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