seqsummarycommand.h

   1 #ifndef SEQSUMMARYCOMMAND_H
   2 #define SEQSUMMARYCOMMAND_H
   3
   4 /*
   5  *  seqcoordcommand.h
   6  *  Mothur
   7  *
   8  *  Created by Pat Schloss on 5/30/09.
   9  *  Copyright 2009 Patrick D. Schloss. All rights reserved.
  10  *
  11  */
  12
  13 #include "mothur.h"
  14 #include "command.hpp"
  15 #include "sequence.hpp"
  16
  17 /**************************************************************************************************/
  18
  19 class SeqSummaryCommand : public Command {
  20 public:
  21         SeqSummaryCommand(string);
  22         SeqSummaryCommand();
  23         ~SeqSummaryCommand(){}
  24
  25         vector<string> setParameters();
  26         string getCommandName()                 { return "summary.seqs";                        }
  27         string getCommandCategory()             { return "Sequence Processing";         }
  28         string getHelpString();
  29         string getCitation() { return "http://www.mothur.org/wiki/Summary.seqs"; }
  30         string getDescription()         { return "summarize the quality of sequences in an unaligned or aligned fasta file"; }
  31
  32         int execute();
  33         void help() { m->mothurOut(getHelpString()); }
  34 private:
  35         bool abort;
  36         string fastafile, outputDir, namefile;
  37         int processors;
  38         vector<string> outputNames;
  39         map<string, int> nameMap;
  40
  41         struct linePair {
  42                 unsigned long long start;
  43                 unsigned long long end;
  44                 linePair(unsigned long long i, unsigned long long j) : start(i), end(j) {}
  45         };
  46
  47         vector<linePair*> lines;
  48         vector<int> processIDS;
  49
  50         int createProcessesCreateSummary(vector<int>&, vector<int>&, vector<int>&, vector<int>&, vector<int>&, string, string);
  51         int driverCreateSummary(vector<int>&, vector<int>&, vector<int>&, vector<int>&, vector<int>&, string, string, linePair*);
  52
  53         #ifdef USE_MPI
  54         int MPICreateSummary(int, int, vector<int>&, vector<int>&, vector<int>&, vector<int>&, vector<int>&, MPI_File&, MPI_File&, vector<unsigned long long>&);
  55         #endif
  56
  57
  58 };
  59
  60 /**************************************************************************************************/
  61 //custom data structure for threads to use.
  62 // This is passed by void pointer so it can be any data type
  63 // that can be passed using a single void pointer (LPVOID).
  64 struct seqSumData {
  65         vector<int>* startPosition;
  66         vector<int>* endPosition;
  67         vector<int>* seqLength;
  68         vector<int>* ambigBases;
  69         vector<int>* longHomoPolymer;
  70         string filename;
  71         string sumFile;
  72         unsigned long long start;
  73         unsigned long long end;
  74         int count;
  75         MothurOut* m;
  76         string namefile;
  77         map<string, int> nameMap;
  78
  79
  80         seqSumData(){}
  81         seqSumData(vector<int>* s, vector<int>* e, vector<int>* l, vector<int>* a, vector<int>* h, string f, string sf, MothurOut* mout, unsigned long long st, unsigned long long en, string na, map<string, int> nam) {
  82                 startPosition = s;
  83                 endPosition = e;
  84                 seqLength = l;
  85                 ambigBases = a;
  86                 longHomoPolymer = h;
  87                 filename = f;
  88                 sumFile = sf;
  89                 m = mout;
  90                 start = st;
  91                 end = en;
  92                 namefile = na;
  93                 nameMap = nam;
  94                 count = 0;
  95         }
  96 };
  97
  98 /**************************************************************************************************/
  99 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 100 #else
 101 static DWORD WINAPI MySeqSumThreadFunction(LPVOID lpParam){
 102         seqSumData* pDataArray;
 103         pDataArray = (seqSumData*)lpParam;
 104
 105         try {
 106                 ofstream outSummary;
 107                 pDataArray->m->openOutputFile(pDataArray->sumFile, outSummary);
 108
 109                 ifstream in;
 110                 pDataArray->m->openInputFile(pDataArray->filename, in);
 111
 112                 //print header if you are process 0
 113                 if ((pDataArray->start == 0) || (pDataArray->start == 1)) {
 114                         outSummary << "seqname\tstart\tend\tnbases\tambigs\tpolymer\tnumSeqs" << endl;
 115                         in.seekg(0);
 116                 }else { //this accounts for the difference in line endings.
 117                         in.seekg(pDataArray->start-1); pDataArray->m->gobble(in);
 118                 }
 119
 120                 pDataArray->count = pDataArray->end;
 121                 for(int i = 0; i < pDataArray->end; i++){ //end is the number of sequences to process
 122
 123                         if (pDataArray->m->control_pressed) { in.close(); outSummary.close(); pDataArray->count = 1; return 1; }
 124
 125                         Sequence current(in); pDataArray->m->gobble(in);
 126
 127                         if (current.getName() != "") {
 128
 129                                 int num = 1;
 130                                 if (pDataArray->namefile != "") {
 131                                         //make sure this sequence is in the namefile, else error
 132                                         map<string, int>::iterator it = pDataArray->nameMap.find(current.getName());
 133
 134                                         if (it == pDataArray->nameMap.end()) { pDataArray->m->mothurOut("[ERROR]: " + current.getName() + " is not in your namefile, please correct."); pDataArray->m->mothurOutEndLine(); pDataArray->m->control_pressed = true; }
 135                                         else { num = it->second; }
 136                                 }
 137
 138                                 //for each sequence this sequence represents
 139                                 for (int i = 0; i < num; i++) {
 140                                         pDataArray->startPosition->push_back(current.getStartPos());
 141                                         pDataArray->endPosition->push_back(current.getEndPos());
 142                                         pDataArray->seqLength->push_back(current.getNumBases());
 143                                         pDataArray->ambigBases->push_back(current.getAmbigBases());
 144                                         pDataArray->longHomoPolymer->push_back(current.getLongHomoPolymer());
 145                                 }
 146
 147                                 outSummary << current.getName() << '\t';
 148                                 outSummary << current.getStartPos() << '\t' << current.getEndPos() << '\t';
 149                                 outSummary << current.getNumBases() << '\t' << current.getAmbigBases() << '\t';
 150                                 outSummary << current.getLongHomoPolymer() << '\t' << num << endl;
 151                         }
 152                 }
 153
 154                 in.close();
 155                 outSummary.close();
 156
 157                 return 0;
 158
 159         }
 160         catch(exception& e) {
 161                 pDataArray->m->errorOut(e, "SeqSummaryCommand", "MySeqSumThreadFunction");
 162                 exit(1);
 163         }
 164 }
 165 #endif
 166
 167
 168
 169
 170 #endif
 171
 172 /**************************************************************************************************/
 173
 174