pintail.h

   1 #ifndef PINTAIL_H
   2 #define PINTAIL_H
   3
   4 /*
   5  *  pintail.h
   6  *  Mothur
   7  *
   8  *  Created by Sarah Westcott on 7/9/09.
   9  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
  10  *
  11  */
  12
  13 #include "chimera.h"
  14 #include "dist.h"
  15
  16 //This class was created using the algorythms described in the
  17 // "At Least 1 in 20 16S rRNA Sequence Records Currently Held in the Public Repositories is Estimated To Contain Substantial Anomalies" paper
  18 //by Kevin E. Ashelford 1, Nadia A. Chuzhanova 3, John C. Fry 1, Antonia J. Jones 2 and Andrew J. Weightman 1.
  19
  20 /***********************************************************/
  21
  22 class Pintail : public Chimera {
  23
  24         public:
  25                 Pintail(string);
  26                 ~Pintail();
  27
  28                 void getChimeras();
  29                 void print(ostream&);
  30
  31
  32         private:
  33
  34                 struct linePair {
  35                         int start;
  36                         int end;
  37                         linePair(int i, int j) : start(i), end(j) {}
  38                 };
  39
  40
  41                 Dist* distCalculator;
  42                 string fastafile;
  43                 int iters;
  44                 vector<linePair*> lines;
  45                 vector<Sequence*> querySeqs;
  46                 vector<Sequence*> templateSeqs;
  47
  48                 map<Sequence*, Sequence*> bestfit;  //maps a query sequence to its most similiar sequence in the template
  49                 map<Sequence*, Sequence*>::iterator itBest;
  50
  51                 map<Sequence*, vector<float> > obsDistance;  //maps a query sequence to its observed distance at each window
  52                 map<Sequence*, vector<float> > expectedDistance;  //maps a query sequence to its expected distance at each window
  53                 map<Sequence*, vector<float> >::iterator itObsDist;
  54                 map<Sequence*, vector<float> >::iterator itExpDist;
  55
  56                 vector<float> averageProbability;                       //Qav
  57                 map<Sequence*, float> seqCoef;                          //maps a sequence to its coefficient
  58                 map<Sequence*, float> DE;                                       //maps a sequence to its deviation
  59                 map<Sequence*, float>::iterator itCoef;
  60
  61                 vector<Sequence*> readSeqs(string);
  62                 vector<float> findQav(vector<float>);
  63                 vector<float> calcFreq(vector<Sequence*>);
  64                 map<Sequence*, float> getCoef(vector<float>);
  65
  66                 void findPairs(int, int);
  67                 void calcObserved(int, int);
  68                 void calcExpected(int, int);
  69                 void calcDE(int, int);
  70
  71                 void createProcessesPairs();
  72                 void createProcessesObserved();
  73                 void createProcessesExpected();
  74                 void createProcessesDE();
  75
  76 };
  77
  78 /***********************************************************/
  79
  80 #endif
  81