                 }else { //this accounts for the difference in line endings.
 144                         inFASTA.seekg(pDataArray->start-1); pDataArray->m->gobble(inFASTA);
 145                 }
 146
 147                 AlignmentDB* templateDB = new AlignmentDB(pDataArray->templateFileName, pDataArray->search, pDataArray->kmerSize, pDataArray->gapOpen, pDataArray->gapExtend, pDataArray->match, pDataArray->misMatch, pDataArray->threadID);
 148
 149                 //moved this into driver to avoid deep copies in windows paralellized version
 150                 Alignment* alignment;
 151                 int longestBase = templateDB->getLongestBase();
 152                 if(pDataArray->align == "gotoh")                        {       alignment = new GotohOverlap(pDataArray->gapOpen, pDataArray->gapExtend, pDataArray->match, pDataArray->misMatch, longestBase);                 }
 153                 else if(pDataArray->align == "needleman")       {       alignment = new NeedlemanOverlap(pDataArray->gapOpen, pDataArray->match, pDataArray->misMatch, longestBase);                            }
 154                 else if(pDataArray->align == "blast")           {       alignment = new BlastAlignment(pDataArray->gapOpen, pDataArray->gapExtend, pDataArray->match, pDataArray->misMatch);            }
 155                 else if(pDataArray->align == "noalign")         {       alignment = new NoAlign();                                                                                                      }
 156                 else {
 157                         pDataArray->m->mothurOut(pDataArray->align + " is not a valid alignment option. I will run the command using needleman.");
 158                         pDataArray->m->mothurOutEndLine();
 159                         alignment = new NeedlemanOverlap(pDataArray->gapOpen, pDataArray->match, pDataArray->misMatch, longestBase);
 160                 }
 161
 162                 pDataArray->count = 0;
 163                 for(int i = 0; i < pDataArray->end; i++){ //end is the number of sequences to process
 164
 165                         if (pDataArray->m->control_pressed) {  break; }
 166
 167                         Sequence* candidateSeq = new Sequence(inFASTA);  pDataArray->m->gobble(inFASTA);
 168                         report.setCandidate(candidateSeq);
 169
 170                         int origNumBases = candidateSeq->getNumBases();
 171                         string originalUnaligned = candidateSeq->getUnaligned();
 172                         int numBasesNeeded = origNumBases * pDataArray->threshold;
 173
 174                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 175                                 if (candidateSeq->getUnaligned().length() > alignment->getnRows()) {
 176                                         alignment->resize(candidateSeq->getUnaligned().length()+1);
 177                                 }
 178
 179                                 Sequence temp = templateDB->findClosestSequence(candidateSeq);
 180                                 Sequence* templateSeq = &temp;
 181
 182                                 float searchScore = templateDB->getSearchScore();
 183
 184                                 Nast* nast = new Nast(alignment, candidateSeq, templateSeq);
 185
 186                                 Sequence* copy;
 187
 188                                 Nast* nast2;
 189                                 bool needToDeleteCopy = false;  //this is needed in case you have you enter the ifs below
 190                                 //since nast does not make a copy of hte sequence passed, and it is used by the reporter below
 191                                 //you can't delete the copy sequence til after you report, but you may choose not to create it in the first place
 192                                 //so this bool tells you if you need to delete it
 193
 194                                 //if there is a possibility that this sequence should be reversed
 195                                 if (candidateSeq->getNumBases() < numBasesNeeded) {
 196
 197                                         string wasBetter =  "";
 198                                         //if the user wants you to try the reverse
 199                                         if (pDataArray->flip) {
 200
 201                                                 //get reverse compliment
 202                                                 copy = new Sequence(candidateSeq->getName(), originalUnaligned);
 203                                                 copy->reverseComplement();
 204
 205                                                 //rerun alignment
 206                                                 Sequence temp2 = templateDB->findClosestSequence(copy);
 207                                                 Sequence* templateSeq2 = &temp2;
 208
 209                                                 searchScore = templateDB->getSearchScore();
 210
 211                                                 nast2 = new Nast(alignment, copy, templateSeq2);
 212
 213                                                 //check if any better
 214                                                 if (copy->getNumBases() > candidateSeq->getNumBases()) {
 215                                                         candidateSeq->setAligned(copy->getAligned());  //use reverse compliments alignment since its better
 216                                                         templateSeq = templateSeq2;
 217                                                         delete nast;
 218                                                         nast = nast2;
 219                                                         needToDeleteCopy = true;
 220                                                         wasBetter = "\treverse complement produced a better alignment, so mothur used the reverse complement.";
 221                                                 }else{
 222                                                         wasBetter = "\treverse complement did NOT produce a better alignment so it was not used, please check sequence.";
 223                                                         delete nast2;
 224                                                         delete copy;
 225                                                 }
 226                                         }
 227
 228                                         //create accnos file with names
 229                                         accnosFile << candidateSeq->getName() << wasBetter << endl;
 230                                 }
 231
 232                                 report.setTemplate(templateSeq);
 233                                 report.setSearchParameters(pDataArray->search, searchScore);
 234                                 report.setAlignmentParameters(pDataArray->align, alignment);
 235                                 report.setNastParameters(*nast);
 236
 237                                 alignmentFile << '>' << candidateSeq->getName() << '\n' << candidateSeq->getAligned() << endl;
 238
 239                                 report.print();
 240                                 delete nast;
 241                                 if (needToDeleteCopy) {   delete copy;   }
 242
 243                                 pDataArray->count++;
 244                         }
 245                         delete candidateSeq;
 246
 247                         //report progress
 248                         if((pDataArray->count) % 100 == 0){     pDataArray->m->mothurOutJustToScreen(toString(pDataArray->count)+"\n");         }
 249
 250                 }
 251                 //report progress
 252                 if((pDataArray->count) % 100 != 0){     pDataArray->m->mothurOutJustToScreen(toString(pDataArray->count)+"\n");         }
 253
 254                 delete alignment;
 255                 delete templateDB;
 256                 alignmentFile.close();
 257                 inFASTA.close();
 258                 accnosFile.close();
 259         }
 260         catch(exception& e) {
 261                 pDataArray->m->errorOut(e, "AlignCommand", "MyAlignThreadFunction");
 262                 exit(1);
 263         }
 264 }
 265 #endif
 266
 267
 268
 269 #endif