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 getOutputFileNameTag(string, string);
  29         string getHelpString();
  30         string getCitation() { return "http://www.mothur.org/wiki/Summary.qual"; }
  31         string getDescription()         { return "summarize the quality of a set of sequences"; }
  32
  33         int execute();
  34         void help() { m->mothurOut(getHelpString()); }
  35
  36 private:
  37         bool abort;
  38         string qualfile, outputDir, namefile, countfile;
  39         vector<string> outputNames;
  40         map<string, int> nameMap;
  41         int processors;
  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< vector<int> >&, string);
  53         int driverCreateSummary(vector<int>&, vector<int>&, vector< vector<int> >&, string, linePair);
  54         int printQual(string, vector<int>&, vector<int>&, vector< vector<int> >&);
  55 };
  56
  57 /**************************************************************************************************/
  58 //custom data structure for threads to use.
  59 // This is passed by void pointer so it can be any data type
  60 // that can be passed using a single void pointer (LPVOID).
  61 struct seqSumQualData {
  62         vector<int> position;
  63         vector<int> averageQ;
  64         vector< vector<int> > scores;
  65         string filename;
  66         unsigned long long start;
  67         unsigned long long end;
  68         int count;
  69         MothurOut* m;
  70     bool hasNameMap;
  71         map<string, int> nameMap;
  72
  73         ~seqSumQualData(){}
  74         seqSumQualData(string f, MothurOut* mout, unsigned long long st, unsigned long long en, bool n, map<string, int> nam) {
  75                 filename = f;
  76                 m = mout;
  77                 start = st;
  78                 end = en;
  79                 hasNameMap = n;
  80                 nameMap = nam;
  81                 count = 0;
  82         }
  83 };
  84
  85 /**************************************************************************************************/
  86 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
  87 #else
  88 static DWORD WINAPI MySeqSumQualThreadFunction(LPVOID lpParam){
  89         seqSumQualData* pDataArray;
  90         pDataArray = (seqSumQualData*)lpParam;
  91
  92         try {
  93                 ifstream in;
  94                 pDataArray->m->openInputFile(pDataArray->filename, in);
  95
  96                 //print header if you are process 0
  97                 if ((pDataArray->start == 0) || (pDataArray->start == 1)) {
  98                         in.seekg(0);
  99                 }else { //this accounts for the difference in line endings.
 100                         in.seekg(pDataArray->start-1); pDataArray->m->gobble(in);
 101                 }
 102
 103                 int count = 0;
 104                 for(int i = 0; i < pDataArray->end; i++){ //end is the number of sequences to process
 105
 106                         if (pDataArray->m->control_pressed) { in.close(); pDataArray->count = 1; return 1; }
 107
 108                         QualityScores current(in); pDataArray->m->gobble(in);
 109
 110                         if (current.getName() != "") {
 111
 112                                 int num = 1;
 113                                 if (pDataArray->hasNameMap) {
 114                                         //make sure this sequence is in the namefile, else error
 115                                         map<string, int>::iterator it = pDataArray->nameMap.find(current.getName());
 116
 117                                         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; }
 118                                         else { num = it->second; }
 119                                 }
 120
 121                                 vector<int> thisScores = current.getQualityScores();
 122
 123                                 //resize to num of positions setting number of seqs with that size to 1
 124                                 if (pDataArray->position.size() < thisScores.size()) { pDataArray->position.resize(thisScores.size(), 0); }
 125                                 if (pDataArray->averageQ.size() < thisScores.size()) { pDataArray->averageQ.resize(thisScores.size(), 0); }
 126                                 if (pDataArray->scores.size() < thisScores.size()) {
 127                                         pDataArray->scores.resize(thisScores.size());
 128                                         for (int i = 0; i < pDataArray->scores.size(); i++) { pDataArray->scores.at(i).resize(41, 0); }
 129                                 }
 130
 131                                 //increase counts of number of seqs with this position
 132                                 //average is really the total, we will average in execute
 133                                 for (int i = 0; i < thisScores.size(); i++) {
 134                                         pDataArray->position.at(i) += num;
 135                                         pDataArray->averageQ.at(i) += (thisScores[i] * num); //weighting for namesfile
 136                                         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; }
 137                                         else { pDataArray->scores.at(i)[thisScores[i]] += num; }
 138                                 }
 139
 140                                 count += num;
 141                         }
 142                 }
 143
 144                 pDataArray->count = count;
 145                 in.close();
 146
 147                 return 0;
 148
 149         }
 150         catch(exception& e) {
 151                 pDataArray->m->errorOut(e, "SummaryQualCommand", "MySeqSumQualThreadFunction");
 152                 exit(1);
 153         }
 154 }
 155 #endif
 156
 157
 158 /**************************************************************************************************/
 159
 160
 161 #endif
 162