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