seqerrorcommand.h

   1 #ifndef SEQERRORCOMMAND
   2 #define SEQERRORCOMMAND
   3
   4 /*
   5  *  seqerrorcommand.h
   6  *  Mothur
   7  *
   8  *  Created by Pat Schloss on 7/15/10.
   9  *  Copyright 2010 Schloss Lab. All rights reserved.
  10  *
  11  */
  12
  13 #include "mothur.h"
  14 #include "command.hpp"
  15 #include "sequence.hpp"
  16
  17 struct Compare {
  18         int AA, AT, AG, AC,     TA, TT, TG, TC, GA, GT, GG, GC, CA, CT, CG, CC, NA, NT, NG, NC, Ai, Ti, Gi, Ci, Ni, dA, dT, dG, dC;
  19         string refName, queryName, sequence;
  20         double errorRate;
  21         int weight, matches, mismatches, total;
  22
  23         Compare(){
  24                 AA=0; AT=0; AG=0; AC=0;
  25                 TA=0; TT=0; TG=0; TC=0;
  26                 GA=0; GT=0; GG=0; GC=0;
  27                 CA=0; CT=0; CG=0; CC=0;
  28                 NA=0; NT=0; NG=0; NC=0;
  29                 Ai=0; Ti=0; Gi=0; Ci=0; Ni=0;
  30                 dA=0; dT=0; dG=0; dC=0;
  31                 refName = "";
  32                 queryName = "";
  33                 weight = 1;
  34                 matches = 0;
  35                 mismatches = 0;
  36                 total = 0;
  37                 errorRate = 1.0000;
  38                 sequence = "";
  39         }
  40 };
  41
  42 class SeqErrorCommand : public Command {
  43 public:
  44         SeqErrorCommand(string);
  45         SeqErrorCommand();
  46         ~SeqErrorCommand() {}
  47
  48         vector<string> setParameters();
  49         string getCommandName()                 { return "seq.error";                           }
  50         string getCommandCategory()             { return "Sequence Processing";         }
  51         string getHelpString();
  52         string getCitation() { return "http://www.mothur.org/wiki/Seq.error"; }
  53
  54         int execute();
  55         void help() { m->mothurOut(getHelpString()); }
  56
  57 private:
  58         bool abort;
  59
  60         struct linePair {
  61                 unsigned long int start;
  62                 unsigned long int end;
  63                 linePair(unsigned long int i, unsigned long int j) : start(i), end(j) {}
  64         };
  65
  66         vector<int> processIDS;   //processid
  67         vector<linePair> lines;
  68         vector<linePair> qLines;
  69         vector<linePair> rLines;
  70
  71         void getReferences();
  72         map<string,int> getWeights();
  73         Compare getErrors(Sequence, Sequence);
  74         void printErrorHeader(ofstream&);
  75         void printErrorData(Compare, int, ofstream&, ofstream&);
  76         void printSubMatrix();
  77         void printErrorFRFile(map<char, vector<int> >, map<char, vector<int> >);
  78         void printErrorQuality(map<char, vector<int> >);
  79         void printQualityFR(vector<vector<int> >, vector<vector<int> >);
  80
  81         int setLines(string, string, string, vector<unsigned long int>&, vector<unsigned long int>&, vector<unsigned long int>&);
  82         int driver(string, string, string, string, string, string, linePair, linePair, linePair);
  83         int createProcesses(string, string, string, string, string, string);
  84
  85         string queryFileName, referenceFileName, qualFileName, reportFileName, namesFileName, outputDir;
  86         double threshold;
  87         bool ignoreChimeras;
  88         int numRefs, processors;
  89         int maxLength, totalBases, totalMatches;
  90         //ofstream errorSummaryFile, errorSeqFile;
  91         vector<string> outputNames;
  92
  93         vector<Sequence> referenceSeqs;
  94         vector<vector<int> > substitutionMatrix;
  95         vector<vector<int> > qualForwardMap;
  96         vector<vector<int> > qualReverseMap;
  97         vector<int> misMatchCounts;
  98         map<char, vector<int> > qScoreErrorMap;
  99         map<char, vector<int> > errorForward;
 100         map<char, vector<int> > errorReverse;
 101         map<string, int> weights;
 102         vector<string> megaAlignVector;
 103
 104 };
 105
 106 #endif