summaryqualcommand.h

   1 #ifndef SUMMARYQUALCOMMAND_H
   2 #define SUMMARYQUALCOMMAND_H
   3
   4 /*
   5  *  summaryqualcommand.h
   6  *  Mothur
   7  *
   8  *  Created by westcott on 11/28/11.
   9  *  Copyright 2011 Schloss Lab. All rights reserved.
  10  *
  11  */
  12
  13
  14 #include "command.hpp"
  15 #include "qualityscores.h"
  16
  17 /**************************************************************************************************/
  18
  19 class SummaryQualCommand : public Command {
  20 public:
  21         SummaryQualCommand(string);
  22         SummaryQualCommand();
  23         ~SummaryQualCommand(){}
  24
  25         vector<string> setParameters();
  26         string getCommandName()                 { return "summary.qual";                        }
  27         string getCommandCategory()             { return "Sequence Processing";         }
  28         string getHelpString();
  29         string getCitation() { return "http://www.mothur.org/wiki/Summary.qual"; }
  30         string getDescription()         { return "summarize the quality of a set of sequences"; }
  31
  32         int execute();
  33         void help() { m->mothurOut(getHelpString()); }
  34
  35 private:
  36         bool abort;
  37         string qualfile, outputDir, namefile;
  38         vector<string> outputNames;
  39         map<string, int> nameMap;
  40         int processors;
  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< vector<int> >&, string);
  52         int driverCreateSummary(vector<int>&, vector<int>&, vector< vector<int> >&, string, linePair);
  53         int printQual(string, vector<int>&, vector<int>&, vector< vector<int> >&);
  54 };
  55
  56 /**************************************************************************************************/
  57 //custom data structure for threads to use.
  58 // This is passed by void pointer so it can be any data type
  59 // that can be passed using a single void pointer (LPVOID).
  60 struct seqSumQualData {
  61         vector<int> position;
  62         vector<int> averageQ;
  63         vector< vector<int> > scores;
  64         string filename, namefile;
  65         unsigned long long start;
  66         unsigned long long end;
  67         int count;
  68         MothurOut* m;
  69         map<string, int> nameMap;
  70
  71         ~seqSumQualData(){}
  72         seqSumQualData(string f, MothurOut* mout, unsigned long long st, unsigned long long en, string n, map<string, int> nam) {
  73                 filename = f;
  74                 m = mout;
  75                 start = st;
  76                 end = en;
  77                 namefile = n;
  78                 nameMap = nam;
  79                 count = 0;
  80         }
  81 };
  82
  83 /**************************************************************************************************/
  84 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
  85 #else
  86 static DWORD WINAPI MySeqSumQualThreadFunction(LPVOID lpParam){
  87         seqSumQualData* pDataArray;
  88         pDataArray = (seqSumQualData*)lpParam;
  89
  90         try {
  91                 ifstream in;
  92                 pDataArray->m->openInputFile(pDataArray->filename, in);
  93
  94                 //print header if you are process 0
  95                 if ((pDataArray->start == 0) || (pDataArray->start == 1)) {
  96                         in.seekg(0);
  97                 }else { //this accounts for the difference in line endings.
  98                         in.seekg(pDataArray->start-1); pDataArray->m->gobble(in);
  99                 }
 100
 101                 int count = 0;
 102                 for(int i = 0; i < pDataArray->end; i++){ //end is the number of sequences to process
 103
 104                         if (pDataArray->m->control_pressed) { in.close(); pDataArray->count = 1; return 1; }
 105
 106                         QualityScores current(in); pDataArray->m->gobble(in);
 107
 108                         if (current.getName() != "") {
 109
 110                                 int num = 1;
 111                                 if (pDataArray->namefile != "") {
 112                                         //make sure this sequence is in the namefile, else error
 113                                         map<string, int>::iterator it = pDataArray->nameMap.find(current.getName());
 114
 115                                         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; }
 116                                         else { num = it->second; }
 117                                 }
 118
 119                                 vector<int> thisScores = current.getQualityScores();
 120
 121                                 //resize to num of positions setting number of seqs with that size to 1
 122                                 if (pDataArray->position.size() < thisScores.size()) { pDataArray->position.resize(thisScores.size(), 0); }
 123                                 if (pDataArray->averageQ.size() < thisScores.size()) { pDataArray->averageQ.resize(thisScores.size(), 0); }
 124                                 if (pDataArray->scores.size() < thisScores.size()) {
 125                                         pDataArray->scores.resize(thisScores.size());
 126                                         for (int i = 0; i < pDataArray->scores.size(); i++) { pDataArray->scores.at(i).resize(41, 0); }
 127                                 }
 128
 129                                 //increase counts of number of seqs with this position
 130                                 //average is really the total, we will average in execute
 131                                 for (int i = 0; i < thisScores.size(); i++) {
 132                                         pDataArray->position.at(i) += num;
 133                                         pDataArray->averageQ.at(i) += (thisScores[i] * num); //weighting for namesfile
 134                                         if (thisScores[i] > 40) { pDataArray->m->mothurOut("[ERROR]: " + current.getName() + " has a quality scores of " + toString(thisScores[i]) + ", expecting values to be less than 40."); pDataArray->m->mothurOutEndLine(); pDataArray->m->control_pressed = true; }
 135                                         else { pDataArray->scores.at(i)[thisScores[i]] += num; }
 136                                 }
 137
 138                                 count += num;
 139                         }
 140                 }
 141
 142                 pDataArray->count = count;
 143                 in.close();
 144
 145                 return 0;
 146
 147         }
 148         catch(exception& e) {
 149                 pDataArray->m->errorOut(e, "SummaryQualCommand", "MySeqSumQualThreadFunction");
 150                 exit(1);
 151         }
 152 }
 153 #endif
 154
 155
 156 /**************************************************************************************************/
 157
 158
 159 #endif
 160